push to specific repo

.github/ISSUE_TEMPLATE/bug-report.yml

@@ -12,83 +12,57 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-name: "🚀 Issue / Bug / Request"
-description: Report a bug, suggest an improvement, or ask a technical question.
+name: "\U0001F41B Bug Report"
+description: Submit a bug report to help us improve LeRobot
 body:
   - type: markdown
     attributes:
       value: |
-        ### Thanks for contributing to LeRobot! 🙌
-        Please choose the most relevant sections below. If this is a general "how-to" question, consider our [Discord](https://discord.gg/s3KuuzsPFb) for faster community support.
-
-  - type: dropdown
-    id: issue-type
-    attributes:
-      label: Ticket Type
-      description: What kind of ticket are you opening?
-      options:
-        - "🐛 Bug Report (Something isn't working)"
-        - "💡 Feature Request / Improvement"
-        - "❓ Technical Question"
-        - "🧹 Maintenance / Documentation"
-    validations:
-      required: true
+        Thanks for taking the time to submit a bug report! 🐛
+        If this is not a bug related to the LeRobot library directly, but instead a general question about your code or the library specifically please use our [discord](https://discord.gg/s3KuuzsPFb).
 
   - type: textarea
     id: system-info
     attributes:
-      label: Environment & System Info
-      description: |
-        For bugs or technical questions, please run `lerobot-info` and paste the output.
-        (Optional for feature requests).
+      label: System Info
+      description: Please share your LeRobot configuration by running `lerobot-info` (if installed) or `python -m lerobot.scripts.display_sys_info` (if not installed) and pasting the output below.
       render: Shell
-      placeholder: lerobot version, OS, python version, etc.
+      placeholder: lerobot version, OS, python version, numpy version, torch version, and lerobot's configuration
+    validations:
+      required: true
+
+  - type: checkboxes
+    id: information-scripts-examples
+    attributes:
+      label: Information
+      description: 'The problem arises when using:'
+      options:
+        - label: "One of the scripts in the examples/ folder of LeRobot"
+        - label: "My own task or dataset (give details below)"
 
   - type: textarea
-    id: description
+    id: reproduction
     validations:
       required: true
     attributes:
-      label: Description
+      label: Reproduction
       description: |
-        Provide a clear summary of the issue or your proposal.
-        - **Bugs:** What is happening?
-        - **Features:** What is the goal/use case?
-        - **Questions:** What are you trying to achieve?
+        If needed, provide a simple code sample that reproduces the problem you ran into. It can be a Colab link or just a code snippet.
+        Sharing error messages or stack traces could be useful as well!
+        Important! Use code tags to correctly format your code. See https://help.github.com/en/github/writing-on-github/creating-and-highlighting-code-blocks#syntax-highlighting
+        Try to avoid screenshots, as they are hard to read and don't allow copy-and-pasting.
+
       placeholder: |
-        A clear and concise description of the issue or suggestion.
+        Steps to reproduce the behavior:
+
+          1.
+          2.
+          3.
 
   - type: textarea
-    id: context-repro
+    id: expected-behavior
+    validations:
+      required: true
     attributes:
-      label: Context & Reproduction
-      description: |
-        Provide a code snippet, steps to reproduce a bug, or technical details about your proposal.
-        Please use code blocks for scripts and CLI commands.
-      placeholder: |
-        Steps to reproduce / Usage example:
-        1.
-        2.
-        3.
-
-  - type: textarea
-    id: logs
-    attributes:
-      label: Relevant logs or stack trace
-      description: If applicable, paste relevant error logs here.
-      render: Shell
-
-  - type: checkboxes
-    id: extras
-    attributes:
-      label: Checklist
-      options:
-        - label: I have searched existing tickets to ensure this isn't a duplicate.
-        - label: I am using the latest version of the `main` branch.
-        - label: I have verified this is not an environment-specific problem.
-
-  - type: textarea
-    id: workaround
-    attributes:
-      label: Additional Info / Workarounds
-      description: Anything else we should know? If you have a workaround, please share it!
+      label: Expected behavior
+      description: "A clear and concise description of what you would expect to happen."

.github/PULL_REQUEST_TEMPLATE.md

@@ -1,55 +1,41 @@
-## Title
+## What this does
 
-Short, imperative summary (e.g., "fix(robots): handle None in sensor parser"). See [CONTRIBUTING.md](../CONTRIBUTING.md) for PR conventions.
+Explain what this PR does. Feel free to tag your PR with the appropriate label(s).
 
-## Type / Scope
+Examples:
+| Title | Label |
+|----------------------|-----------------|
+| Fixes #[issue] | (🐛 Bug) |
+| Adds new dataset | (🗃️ Dataset) |
+| Optimizes something | (⚡️ Performance) |
 
-- **Type**: (Bug | Feature | Docs | Performance | Test | CI | Chore)
-- **Scope**: (optional — name of module or package affected)
+## How it was tested
 
-## Summary / Motivation
+Explain/show how you tested your changes.
 
-- One-paragraph description of what changes and why.
-- Why this change is needed and any trade-offs or design notes.
+Examples:
 
-## Related issues
+- Added `test_something` in `tests/test_stuff.py`.
+- Added `new_feature` and checked that training converges with policy X on dataset/environment Y.
+- Optimized `some_function`, it now runs X times faster than previously.
 
-- Fixes / Closes: # (if any)
-- Related: # (if any)
+## How to checkout & try? (for the reviewer)
 
-## What changed
+Provide a simple way for the reviewer to try out your changes.
 
-- Short, concrete bullets of the modifications (files/behaviour).
-- Short note if this introduces breaking changes and migration steps.
+Examples:
 
-## How was this tested (or how to run locally)
+```bash
+pytest -sx tests/test_stuff.py::test_something
+```
 
-- Tests added: list new tests or test files.
-- Manual checks / dataset runs performed.
-- Instructions for the reviewer
+```bash
+lerobot-train --some.option=true
+```
 
-Example:
+## SECTION TO REMOVE BEFORE SUBMITTING YOUR PR
 
-- Ran the relevant tests:
+**Note**: Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
+members/contributors who may be interested in your PR. Try to avoid tagging more than 3 people.
 
-  ```bash
-  pytest -q tests/ -k <keyword>
-  ```
-
-- Reproduce with a quick example or CLI (if applicable):
-
-  ```bash
-  lerobot-train --some.option=true
-  ```
-
-## Checklist (required before merge)
-
-- [ ] Linting/formatting run (`pre-commit run -a`)
-- [ ] All tests pass locally (`pytest`)
-- [ ] Documentation updated
-- [ ] CI is green
-
-## Reviewer notes
-
-- Anything the reviewer should focus on (performance, edge-cases, specific files) or general notes.
-- Anyone in the community is free to review the PR.
+**Note**: Before submitting this PR, please read the [contributor guideline](https://github.com/huggingface/lerobot/blob/main/CONTRIBUTING.md#submitting-a-pull-request-pr).

.github/labeler.yml

@@ -1,69 +0,0 @@
-# Copyright 2025 The HuggingFace Inc. team. All rights reserved.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-CI:
-  - changed-files:
-      - any-glob-to-any-file:
-        - '.github/**'
-        - 'docker/**'
-
-github_actions:
-  - changed-files:
-      - any-glob-to-any-file: '.github/**'
-
-documentation:
-  - changed-files:
-      - any-glob-to-any-file:
-          - '**/*.md'
-          - '**/*.mdx'
-          - 'docs/**'
-
-examples:
-  - changed-files:
-      - any-glob-to-any-file: 'examples/**'
-
-tests:
-  - changed-files:
-      - any-glob-to-any-file: 'tests/**'
-
-sensors:
-  - changed-files:
-      - any-glob-to-any-file: 'src/lerobot/cameras/**'
-
-configuration:
-  - changed-files:
-      - any-glob-to-any-file: 'src/lerobot/configs/**'
-
-dataset:
-  - changed-files:
-      - any-glob-to-any-file: 'src/lerobot/datasets/**'
-
-evaluation:
-  - changed-files:
-      - any-glob-to-any-file: 'src/lerobot/envs/**'
-
-robots:
-  - changed-files:
-      - any-glob-to-any-file:
-          - 'src/lerobot/teleoperators/**'
-          - 'src/lerobot/robots/**'
-          - 'src/lerobot/motors/**'
-
-policies:
-  - changed-files:
-      - any-glob-to-any-file: 'src/lerobot/policies/**'
-
-processor:
-  - changed-files:
-      - any-glob-to-any-file: 'src/lerobot/processor/**'

.github/workflows/documentation-upload-pr.yml

@@ -31,8 +31,7 @@ jobs:
     name: Upload Preview and Comment
     if: >
       github.event.workflow_run.event == 'pull_request' &&
-      github.event.workflow_run.conclusion == 'success' &&
-      github.repository == 'huggingface/lerobot'
+      github.event.workflow_run.conclusion == 'success'
     uses: huggingface/doc-builder/.github/workflows/upload_pr_documentation.yml@main
     with:
       package_name: lerobot

.github/workflows/documentation.yml

@@ -18,11 +18,6 @@ name: Documentation
 on:
   # Allows running this workflow manually from the Actions tab
   workflow_dispatch:
-    inputs:
-      version:
-        description: 'Version tag (e.g. v0.1.2) - Leave empty for standard main build'
-        required: false
-        type: string
 
   # Triggers the workflow on push events to main for the docs folder
   push:
@@ -38,9 +33,6 @@ on:
     paths:
       - "docs/**"
 
-  release:
-    types: [published]
-
 # Ensures that only the latest commit for a PR or branch is built, canceling older runs.
 concurrency:
   group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
@@ -50,22 +42,14 @@ jobs:
   # This job builds and deploys the official documentation.
   build_main_docs:
     name: Build Main Docs
-    if: >
-      (github.event_name == 'push' || github.event_name == 'workflow_dispatch' || github.event_name == 'release') &&
-      github.repository == 'huggingface/lerobot'
+    if: github.event_name == 'push' || github.event_name == 'workflow_dispatch'
     permissions:
       contents: read
     uses: huggingface/doc-builder/.github/workflows/build_main_documentation.yml@main
     with:
       commit_sha: ${{ github.sha }}
       package: lerobot
-      additional_args: >-
-        --not_python_module
-        ${{
-          (github.event_name == 'release' && format('--version {0}', github.event.release.tag_name)) ||
-          (inputs.version != '' && format('--version {0}', inputs.version)) ||
-          ''
-        }}
+      additional_args: --not_python_module
     secrets:
       token: ${{ secrets.HUGGINGFACE_PUSH }}
       hf_token: ${{ secrets.HF_DOC_BUILD_PUSH }}
@@ -74,7 +58,7 @@ jobs:
   # The result of this job triggers the 'Upload PR Documentation' workflow.
   build_pr_docs:
     name: Build PR Docs
-    if: github.event_name == 'pull_request' && github.repository == 'huggingface/lerobot'
+    if: github.event_name == 'pull_request'
     permissions:
       contents: read
       pull-requests: write

.github/workflows/fast_tests.yml

@@ -45,6 +45,7 @@ permissions:
 env:
   UV_VERSION: "0.8.0"
   PYTHON_VERSION: "3.10"
+  DOCKER_IMAGE_NAME: huggingface/lerobot-gpu
 
 # Ensures that only the latest commit for a PR or branch is built, canceling older runs.
 concurrency:
@@ -62,7 +63,7 @@ jobs:
       HF_HOME: /mnt/cache/.cache/huggingface
       HF_LEROBOT_HOME: /mnt/cache/.cache/huggingface/lerobot
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v4
         with:
           persist-credentials: false
           lfs: true

.github/workflows/full_tests.yml

@@ -61,7 +61,7 @@ jobs:
       HF_HOME: /mnt/cache/.cache/huggingface
       HF_LEROBOT_HOME: /mnt/cache/.cache/huggingface/lerobot
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v4
         with:
           lfs: true
           persist-credentials: false
@@ -85,7 +85,7 @@ jobs:
           python-version: ${{ env.PYTHON_VERSION }}
 
       - name: Install lerobot with all extras
-        run: uv sync --extra all # TODO(Steven): Make flash-attn optional
+        run: uv sync --all-extras --no-extra groot # TODO(Steven): Make flash-attn optional
 
       - name: Run pytest (all extras)
         run: uv run pytest tests -vv --maxfail=10
@@ -127,7 +127,7 @@ jobs:
           sudo apt-get update
           sudo apt-get install git-lfs
           git lfs install
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v4
         with:
           lfs: true
           persist-credentials: false
@@ -186,18 +186,15 @@ jobs:
     steps:
       - name: Get Docker Hub Token and Delete Image
         # zizmor: ignore[template-injection]
-        env:
-          DOCKERHUB_LEROBOT_USERNAME: ${{ secrets.DOCKERHUB_LEROBOT_USERNAME }}
-          DOCKERHUB_LEROBOT_PASSWORD: ${{ secrets.DOCKERHUB_LEROBOT_PASSWORD }}
-          IMAGE_FULL: ${{ needs.build-and-push-docker.outputs.image_tag }}
         run: |
-          IMAGE_NAME=$(echo "$IMAGE_FULL" | cut -d':' -f1)
-          IMAGE_TAG=$(echo "$IMAGE_FULL" | cut -d':' -f2-)
+          IMAGE_NAME=$(echo "${{ needs.build-and-push-docker.outputs.image_tag }}" | cut -d':' -f1)
+          IMAGE_TAG=$(echo "${{ needs.build-and-push-docker.outputs.image_tag }}" | cut -d':' -f2)
+
           echo "Attempting to delete image: $IMAGE_NAME:$IMAGE_TAG"
 
           TOKEN=$(curl -s -H "Content-Type: application/json" \
                        -X POST \
-                       -d "{\"username\": \"$DOCKERHUB_LEROBOT_USERNAME\", \"password\": \"$DOCKERHUB_LEROBOT_PASSWORD\"}" \
+                       -d '{"username": "${{ secrets.DOCKERHUB_LEROBOT_USERNAME }}", "password": "${{ secrets.DOCKERHUB_LEROBOT_PASSWORD }}"}' \
                        https://hub.docker.com/v2/users/login/ | jq -r .token)
 
           if [ "$TOKEN" == "null" ] || [ -z "$TOKEN" ]; then
@@ -208,7 +205,7 @@ jobs:
           HTTP_RESPONSE=$(curl -s -o /dev/null -w "%{http_code}" \
                                -H "Authorization: JWT ${TOKEN}" \
                                -X DELETE \
-                               https://hub.docker.com/v2/repositories/${IMAGE_NAME}/tags/$IMAGE_TAG)
+                               https://hub.docker.com/v2/repositories/${IMAGE_NAME}/tags/${IMAGE_TAG}/)
 
           if [ "$HTTP_RESPONSE" -eq 204 ]; then
             echo "Successfully deleted Docker image tag: $IMAGE_NAME:$IMAGE_TAG"

.github/workflows/issue_labeler.yml

@@ -1,77 +0,0 @@
-# Copyright 2025 The HuggingFace Inc. team. All rights reserved.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-# This workflow automatically labels issues based on their content.
-name: Issue Labeler
-on:
-  # Trigger on new issues and edits to existing issues
-  issues:
-    types: [opened, edited]
-
-permissions:
-  contents: read
-  issues: write
-
-jobs:
-  label-issue:
-    name: Auto Label Issue
-    runs-on: ubuntu-latest
-    if: github.repository == 'huggingface/lerobot'
-    steps:
-      - uses: actions/github-script@v8
-        with:
-          script: |
-            // Setup Input Text
-            const body = (context.payload.issue.body || '');
-            const title = (context.payload.issue.title || '');
-            const cleanBody = body.replace(/```[\s\S]*?```/g, '');
-            const text = `${title}\n${cleanBody}`.toLowerCase();
-            const labelsToAdd = new Set();
-            const matches = (re) => re.test(text);
-
-            // Keyword Heuristics
-
-            if (matches(/\b(bug|error|crash|exception)\b/i)) labelsToAdd.add('bug');
-            if (matches(/\b(new feature|enhancement|improvement|proposal|feature request)\b/i)) labelsToAdd.add('enhancement');
-            if (matches(/\b(question|how to|clarify|explain|how do i|help me|question about)\b/i)) labelsToAdd.add('question');
-            if (matches(/\b(documentation|docs?|readme|tutorial|wiki|typo|docstring)\b/i)) labelsToAdd.add('documentation');
-            if (matches(/\b(example|sample|demo|notebook)s?\b/i)) labelsToAdd.add('examples');
-            if (matches(/\b(datasets?|data loader|data augmentation|data preprocessing)\b/i)) labelsToAdd.add('dataset');
-            if (matches(/\b(mujoco|isaac|simulation|sim)\b/i)) labelsToAdd.add('simulation');
-            if (matches(/\b(train|training|optimizer|gradient|wandb|sac)\b/i)) labelsToAdd.add('training');
-            if (matches(/\b(rerun|plot|render|rendering|visualizer)/i)) labelsToAdd.add('visualization');
-            if (matches(/\b(cameras?|opencv|realsense|lidars?|sensors?|imus?|microphones?|rgbd|encoders?)\b/i)) labelsToAdd.add('sensors');
-            if (matches(/\b(urdf|actuators?|calibration|end-effector|kinematics)\b/i)) labelsToAdd.add('robots');
-            if (matches(/\b(teleop|teleoperator|controller|leader|follower|joystick|gamepad)\b/i)) labelsToAdd.add('teleoperators');
-            if (matches(/\b(policy|policies|model?)\b/i)) labelsToAdd.add('policies');
-            if (matches(/\b(processor|pipeline|preprocessor|postprocessor)s?\b/i)) labelsToAdd.add('processor');
-            if (matches(/\b(eval|evaluate|evaluation|metrics?|score|benchmarks?)\b/i)) labelsToAdd.add('evaluation');
-            if (matches(/\b(tests?|pytest|unittest|failing test)\b/i)) labelsToAdd.add('tests');
-            if (matches(/\b(ci|github actions?|github workflows?|gha|docker|pypi)\b/i)) labelsToAdd.add('CI');
-            if (matches(/\b(perf|latency|throughput|fps|speed|performance|slow|fast|slower|faster|memory usage)\b/i)) labelsToAdd.add('performance');
-            if (matches(/\b(dependency|dependencies|pip|install error|importerror|package not found|pyproject)\b/i)) labelsToAdd.add('dependencies');
-            if (matches(/\b(configuration|config|arguments?|input feature|dracuss)\b/i)) labelsToAdd.add('configuration');
-
-            // Apply Labels
-            const labels = Array.from(labelsToAdd).filter(Boolean);
-
-            if (labels.length > 0) {
-              console.log(`Adding labels: ${labels.join(', ')}`);
-              await github.rest.issues.addLabels({
-                owner: context.repo.owner,
-                repo: context.repo.repo,
-                issue_number: context.issue.number,
-                labels,
-              });
-            }

.github/workflows/nightly.yml

@@ -43,7 +43,6 @@ jobs:
     name: Build CPU Docker for Nightly
     runs-on:
       group: aws-general-8-plus
-    if: github.repository == 'huggingface/lerobot'
     outputs:
       image_tag: ${{ env.DOCKER_IMAGE_NAME_CPU }}
     steps:
@@ -52,7 +51,7 @@ jobs:
           sudo apt-get update
           sudo apt-get install git-lfs
           git lfs install
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v4
         with:
           lfs: true
           persist-credentials: false
@@ -78,7 +77,6 @@ jobs:
     name: Build GPU Docker for Nightly
     runs-on:
       group: aws-general-8-plus
-    if: github.repository == 'huggingface/lerobot'
     outputs:
       image_tag: ${{ env.DOCKER_IMAGE_NAME_GPU }}
     steps:
@@ -87,7 +85,7 @@ jobs:
           sudo apt-get update
           sudo apt-get install git-lfs
           git lfs install
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v4
         with:
           lfs: true
           persist-credentials: false

.github/workflows/pr_labeler.yml

@@ -1,39 +0,0 @@
-# Copyright 2025 The HuggingFace Inc. team. All rights reserved.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-# This workflow labels pull requests based on the files that were changed.
-name: Pull Request Labeler
-
-on:
-  # Allows labeling pull requests when they are opened or updated
-  # zizmor: ignore[dangerous-triggers] Needed to label PRs from forks
-  pull_request_target:
-    branches:
-      - main
-    types: [opened, synchronize, reopened, ready_for_review]
-
-permissions:
-  contents: read
-  pull-requests: write
-
-jobs:
-  triage:
-    name: Label PR
-    runs-on: ubuntu-latest
-    if: github.repository == 'huggingface/lerobot' && !github.event.pull_request.draft
-    steps:
-      - uses: actions/labeler@v6
-        with:
-          repo-token: ${{ secrets.GITHUB_TOKEN }}
-          sync-labels: true # Removes labels if files are removed from the PR

.github/workflows/quality.yml

@@ -43,12 +43,12 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout code
-        uses: actions/checkout@v6
+        uses: actions/checkout@v4
         with:
           persist-credentials: false
 
       - name: Set up Python
-        uses: actions/setup-python@v6
+        uses: actions/setup-python@v5
         with:
           python-version: '3.10'
 

.github/workflows/release.yml

@@ -29,7 +29,6 @@ jobs:
   build-and-publish:
     name: Build and publish Python distributions
     runs-on: ubuntu-latest
-    if: github.repository == 'huggingface/lerobot'
     outputs:
       version: ${{ steps.extract_info.outputs.tag_version }}
     permissions:
@@ -38,12 +37,12 @@ jobs:
 
     steps:
       - name: Checkout code
-        uses: actions/checkout@v6
+        uses: actions/checkout@v4
         with:
           persist-credentials: false
 
       - name: Set up Python
-        uses: actions/setup-python@v6
+        uses: actions/setup-python@v5
         with:
           python-version: '3.10'
 
@@ -135,7 +134,7 @@ jobs:
     env:
       MUJOCO_GL: egl
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v4
         with:
           lfs: true
           persist-credentials: false
@@ -177,3 +176,4 @@ jobs:
 
 # TODO(Steven): Publish draft/pre-release and to test pypi weekly
 # TODO(Steven): Separate build and publish job
+# TODO(Steven): Tag documentation with the same version as the package

.github/workflows/security.yml

@@ -43,7 +43,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout code
-        uses: actions/checkout@v6 # zizmor: ignore[unpinned-uses]
+        uses: actions/checkout@v4 # zizmor: ignore[unpinned-uses]
         with:
           fetch-depth: 0
           persist-credentials: false

.github/workflows/stale.yml

@@ -45,7 +45,6 @@ jobs:
   stale:
     name: Close Stale Issues and PRs
     runs-on: ubuntu-latest
-    if: github.repository == 'huggingface/lerobot'
     permissions:
       actions: write
       contents: write # only for delete-branch option

.github/workflows/unbound_deps_tests.yml

@@ -20,8 +20,8 @@ on:
   workflow_dispatch:
 
   # Run on the 1st and 15th of every month at 09:00 UTC
-  # schedule:
-  #  - cron: '0 2 1,15 * *'
+  schedule:
+    - cron: '0 2 1,15 * *'
 
 permissions:
   contents: read
@@ -43,13 +43,12 @@ jobs:
   full-tests:
     name: Full Unbound Tests
     runs-on: ubuntu-latest
-    if: github.repository == 'huggingface/lerobot'
     env:
       MUJOCO_GL: egl
       HF_HOME: /mnt/cache/.cache/huggingface
       HF_LEROBOT_HOME: /mnt/cache/.cache/huggingface/lerobot
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v4
         with:
           lfs: true
           persist-credentials: false
@@ -78,7 +77,7 @@ jobs:
           echo "Dependencies unbound:" && cat pyproject.toml
 
       - name: Install lerobot with all extras
-        run: uv sync --extra all # TODO(Steven): Make flash-attn optional
+        run: uv sync --all-extras --no-extra groot # TODO(Steven): Make flash-attn optional
 
       - name: Run pytest (all extras)
         run: uv run pytest tests -vv
@@ -101,7 +100,7 @@ jobs:
           sudo apt-get update
           sudo apt-get install git-lfs
           git lfs install
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v4
         with:
           lfs: true
           persist-credentials: false
@@ -162,19 +161,15 @@ jobs:
     steps:
       - name: Get Docker Hub Token and Delete Image
         # zizmor: ignore[template-injection]
-        env:
-          DOCKERHUB_LEROBOT_USERNAME: ${{ secrets.DOCKERHUB_LEROBOT_USERNAME }}
-          DOCKERHUB_LEROBOT_PASSWORD: ${{ secrets.DOCKERHUB_LEROBOT_PASSWORD }}
-          IMAGE_FULL: ${{ needs.build-and-push-docker.outputs.image_tag }}
         run: |
-          IMAGE_NAME=$(echo "$IMAGE_FULL" | cut -d':' -f1)
-          IMAGE_TAG=$(echo "$IMAGE_FULL" | cut -d':' -f2)
+          IMAGE_NAME=$(echo "${{ needs.build-and-push-docker.outputs.image_tag }}" | cut -d':' -f1)
+          IMAGE_TAG=$(echo "${{ needs.build-and-push-docker.outputs.image_tag }}" | cut -d':' -f2)
 
           echo "Attempting to delete image: $IMAGE_NAME:$IMAGE_TAG"
 
           TOKEN=$(curl -s -H "Content-Type: application/json" \
                        -X POST \
-                       -d "{\"username\": \"$DOCKERHUB_LEROBOT_USERNAME\", \"password\": \"$DOCKERHUB_LEROBOT_PASSWORD\"}" \
+                       -d '{"username": "${{ secrets.DOCKERHUB_LEROBOT_USERNAME }}", "password": "${{ secrets.DOCKERHUB_LEROBOT_PASSWORD }}"}' \
                        https://hub.docker.com/v2/users/login/ | jq -r .token)
 
           if [ "$TOKEN" == "null" ] || [ -z "$TOKEN" ]; then
@@ -185,7 +180,7 @@ jobs:
           HTTP_RESPONSE=$(curl -s -o /dev/null -w "%{http_code}" \
                                -H "Authorization: JWT ${TOKEN}" \
                                -X DELETE \
-                               https://hub.docker.com/v2/repositories/${IMAGE_NAME}/tags/$IMAGE_TAG)
+                               https://hub.docker.com/v2/repositories/${IMAGE_NAME}/tags/${IMAGE_TAG}/)
 
           if [ "$HTTP_RESPONSE" -eq 204 ]; then
             echo "Successfully deleted Docker image tag: $IMAGE_NAME:$IMAGE_TAG"

.pre-commit-config.yaml

@@ -87,7 +87,7 @@ repos:
   # TODO(Steven): Uncomment when ready to use
   ##### Static Analysis & Typing #####
   - repo: https://github.com/pre-commit/mirrors-mypy
-    rev: v1.19.1
+    rev: v1.18.2
     hooks:
       - id: mypy
         args: [--config-file=pyproject.toml]

CODE_OF_CONDUCT.md

@@ -52,7 +52,7 @@ decisions when appropriate.
 
 This Code of Conduct applies within all community spaces, and also applies when
 an individual is officially representing the community in public spaces.
-Examples of representing our community include using an official e-mail address,
+Examples of representing our community include using an official email address,
 posting via an official social media account, or acting as an appointed
 representative at an online or offline event.
 
@@ -60,7 +60,7 @@ representative at an online or offline event.
 
 Instances of abusive, harassing, or otherwise unacceptable behavior may be
 reported to the community leaders responsible for enforcement at
-feedback@huggingface.co.
+[feedback@huggingface.co](mailto:feedback@huggingface.co).
 All complaints will be reviewed and investigated promptly and fairly.
 
 All community leaders are obligated to respect the privacy and security of the

CONTRIBUTING.md

@@ -1,83 +1,323 @@
-# How to contribute to 🤗 LeRobot
+# How to contribute to 🤗 LeRobot?
 
-Everyone is welcome to contribute, and we value everybody's contribution. Code is not the only way to help the community. Answering questions, helping others, reaching out, and improving the documentation are immensely valuable.
+Everyone is welcome to contribute, and we value everybody's contribution. Code
+is thus not the only way to help the community. Answering questions, helping
+others, reaching out and improving the documentations are immensely valuable to
+the community.
 
-Whichever way you choose to contribute, please be mindful to respect our [code of conduct](./CODE_OF_CONDUCT.md).
+It also helps us if you spread the word: reference the library from blog posts
+on the awesome projects it made possible, shout out on Twitter when it has
+helped you, or simply ⭐️ the repo to say "thank you".
 
-## Ways to Contribute
+Whichever way you choose to contribute, please be mindful to respect our
+[code of conduct](https://github.com/huggingface/lerobot/blob/main/CODE_OF_CONDUCT.md).
 
-You can contribute in many ways:
+## You can contribute in so many ways!
 
-- **Fixing issues:** Resolve bugs or improve existing code.
-- **New features:** Develop new features.
-- **Extend:** Implement new models/policies, robots, or simulation environments and upload datasets to the Hugging Face Hub.
-- **Documentation:** Improve examples, guides, and docstrings.
-- **Feedback:** Submit tickets related to bugs or desired new features.
+Some of the ways you can contribute to 🤗 LeRobot:
 
-If you are unsure where to start, join our [Discord Channel](https://discord.gg/q8Dzzpym3f).
+- Fixing outstanding issues with the existing code.
+- Implementing new models, datasets or simulation environments.
+- Contributing to the examples or to the documentation.
+- Submitting issues related to bugs or desired new features.
 
-## Development Setup
+Following the guides below, feel free to open issues and PRs and to coordinate your efforts with the community on our [Discord Channel](https://discord.gg/VjFz58wn3R). For specific inquiries, reach out to [Remi Cadene](mailto:remi.cadene@huggingface.co).
 
-To contribute code, you need to set up a development environment.
+If you are not sure how to contribute or want to know the next features we working on, look on this project page: [LeRobot TODO](https://github.com/orgs/huggingface/projects/46)
 
-### 1. Fork and Clone
+## Submitting a new issue or feature request
 
-Fork the repository on GitHub, then clone your fork:
-
-```bash
-git clone https://github.com/<your-handle>/lerobot.git
-cd lerobot
-git remote add upstream https://github.com/huggingface/lerobot.git
-```
-
-### 2. Environment Installation
-
-Please follow our [Installation Guide](./docs/source/installation.mdx) for the environment setup & installation from source.
-
-## Running Tests & Quality Checks
-
-### Code Style (Pre-commit)
-
-Install `pre-commit` hooks to run checks automatically before you commit:
-
-```bash
-pre-commit install
-```
-
-To run checks manually on all files:
-
-```bash
-pre-commit run --all-files
-```
-
-### Running Tests
-
-We use `pytest`. First, ensure you have test artifacts by installing **git-lfs**:
+Do your best to follow these guidelines when submitting an issue or a feature
+request. It will make it easier for us to come back to you quickly and with good
+feedback.
+
+### Did you find a bug?
+
+The 🤗 LeRobot library is robust and reliable thanks to the users who notify us of
+the problems they encounter. So thank you for reporting an issue.
+
+First, we would really appreciate it if you could **make sure the bug was not
+already reported** (use the search bar on Github under Issues).
+
+Did not find it? :( So we can act quickly on it, please follow these steps:
+
+- Include your **OS type and version**, the versions of **Python** and **PyTorch**.
+- A short, self-contained, code snippet that allows us to reproduce the bug in
+  less than 30s.
+- The full traceback if an exception is raised.
+- Attach any other additional information, like screenshots, you think may help.
+
+### Do you want a new feature?
+
+A good feature request addresses the following points:
+
+1. Motivation first:
+
+- Is it related to a problem/frustration with the library? If so, please explain
+  why. Providing a code snippet that demonstrates the problem is best.
+- Is it related to something you would need for a project? We'd love to hear
+  about it!
+- Is it something you worked on and think could benefit the community?
+  Awesome! Tell us what problem it solved for you.
+
+2. Write a _paragraph_ describing the feature.
+3. Provide a **code snippet** that demonstrates its future use.
+4. In case this is related to a paper, please attach a link.
+5. Attach any additional information (drawings, screenshots, etc.) you think may help.
+
+If your issue is well written we're already 80% of the way there by the time you
+post it.
+
+## Adding new policies, datasets or environments
+
+Look at our implementations for [datasets](./src/lerobot/datasets/), [policies](./src/lerobot/policies/),
+environments ([aloha](https://github.com/huggingface/gym-aloha),
+[pusht](https://github.com/huggingface/gym-pusht))
+and follow the same api design.
+
+When implementing a new dataset loadable with LeRobotDataset follow these steps:
+
+- Update `available_datasets_per_env` in `lerobot/__init__.py`
+
+When implementing a new environment (e.g. `gym_aloha`), follow these steps:
+
+- Update `available_tasks_per_env` and `available_datasets_per_env` in `lerobot/__init__.py`
+
+When implementing a new policy class (e.g. `DiffusionPolicy`) follow these steps:
+
+- Update `available_policies` and `available_policies_per_env`, in `lerobot/__init__.py`
+- Set the required `name` class attribute.
+- Update variables in `tests/test_available.py` by importing your new Policy class
+
+## Submitting a pull request (PR)
+
+Before writing code, we strongly advise you to search through the existing PRs or
+issues to make sure that nobody is already working on the same thing. If you are
+unsure, it is always a good idea to open an issue to get some feedback.
+
+You will need basic `git` proficiency to be able to contribute to
+🤗 LeRobot. `git` is not the easiest tool to use but it has the greatest
+manual. Type `git --help` in a shell and enjoy. If you prefer books, [Pro
+Git](https://git-scm.com/book/en/v2) is a very good reference.
+
+Follow these steps to start contributing:
+
+1. Fork the [repository](https://github.com/huggingface/lerobot) by
+   clicking on the 'Fork' button on the repository's page. This creates a copy of the code
+   under your GitHub user account.
+
+2. Clone your fork to your local disk, and add the base repository as a remote. The following command
+   assumes you have your public SSH key uploaded to GitHub. See the following guide for more
+   [information](https://docs.github.com/en/repositories/creating-and-managing-repositories/cloning-a-repository).
+
+   ```bash
+   git clone git@github.com:<your Github handle>/lerobot.git
+   cd lerobot
+   git remote add upstream https://github.com/huggingface/lerobot.git
+   ```
+
+3. Create a new branch to hold your development changes, and do this for every new PR you work on.
+
+   Start by synchronizing your `main` branch with the `upstream/main` branch (more details in the [GitHub Docs](https://docs.github.com/en/github/collaborating-with-issues-and-pull-requests/syncing-a-fork)):
+
+   ```bash
+   git checkout main
+   git fetch upstream
+   git rebase upstream/main
+   ```
+
+   Once your `main` branch is synchronized, create a new branch from it:
+
+   ```bash
+   git checkout -b a-descriptive-name-for-my-changes
+   ```
+
+   🚨 **Do not** work on the `main` branch.
+
+4. for development, we advise to use a tool like `poetry` or `uv` instead of just `pip` to easily track our dependencies.
+   Follow the instructions to [install poetry](https://python-poetry.org/docs/#installation) (use a version >=2.1.0) or to [install uv](https://docs.astral.sh/uv/getting-started/installation/#installation-methods) if you don't have one of them already.
+
+   Set up a development environment with conda:
+
+   ```bash
+   conda create -y -n lerobot-dev python=3.10 && conda activate lerobot-dev
+   ```
+
+   If you're using `uv`, it can manage python versions so you can instead do:
+
+   ```bash
+   uv venv --python 3.10 && source .venv/bin/activate
+   ```
+
+   To develop on 🤗 LeRobot, you will at least need to install the `dev` and `test` extras dependencies along with the core library:
+
+   using `poetry`
+
+   ```bash
+   poetry sync --extras "dev test"
+   ```
+
+   using `uv`
+
+   ```bash
+   uv sync --extra dev --extra test
+   ```
+
+   You can also install the project with all its dependencies (including environments):
+
+   using `poetry`
+
+   ```bash
+   poetry sync --all-extras
+   ```
+
+   using `uv`
+
+   ```bash
+   uv sync --all-extras
+   ```
+
+   > **Note:** If you don't install simulation environments with `--all-extras`, the tests that require them will be skipped when running the pytest suite locally. However, they _will_ be tested in the CI. In general, we advise you to install everything and test locally before pushing.
+
+   Whichever command you chose to install the project (e.g. `poetry sync --all-extras`), you should run it again when pulling code with an updated version of `pyproject.toml` and `poetry.lock` in order to synchronize your virtual environment with the new dependencies.
+
+   The equivalent of `pip install some-package`, would just be:
+
+   using `poetry`
+
+   ```bash
+   poetry add some-package
+   ```
+
+   using `uv`
+
+   ```bash
+   uv add some-package
+   ```
+
+   When making changes to the poetry sections of the `pyproject.toml`, you should run the following command to lock dependencies.
+   using `poetry`
+
+   ```bash
+   poetry lock
+   ```
+
+   using `uv`
+
+   ```bash
+   uv lock
+   ```
+
+5. Develop the features on your branch.
+
+   As you work on the features, you should make sure that the test suite
+   passes. You should run the tests impacted by your changes like this (see
+   below an explanation regarding the environment variable):
+
+   ```bash
+   pytest tests/<TEST_TO_RUN>.py
+   ```
+
+6. Follow our style.
+
+   `lerobot` relies on `ruff` to format its source code
+   consistently. Set up [`pre-commit`](https://pre-commit.com/) to run these checks
+   automatically as Git commit hooks.
+
+   Install `pre-commit` hooks:
+
+   ```bash
+   pre-commit install
+   ```
+
+   You can run these hooks whenever you need on staged files with:
+
+   ```bash
+   pre-commit
+   ```
+
+   Once you're happy with your changes, add changed files using `git add` and
+   make a commit with `git commit` to record your changes locally:
+
+   ```bash
+   git add modified_file.py
+   git commit
+   ```
+
+   Note, if you already committed some changes that have a wrong formatting, you can use:
+
+   ```bash
+   pre-commit run --all-files
+   ```
+
+   Please write [good commit messages](https://chris.beams.io/posts/git-commit/).
+
+   It is a good idea to sync your copy of the code with the original
+   repository regularly. This way you can quickly account for changes:
+
+   ```bash
+   git fetch upstream
+   git rebase upstream/main
+   ```
+
+   Push the changes to your account using:
+
+   ```bash
+   git push -u origin a-descriptive-name-for-my-changes
+   ```
+
+7. Once you are satisfied (**and the checklist below is happy too**), go to the
+   webpage of your fork on GitHub. Click on 'Pull request' to send your changes
+   to the project maintainers for review.
+
+8. It's ok if maintainers ask you for changes. It happens to core contributors
+   too! So everyone can see the changes in the Pull request, work in your local
+   branch and push the changes to your fork. They will automatically appear in
+   the pull request.
+
+### Checklist
+
+1. The title of your pull request should be a summary of its contribution;
+2. If your pull request addresses an issue, please mention the issue number in
+   the pull request description to make sure they are linked (and people
+   consulting the issue know you are working on it);
+3. To indicate a work in progress please prefix the title with `[WIP]`, or preferably mark
+   the PR as a draft PR. These are useful to avoid duplicated work, and to differentiate
+   it from PRs ready to be merged;
+4. Make sure existing tests pass;
+
+### Tests
+
+An extensive test suite is included to test the library behavior and several examples. Library tests can be found in the [tests folder](https://github.com/huggingface/lerobot/tree/main/tests).
+
+Install [git lfs](https://git-lfs.com/) to retrieve test artifacts (if you don't have it already).
+
+On Mac:
 
 ```bash
+brew install git-lfs
 git lfs install
+```
+
+On Ubuntu:
+
+```bash
+sudo apt-get install git-lfs
+git lfs install
+```
+
+Pull artifacts if they're not in [tests/artifacts](tests/artifacts)
+
+```bash
 git lfs pull
 ```
 
-Run the full suite (this may require extras installed):
+We use `pytest` in order to run the tests. From the root of the
+repository, here's how to run tests with `pytest` for the library:
 
 ```bash
-pytest -sv ./tests
+python -m pytest -sv ./tests
 ```
 
-Or run a specific test file during development:
-
-```bash
-pytest -sv tests/test_specific_feature.py
-```
-
-## Submitting Issues & Pull Requests
-
-Use the templates for required fields and examples.
-
-- **Issues:** Follow the [ticket template](./.github/ISSUE_TEMPLATE/bug-report.yml).
-- **Pull requests:** Rebase on `upstream/main`, use a descriptive branch (don't work on `main`), run `pre-commit` and tests locally, and follow the [PR template](./.github/PULL_REQUEST_TEMPLATE.md).
-
-One member of the LeRobot team will then review your contribution.
-
-Thank you for contributing to LeRobot!
+You can specify a smaller set of tests in order to test only the feature
+you're working on.

README.md

@@ -1,5 +1,7 @@
 <p align="center">
-  <img alt="LeRobot, Hugging Face Robotics Library" src="./media/readme/lerobot-logo-thumbnail.png" width="100%">
+  <img alt="LeRobot, Hugging Face Robotics Library" src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/lerobot-logo-thumbnail.png" width="100%">
+  <br/>
+  <br/>
 </p>
 
 <div align="center">
@@ -10,132 +12,323 @@
 [![Status](https://img.shields.io/pypi/status/lerobot)](https://pypi.org/project/lerobot/)
 [![Version](https://img.shields.io/pypi/v/lerobot)](https://pypi.org/project/lerobot/)
 [![Contributor Covenant](https://img.shields.io/badge/Contributor%20Covenant-v2.1-ff69b4.svg)](https://github.com/huggingface/lerobot/blob/main/CODE_OF_CONDUCT.md)
-[![Discord](https://img.shields.io/badge/Discord-Join_Us-5865F2?style=flat&logo=discord&logoColor=white)](https://discord.gg/q8Dzzpym3f)
+[![Discord](https://dcbadge.vercel.app/api/server/C5P34WJ68S?style=flat)](https://discord.gg/s3KuuzsPFb)
+
+<!-- [![Coverage](https://codecov.io/gh/huggingface/lerobot/branch/main/graph/badge.svg?token=TODO)](https://codecov.io/gh/huggingface/lerobot) -->
 
 </div>
 
-**LeRobot** aims to provide models, datasets, and tools for real-world robotics in PyTorch. The goal is to lower the barrier to entry so that everyone can contribute to and benefit from shared datasets and pretrained models.
+<h2 align="center">
+    <p><a href="https://huggingface.co/docs/lerobot/hope_jr">
+        Build Your Own HopeJR Robot!</a></p>
+</h2>
 
-🤗 A hardware-agnostic, Python-native interface that standardizes control across diverse platforms, from low-cost arms (SO-100) to humanoids.
+<div align="center">
+  <img
+    src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/hope_jr/hopejr.png"
+    alt="HopeJR robot"
+    title="HopeJR robot"
+    width="60%"
+  />
 
-🤗 A standardized, scalable LeRobotDataset format (Parquet + MP4 or images) hosted on the Hugging Face Hub, enabling efficient storage, streaming and visualization of massive robotic datasets.
+  <p><strong>Meet HopeJR – A humanoid robot arm and hand for dexterous manipulation!</strong></p>
+  <p>Control it with exoskeletons and gloves for precise hand movements.</p>
+  <p>Perfect for advanced manipulation tasks! 🤖</p>
 
-🤗 State-of-the-art policies that have been shown to transfer to the real-world ready for training and deployment.
+  <p><a href="https://huggingface.co/docs/lerobot/hope_jr">
+      See the full HopeJR tutorial here.</a></p>
+</div>
 
-🤗 Comprehensive support for the open-source ecosystem to democratize physical AI.
+<br/>
 
-## Quick Start
+<h2 align="center">
+    <p><a href="https://huggingface.co/docs/lerobot/so101">
+        Build Your Own SO-101 Robot!</a></p>
+</h2>
 
-LeRobot can be installed directly from PyPI.
+<div align="center">
+  <table>
+    <tr>
+      <td align="center"><img src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/so101/so101.webp" alt="SO-101 follower arm" title="SO-101 follower arm" width="90%"/></td>
+      <td align="center"><img src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/so101/so101-leader.webp" alt="SO-101 leader arm" title="SO-101 leader arm" width="90%"/></td>
+    </tr>
+  </table>
+
+  <p><strong>Meet the updated SO100, the SO-101 – Just €114 per arm!</strong></p>
+  <p>Train it in minutes with a few simple moves on your laptop.</p>
+  <p>Then sit back and watch your creation act autonomously! 🤯</p>
+
+  <p><a href="https://huggingface.co/docs/lerobot/so101">
+      See the full SO-101 tutorial here.</a></p>
+
+  <p>Want to take it to the next level? Make your SO-101 mobile by building LeKiwi!</p>
+  <p>Check out the <a href="https://huggingface.co/docs/lerobot/lekiwi">LeKiwi tutorial</a> and bring your robot to life on wheels.</p>
+
+  <img src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/lekiwi/kiwi.webp" alt="LeKiwi mobile robot" title="LeKiwi mobile robot" width="50%">
+</div>
+
+<br/>
+
+<h3 align="center">
+    <p>LeRobot: State-of-the-art AI for real-world robotics</p>
+</h3>
+
+---
+
+🤗 LeRobot aims to provide models, datasets, and tools for real-world robotics in PyTorch. The goal is to lower the barrier to entry to robotics so that everyone can contribute and benefit from sharing datasets and pretrained models.
+
+🤗 LeRobot contains state-of-the-art approaches that have been shown to transfer to the real-world with a focus on imitation learning and reinforcement learning.
+
+🤗 LeRobot already provides a set of pretrained models, datasets with human collected demonstrations, and simulation environments to get started without assembling a robot. In the coming weeks, the plan is to add more and more support for real-world robotics on the most affordable and capable robots out there.
+
+🤗 LeRobot hosts pretrained models and datasets on this Hugging Face community page: [huggingface.co/lerobot](https://huggingface.co/lerobot)
+
+#### Examples of pretrained models on simulation environments
+
+<table>
+  <tr>
+    <td><img src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/gym/aloha_act.gif" width="100%" alt="ACT policy on ALOHA env"/></td>
+    <td><img src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/gym/simxarm_tdmpc.gif" width="100%" alt="TDMPC policy on SimXArm env"/></td>
+    <td><img src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/gym/pusht_diffusion.gif" width="100%" alt="Diffusion policy on PushT env"/></td>
+  </tr>
+  <tr>
+    <td align="center">ACT policy on ALOHA env</td>
+    <td align="center">TDMPC policy on SimXArm env</td>
+    <td align="center">Diffusion policy on PushT env</td>
+  </tr>
+</table>
+
+## Installation
+
+LeRobot works with Python 3.10+ and PyTorch 2.2+.
+
+### Environment Setup
+
+Create a virtual environment with Python 3.10 and activate it, e.g. with [`miniforge`](https://conda-forge.org/download/):
+
+```bash
+conda create -y -n lerobot python=3.10
+conda activate lerobot
+```
+
+When using `conda`, install `ffmpeg` in your environment:
+
+```bash
+conda install ffmpeg -c conda-forge
+```
+
+> **NOTE:** This usually installs `ffmpeg 7.X` for your platform compiled with the `libsvtav1` encoder. If `libsvtav1` is not supported (check supported encoders with `ffmpeg -encoders`), you can:
+>
+> - _[On any platform]_ Explicitly install `ffmpeg 7.X` using:
+>
+> ```bash
+> conda install ffmpeg=7.1.1 -c conda-forge
+> ```
+>
+> - _[On Linux only]_ Install [ffmpeg build dependencies](https://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu#GettheDependencies) and [compile ffmpeg from source with libsvtav1](https://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu#libsvtav1), and make sure you use the corresponding ffmpeg binary to your install with `which ffmpeg`.
+
+### Install LeRobot 🤗
+
+#### From Source
+
+First, clone the repository and navigate into the directory:
+
+```bash
+git clone https://github.com/huggingface/lerobot.git
+cd lerobot
+```
+
+Then, install the library in editable mode. This is useful if you plan to contribute to the code.
+
+```bash
+pip install -e .
+```
+
+> **NOTE:** If you encounter build errors, you may need to install additional dependencies (`cmake`, `build-essential`, and `ffmpeg libs`). On Linux, run:
+> `sudo apt-get install cmake build-essential python3-dev pkg-config libavformat-dev libavcodec-dev libavdevice-dev libavutil-dev libswscale-dev libswresample-dev libavfilter-dev`. For other systems, see: [Compiling PyAV](https://pyav.org/docs/develop/overview/installation.html#bring-your-own-ffmpeg)
+
+For simulations, 🤗 LeRobot comes with gymnasium environments that can be installed as extras:
+
+- [aloha](https://github.com/huggingface/gym-aloha)
+- [xarm](https://github.com/huggingface/gym-xarm)
+- [pusht](https://github.com/huggingface/gym-pusht)
+
+For instance, to install 🤗 LeRobot with aloha and pusht, use:
+
+```bash
+pip install -e ".[aloha, pusht]"
+```
+
+### Installation from PyPI
+
+**Core Library:**
+Install the base package with:
 
 ```bash
 pip install lerobot
-lerobot-info
 ```
 
-> [!IMPORTANT]
-> For detailed installation guide, please see the [Installation Documentation](https://huggingface.co/docs/lerobot/installation).
+_This installs only the default dependencies._
 
-## Robots & Control
-
-<div align="center">
-  <img src="./media/readme/robots_control_video.webp" width="640px" alt="Reachy 2 Demo">
-</div>
-
-LeRobot provides a unified `Robot` class interface that decouples control logic from hardware specifics. It supports a wide range of robots and teleoperation devices.
-
-```python
-from lerobot.robots.myrobot import MyRobot
-
-# Connect to a robot
-robot = MyRobot(config=...)
-robot.connect()
-
-# Read observation and send action
-obs = robot.get_observation()
-action = model.select_action(obs)
-robot.send_action(action)
-```
-
-**Supported Hardware:** SO100, LeKiwi, Koch, HopeJR, OMX, EarthRover, Reachy2, Gamepads, Keyboards, Phones, OpenARM, Unitree G1.
-
-While these devices are natively integrated into the LeRobot codebase, the library is designed to be extensible. You can easily implement the Robot interface to utilize LeRobot's data collection, training, and visualization tools for your own custom robot.
-
-For detailed hardware setup guides, see the [Hardware Documentation](https://huggingface.co/docs/lerobot/integrate_hardware).
-
-## LeRobot Dataset
-
-To solve the data fragmentation problem in robotics, we utilize the **LeRobotDataset** format.
-
-- **Structure:** Synchronized MP4 videos (or images) for vision and Parquet files for state/action data.
-- **HF Hub Integration:** Explore thousands of robotics datasets on the [Hugging Face Hub](https://huggingface.co/lerobot).
-- **Tools:** Seamlessly delete episodes, split by indices/fractions, add/remove features, and merge multiple datasets.
-
-```python
-from lerobot.datasets.lerobot_dataset import LeRobotDataset
-
-# Load a dataset from the Hub
-dataset = LeRobotDataset("lerobot/aloha_mobile_cabinet")
-
-# Access data (automatically handles video decoding)
-episode_index=0
-print(f"{dataset[episode_index]['action'].shape=}\n")
-```
-
-Learn more about it in the [LeRobotDataset Documentation](https://huggingface.co/docs/lerobot/lerobot-dataset-v3)
-
-## SoTA Models
-
-LeRobot implements state-of-the-art policies in pure PyTorch, covering Imitation Learning, Reinforcement Learning, and Vision-Language-Action (VLA) models, with more coming soon. It also provides you with the tools to instrument and inspect your training process.
-
-<p align="center">
-  <img alt="Gr00t Architecture" src="./media/readme/VLA_architecture.jpg" width="640px">
-</p>
-
-Training a policy is as simple as running a script configuration:
+**Extra Features:**
+To install additional functionality, use one of the following:
 
 ```bash
-lerobot-train \
-  --policy=act \
-  --dataset.repo_id=lerobot/aloha_mobile_cabinet
+pip install 'lerobot[all]'          # All available features
+pip install 'lerobot[aloha,pusht]'  # Specific features (Aloha & Pusht)
+pip install 'lerobot[feetech]'      # Feetech motor support
 ```
 
-| Category                   | Models                                                                                                                                                                                                       |
-| -------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
-| **Imitation Learning**     | [ACT](./docs/source/policy_act_README.md), [Diffusion](./docs/source/policy_diffusion_README.md), [VQ-BeT](./docs/source/policy_vqbet_README.md)                                                             |
-| **Reinforcement Learning** | [HIL-SERL](./docs/source/hilserl.mdx), [TDMPC](./docs/source/policy_tdmpc_README.md) & QC-FQL (coming soon)                                                                                                  |
-| **VLAs Models**            | [Pi0Fast](./docs/source/pi0fast.mdx), [Pi0.5](./docs/source/pi05.mdx), [GR00T N1.5](./docs/source/policy_groot_README.md), [SmolVLA](./docs/source/policy_smolvla_README.md), [XVLA](./docs/source/xvla.mdx) |
+_Replace `[...]` with your desired features._
 
-Similarly to the hardware, you can easily implement your own policy & leverage LeRobot's data collection, training, and visualization tools, and share your model to the HF Hub
+**Available Tags:**
+For a full list of optional dependencies, see:
+https://pypi.org/project/lerobot/
 
-For detailed policy setup guides, see the [Policy Documentation](https://huggingface.co/docs/lerobot/bring_your_own_policies).
+> [!NOTE]
+> For lerobot 0.4.0, if you want to install pi tags, you will have to do: `pip install "lerobot[pi]@git+https://github.com/huggingface/lerobot.git"`.
+>
+> This will be solved in the next patch release
 
-## Inference & Evaluation
+### Weights & Biases
 
-Evaluate your policies in simulation or on real hardware using the unified evaluation script. LeRobot supports standard benchmarks like **LIBERO**, **MetaWorld** and more to come.
+To use [Weights and Biases](https://docs.wandb.ai/quickstart) for experiment tracking, log in with
 
 ```bash
-# Evaluate a policy on the LIBERO benchmark
-lerobot-eval \
-  --policy.path=lerobot/pi0_libero_finetuned \
-  --env.type=libero \
-  --env.task=libero_object \
-  --eval.n_episodes=10
+wandb login
 ```
 
-Learn how to implement your own simulation environment or benchmark and distribute it from the HF Hub by following the [EnvHub Documentation](https://huggingface.co/docs/lerobot/envhub)
+(note: you will also need to enable WandB in the configuration. See below.)
 
-## Resources
+### Visualize datasets
 
-- **[Documentation](https://huggingface.co/docs/lerobot/index):** The complete guide to tutorials & API.
-- **[Chinese Tutorials: LeRobot+SO-ARM101中文教程-同济子豪兄](https://zihao-ai.feishu.cn/wiki/space/7589642043471924447)** Detailed doc for assembling, teleoperate, dataset, train, deploy. Verified by Seed Studio and 5 global hackathon players.
-- **[Discord](https://discord.gg/q8Dzzpym3f):** Join the `LeRobot` server to discuss with the community.
-- **[X](https://x.com/LeRobotHF):** Follow us on X to stay up-to-date with the latest developments.
-- **[Robot Learning Tutorial](https://huggingface.co/spaces/lerobot/robot-learning-tutorial):** A free, hands-on course to learn robot learning using LeRobot.
+Check out [example 1](https://github.com/huggingface/lerobot/blob/main/examples/dataset/load_lerobot_dataset.py) that illustrates how to use our dataset class which automatically downloads data from the Hugging Face hub.
+
+You can also locally visualize episodes from a dataset on the hub by executing our script from the command line:
+
+```bash
+lerobot-dataset-viz \
+    --repo-id lerobot/pusht \
+    --episode-index 0
+```
+
+or from a dataset in a local folder with the `root` option and the `--mode local` (in the following case the dataset will be searched for in `./my_local_data_dir/lerobot/pusht`)
+
+```bash
+lerobot-dataset-viz \
+    --repo-id lerobot/pusht \
+    --root ./my_local_data_dir \
+    --mode local \
+    --episode-index 0
+```
+
+It will open `rerun.io` and display the camera streams, robot states and actions, like this:
+
+https://github-production-user-asset-6210df.s3.amazonaws.com/4681518/328035972-fd46b787-b532-47e2-bb6f-fd536a55a7ed.mov?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAVCODYLSA53PQK4ZA%2F20240505%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20240505T172924Z&X-Amz-Expires=300&X-Amz-Signature=d680b26c532eeaf80740f08af3320d22ad0b8a4e4da1bcc4f33142c15b509eda&X-Amz-SignedHeaders=host&actor_id=24889239&key_id=0&repo_id=748713144
+
+Our script can also visualize datasets stored on a distant server. See `lerobot-dataset-viz --help` for more instructions.
+
+### The `LeRobotDataset` format
+
+A dataset in `LeRobotDataset` format is very simple to use. It can be loaded from a repository on the Hugging Face hub or a local folder simply with e.g. `dataset = LeRobotDataset("lerobot/aloha_static_coffee")` and can be indexed into like any Hugging Face and PyTorch dataset. For instance `dataset[0]` will retrieve a single temporal frame from the dataset containing observation(s) and an action as PyTorch tensors ready to be fed to a model.
+
+A specificity of `LeRobotDataset` is that, rather than retrieving a single frame by its index, we can retrieve several frames based on their temporal relationship with the indexed frame, by setting `delta_timestamps` to a list of relative times with respect to the indexed frame. For example, with `delta_timestamps = {"observation.image": [-1, -0.5, -0.2, 0]}` one can retrieve, for a given index, 4 frames: 3 "previous" frames 1 second, 0.5 seconds, and 0.2 seconds before the indexed frame, and the indexed frame itself (corresponding to the 0 entry). See example [1_load_lerobot_dataset.py](https://github.com/huggingface/lerobot/blob/main/examples/dataset/load_lerobot_dataset.py) for more details on `delta_timestamps`.
+
+Under the hood, the `LeRobotDataset` format makes use of several ways to serialize data which can be useful to understand if you plan to work more closely with this format. We tried to make a flexible yet simple dataset format that would cover most type of features and specificities present in reinforcement learning and robotics, in simulation and in real-world, with a focus on cameras and robot states but easily extended to other types of sensory inputs as long as they can be represented by a tensor.
+
+Here are the important details and internal structure organization of a typical `LeRobotDataset` instantiated with `dataset = LeRobotDataset("lerobot/aloha_static_coffee")`. The exact features will change from dataset to dataset but not the main aspects:
+
+```
+dataset attributes:
+  ├ hf_dataset: a Hugging Face dataset (backed by Arrow/parquet). Typical features example:
+  │  ├ observation.images.cam_high (VideoFrame):
+  │  │   VideoFrame = {'path': path to a mp4 video, 'timestamp' (float32): timestamp in the video}
+  │  ├ observation.state (list of float32): position of an arm joints (for instance)
+  │  ... (more observations)
+  │  ├ action (list of float32): goal position of an arm joints (for instance)
+  │  ├ episode_index (int64): index of the episode for this sample
+  │  ├ frame_index (int64): index of the frame for this sample in the episode ; starts at 0 for each episode
+  │  ├ timestamp (float32): timestamp in the episode
+  │  ├ next.done (bool): indicates the end of an episode ; True for the last frame in each episode
+  │  └ index (int64): general index in the whole dataset
+  ├ meta: a LeRobotDatasetMetadata object containing:
+  │  ├ info: a dictionary of metadata on the dataset
+  │  │  ├ codebase_version (str): this is to keep track of the codebase version the dataset was created with
+  │  │  ├ fps (int): frame per second the dataset is recorded/synchronized to
+  │  │  ├ features (dict): all features contained in the dataset with their shapes and types
+  │  │  ├ total_episodes (int): total number of episodes in the dataset
+  │  │  ├ total_frames (int): total number of frames in the dataset
+  │  │  ├ robot_type (str): robot type used for recording
+  │  │  ├ data_path (str): formattable string for the parquet files
+  │  │  └ video_path (str): formattable string for the video files (if using videos)
+  │  ├ episodes: a DataFrame containing episode metadata with columns:
+  │  │  ├ episode_index (int): index of the episode
+  │  │  ├ tasks (list): list of tasks for this episode
+  │  │  ├ length (int): number of frames in this episode
+  │  │  ├ dataset_from_index (int): start index of this episode in the dataset
+  │  │  └ dataset_to_index (int): end index of this episode in the dataset
+  │  ├ stats: a dictionary of statistics (max, mean, min, std) for each feature in the dataset, for instance
+  │  │  ├ observation.images.front_cam: {'max': tensor with same number of dimensions (e.g. `(c, 1, 1)` for images, `(c,)` for states), etc.}
+  │  │  └ ...
+  │  └ tasks: a DataFrame containing task information with task names as index and task_index as values
+  ├ root (Path): local directory where the dataset is stored
+  ├ image_transforms (Callable): optional image transformations to apply to visual modalities
+  └ delta_timestamps (dict): optional delta timestamps for temporal queries
+```
+
+A `LeRobotDataset` is serialised using several widespread file formats for each of its parts, namely:
+
+- hf_dataset stored using Hugging Face datasets library serialization to parquet
+- videos are stored in mp4 format to save space
+- metadata are stored in plain json/jsonl files
+
+Dataset can be uploaded/downloaded from the HuggingFace hub seamlessly. To work on a local dataset, you can specify its location with the `root` argument if it's not in the default `~/.cache/huggingface/lerobot` location.
+
+#### Reproduce state-of-the-art (SOTA)
+
+We provide some pretrained policies on our [hub page](https://huggingface.co/lerobot) that can achieve state-of-the-art performances.
+You can reproduce their training by loading the config from their run. Simply running:
+
+```bash
+lerobot-train --config_path=lerobot/diffusion_pusht
+```
+
+reproduces SOTA results for Diffusion Policy on the PushT task.
+
+## Contribute
+
+If you would like to contribute to 🤗 LeRobot, please check out our [contribution guide](https://github.com/huggingface/lerobot/blob/main/CONTRIBUTING.md).
+
+### Add a pretrained policy
+
+Once you have trained a policy you may upload it to the Hugging Face hub using a hub id that looks like `${hf_user}/${repo_name}` (e.g. [lerobot/diffusion_pusht](https://huggingface.co/lerobot/diffusion_pusht)).
+
+You first need to find the checkpoint folder located inside your experiment directory (e.g. `outputs/train/2024-05-05/20-21-12_aloha_act_default/checkpoints/002500`). Within that there is a `pretrained_model` directory which should contain:
+
+- `config.json`: A serialized version of the policy configuration (following the policy's dataclass config).
+- `model.safetensors`: A set of `torch.nn.Module` parameters, saved in [Hugging Face Safetensors](https://huggingface.co/docs/safetensors/index) format.
+- `train_config.json`: A consolidated configuration containing all parameters used for training. The policy configuration should match `config.json` exactly. This is useful for anyone who wants to evaluate your policy or for reproducibility.
+
+To upload these to the hub, run the following:
+
+```bash
+huggingface-cli upload ${hf_user}/${repo_name} path/to/pretrained_model
+```
+
+See [lerobot_eval.py](https://github.com/huggingface/lerobot/blob/main/src/lerobot/scripts/lerobot_eval.py) for an example of how other people may use your policy.
+
+### Acknowledgment
+
+- The LeRobot team 🤗 for building SmolVLA [Paper](https://arxiv.org/abs/2506.01844), [Blog](https://huggingface.co/blog/smolvla).
+- Thanks to Tony Zhao, Zipeng Fu and colleagues for open sourcing ACT policy, ALOHA environments and datasets. Ours are adapted from [ALOHA](https://tonyzhaozh.github.io/aloha) and [Mobile ALOHA](https://mobile-aloha.github.io).
+- Thanks to Cheng Chi, Zhenjia Xu and colleagues for open sourcing Diffusion policy, Pusht environment and datasets, as well as UMI datasets. Ours are adapted from [Diffusion Policy](https://diffusion-policy.cs.columbia.edu) and [UMI Gripper](https://umi-gripper.github.io).
+- Thanks to Nicklas Hansen, Yunhai Feng and colleagues for open sourcing TDMPC policy, Simxarm environments and datasets. Ours are adapted from [TDMPC](https://github.com/nicklashansen/tdmpc) and [FOWM](https://www.yunhaifeng.com/FOWM).
+- Thanks to Antonio Loquercio and Ashish Kumar for their early support.
+- Thanks to [Seungjae (Jay) Lee](https://sjlee.cc/), [Mahi Shafiullah](https://mahis.life/) and colleagues for open sourcing [VQ-BeT](https://sjlee.cc/vq-bet/) policy and helping us adapt the codebase to our repository. The policy is adapted from [VQ-BeT repo](https://github.com/jayLEE0301/vq_bet_official).
 
 ## Citation
 
-If you use LeRobot in your research, please cite:
+If you want, you can cite this work with:
 
 ```bibtex
 @misc{cadene2024lerobot,
@@ -146,14 +339,6 @@ If you use LeRobot in your research, please cite:
 }
 ```
 
-## Contribute
+## Star History
 
-We welcome contributions from everyone in the community! To get started, please read our [CONTRIBUTING.md](./CONTRIBUTING.md) guide. Whether you're adding a new feature, improving documentation, or fixing a bug, your help and feedback are invaluable. We're incredibly excited about the future of open-source robotics and can't wait to work with you on what's next—thank you for your support!
-
-<p align="center">
-  <img alt="SO101 Video" src="./media/readme/so100_video.webp" width="640px">
-</p>
-
-<div align="center">
-<sub>Built by the <a href="https://huggingface.co/lerobot">LeRobot</a> team at <a href="https://huggingface.co">Hugging Face</a> with ❤️</sub>
-</div>
+[![Star History Chart](https://api.star-history.com/svg?repos=huggingface/lerobot&type=Timeline)](https://star-history.com/#huggingface/lerobot&Timeline)

SECURITY.md

@@ -1,48 +0,0 @@
-# Security Policy
-
-## Project Status & Philosophy
-
-`lerobot` has so far been primarily a research and prototyping tool, which is why deployment security hasn’t been a strong focus until now. As `lerobot` continues to be adopted and deployed in production, we are paying much closer attention to these kinds of issues.
-
-Fortunately, being an open-source project, the community can also help by reporting and fixing vulnerabilities. We appreciate your efforts to responsibly disclose your findings and will make every effort to acknowledge your contributions.
-
-## Reporting a Vulnerability
-
-To report a security issue, please use the GitHub Security Advisory ["Report a Vulnerability"](https://github.com/huggingface/lerobot/security/advisories/new) tab.
-
-The `lerobot` team will send a response indicating the next steps in handling your report. After the initial reply to your report, the security team will keep you informed of the progress towards a fix and full announcement, and may ask for additional information or guidance.
-
-#### Hugging Face Security Team
-
-Since this project is part of the Hugging Face ecosystem, feel free to submit vulnerability reports directly to: **[security@huggingface.co](mailto:security@huggingface.co)**. Someone from the HF security team will review the report and recommend next steps.
-
-#### Open Source Disclosures
-
-If reporting a vulnerability specific to the open-source codebase (and not the underlying Hub infrastructure), you may also use [Huntr](https://huntr.com), a vulnerability disclosure program for open source software.
-
-## Supported Versions
-
-Currently, we treat `lerobot` as a rolling release. We prioritize security updates for the latest available version (`main` branch).
-
-| Version  | Supported |
-| -------- | --------- |
-| Latest   | ✅        |
-| < Latest | ❌        |
-
-## Secure Usage Guidelines
-
-`lerobot` is tightly coupled to the Hugging Face Hub for sharing data and pretrained policies. When downloading artifacts uploaded by others, you expose yourself to risks. Please read below for recommendations to keep your runtime and robot environment safe.
-
-### Remote Artefacts (Weights & Policies)
-
-Models and policies uploaded to the Hugging Face Hub come in different formats. We heavily recommend uploading and downloading models in the [`safetensors`](https://github.com/huggingface/safetensors) format.
-
-`safetensors` was developed specifically to prevent arbitrary code execution on your system, which is critical when running software on physical hardware/robots.
-
-To avoid loading models from unsafe formats (e.g., `pickle`), you should ensure you are prioritizing `safetensors` files.
-
-### Remote Code
-
-Some models or environments on the Hub may require `trust_remote_code=True` to run custom architecture code.
-
-Please **always** verify the content of the modeling files when using this argument. We recommend setting a specific `revision` (commit hash) when loading remote code to ensure you protect yourself from unverified updates to the repository.

docker/Dockerfile.internal

@@ -73,7 +73,7 @@ ENV HOME=/home/user_lerobot \
 RUN uv venv --python python${PYTHON_VERSION}
 
 # Install Python dependencies for caching
-COPY --chown=user_lerobot:user_lerobot setup.py pyproject.toml README.md MANIFEST.in ./
+COPY --chown=user_lerobot:user_lerobot pyproject.toml README.md MANIFEST.in ./
 COPY --chown=user_lerobot:user_lerobot src/ src/
 
 ARG UNBOUND_DEPS=false

docker/Dockerfile.user

@@ -59,7 +59,7 @@ ENV HOME=/home/user_lerobot \
 RUN uv venv
 
 # Install Python dependencies for caching
-COPY --chown=user_lerobot:user_lerobot setup.py pyproject.toml README.md MANIFEST.in ./
+COPY --chown=user_lerobot:user_lerobot pyproject.toml README.md MANIFEST.in ./
 COPY --chown=user_lerobot:user_lerobot src/ src/
 
 ARG UNBOUND_DEPS=false

docs/source/_toctree.yml

@@ -9,8 +9,6 @@
     title: Imitation Learning for Robots
   - local: cameras
     title: Cameras
-  - local: bring_your_own_policies
-    title: Bring Your Own Policies
   - local: integrate_hardware
     title: Bring Your Own Hardware
   - local: hilserl
@@ -19,8 +17,6 @@
     title: Train RL in Simulation
   - local: multi_gpu_training
     title: Multi GPU training
-  - local: peft_training
-    title: Training with PEFT (e.g., LoRA)
   title: "Tutorials"
 - sections:
   - local: lerobot-dataset-v3
@@ -37,21 +33,11 @@
     title: SmolVLA
   - local: pi0
     title: π₀ (Pi0)
-  - local: pi0fast
-    title: π₀-FAST (Pi0Fast)
   - local: pi05
     title: π₀.₅ (Pi05)
   - local: groot
     title: NVIDIA GR00T N1.5
-  - local: xvla
-    title: X-VLA
-  - local: walloss
-    title: WALL-OSS
   title: "Policies"
-- sections:
-  - local: sarm
-    title: SARM
-  title: "Reward Models"
 - sections:
   - local: async
     title: Use Async Inference
@@ -63,8 +49,6 @@
     title: Environments from the Hub
   - local: envhub_leisaac
     title: Control & Train Robots in Sim (LeIsaac)
-  - local: envhub_isaaclab_arena
-    title: NVIDIA IsaacLab Arena Environments
   - local: libero
     title: Using Libero
   - local: metaworld
@@ -97,17 +81,11 @@
     title: Reachy 2
   - local: unitree_g1
     title: Unitree G1
-  - local: earthrover_mini_plus
-    title: Earth Rover Mini
   title: "Robots"
 - sections:
   - local: phone_teleop
     title: Phone
   title: "Teleoperators"
-- sections:
-  - local: torch_accelerators
-    title: PyTorch accelerators
-  title: "Supported Hardware"
 - sections:
   - local: notebooks
     title: Notebooks

docs/source/async.mdx

@@ -169,7 +169,7 @@ python -m lerobot.async_inference.robot_client \
 <!-- prettier-ignore-start -->
 ```python
 import threading
-from lerobot.robots.so_follower import SO100FollowerConfig
+from lerobot.robots.so100_follower import SO100FollowerConfig
 from lerobot.cameras.opencv.configuration_opencv import OpenCVCameraConfig
 from lerobot.async_inference.configs import RobotClientConfig
 from lerobot.async_inference.robot_client import RobotClient
@@ -195,7 +195,6 @@ client_cfg = RobotClientConfig(
     robot=robot_cfg,
     server_address="localhost:8080",
     policy_device="mps",
-    client_device="cpu",
     policy_type="smolvla",
     pretrained_name_or_path="<user>/smolvla_async",
     chunk_size_threshold=0.5,
@@ -279,7 +278,7 @@ We found the default values of `actions_per_chunk` and `chunk_size_threshold` to
 2. **Adjust your `fps` based on inference latency.** While the server generates a new action chunk, the client is not idle and is stepping through its current action queue. If the two processes happen at fundamentally different speeds, the client might end up with an empty queue. As such, you should reduce your fps if you consistently run out of actions in queue.
 3. **Adjust `chunk_size_threshold`**.
    - Values closer to `0.0` result in almost sequential behavior. Values closer to `1.0` → send observation every step (more bandwidth, relies on good world-model).
-   - We found values around 0.5-0.6 to work well. If you want to tweak this, spin up a `RobotClient` setting the `--debug_visualize_queue_size` to `True`. This will plot the action queue size evolution at runtime, and you can use it to find the value of `chunk_size_threshold` that works best for your setup.
+   - We found values around 0.5-0.6 to work well. If you want to tweak this, spin up a `RobotClient` setting the `--debug-visualize-queue-size` to `True`. This will plot the action queue size evolution at runtime, and you can use it to find the value of `chunk_size_threshold` that works best for your setup.
 
 <p align="center">
   <img
@@ -290,7 +289,7 @@ We found the default values of `actions_per_chunk` and `chunk_size_threshold` to
 <p align="center">
   <i>
     The action queue size is plotted at runtime when the
-    `--debug_visualize_queue_size` flag is passed, for various levels of
+    `--debug-visualize-queue-size` flag is passed, for various levels of
     `chunk_size_threshold` (`g` in the SmolVLA paper).
   </i>
 </p>

docs/source/bring_your_own_policies.mdx

@@ -1,175 +0,0 @@
-# Bring Your Own Policies
-
-This tutorial explains how to integrate your own custom policy implementations into the LeRobot ecosystem, allowing you to leverage all LeRobot tools for training, evaluation, and deployment while using your own algorithms.
-
-## Step 1: Create a Policy Package
-
-Your custom policy should be organized as an installable Python package following LeRobot's plugin conventions.
-
-### Package Structure
-
-Create a package with the prefix `lerobot_policy_` (IMPORTANT!) followed by your policy name:
-
-```bash
-lerobot_policy_my_custom_policy/
-├── pyproject.toml
-└── src/
-    └── lerobot_policy_my_custom_policy/
-        ├── __init__.py
-        ├── configuration_my_custom_policy.py
-        ├── modeling_my_custom_policy.py
-        └── processor_my_custom_policy.py
-```
-
-### Package Configuration
-
-Set up your `pyproject.toml`:
-
-```toml
-[project]
-name = "lerobot_policy_my_custom_policy"
-version = "0.1.0"
-dependencies = [
-    # your policy-specific dependencies
-]
-requires-python = ">= 3.11"
-
-[build-system]
-build-backend = # your-build-backend
-requires = # your-build-system
-```
-
-## Step 2: Define the Policy Configuration
-
-Create a configuration class that inherits from `PreTrainedConfig` and registers your policy type:
-
-```python
-# configuration_my_custom_policy.py
-from dataclasses import dataclass, field
-from lerobot.configs.policies import PreTrainedConfig
-from lerobot.configs.types import NormalizationMode
-
-@PreTrainedConfig.register_subclass("my_custom_policy")
-@dataclass
-class MyCustomPolicyConfig(PreTrainedConfig):
-    """Configuration class for MyCustomPolicy.
-
-    Args:
-        n_obs_steps: Number of observation steps to use as input
-        horizon: Action prediction horizon
-        n_action_steps: Number of action steps to execute
-        hidden_dim: Hidden dimension for the policy network
-        # Add your policy-specific parameters here
-    """
-    # ...PreTrainedConfig fields...
-    pass
-
-    def __post_init__(self):
-        super().__post_init__()
-        # Add any validation logic here
-
-    def validate_features(self) -> None:
-        """Validate input/output feature compatibility."""
-        # Implement validation logic for your policy's requirements
-        pass
-```
-
-## Step 3: Implement the Policy Class
-
-Create your policy implementation by inheriting from LeRobot's base `PreTrainedPolicy` class:
-
-```python
-# modeling_my_custom_policy.py
-import torch
-import torch.nn as nn
-from typing import Dict, Any
-
-from lerobot.policies.pretrained import PreTrainedPolicy
-from .configuration_my_custom_policy import MyCustomPolicyConfig
-
-class MyCustomPolicy(PreTrainedPolicy):
-    config_class = MyCustomPolicyConfig
-    name = "my_custom_policy"
-
-    def __init__(self, config: MyCustomPolicyConfig, dataset_stats: Dict[str, Any] = None):
-        super().__init__(config, dataset_stats)
-        ...
-```
-
-## Step 4: Add Data Processors
-
-Create processor functions:
-
-```python
-# processor_my_custom_policy.py
-from typing import Dict, Any
-import torch
-
-
-def make_my_custom_policy_pre_post_processors(
-    config,
-) -> tuple[
-    PolicyProcessorPipeline[dict[str, Any], dict[str, Any]],
-    PolicyProcessorPipeline[PolicyAction, PolicyAction],
-]:
-    """Create preprocessing and postprocessing functions for your policy."""
-    pass  # Define your preprocessing and postprocessing logic here
-
-```
-
-## Step 5: Package Initialization
-
-Expose your classes in the package's `__init__.py`:
-
-```python
-# __init__.py
-"""Custom policy package for LeRobot."""
-
-try:
-    import lerobot  # noqa: F401
-except ImportError:
-    raise ImportError(
-        "lerobot is not installed. Please install lerobot to use this policy package."
-    )
-
-from .configuration_my_custom_policy import MyCustomPolicyConfig
-from .modeling_my_custom_policy import MyCustomPolicy
-from .processor_my_custom_policy import make_my_custom_policy_pre_post_processors
-
-__all__ = [
-    "MyCustomPolicyConfig",
-    "MyCustomPolicy",
-    "make_my_custom_policy_pre_post_processors",
-]
-```
-
-## Step 6: Installation and Usage
-
-### Install Your Policy Package
-
-```bash
-cd lerobot_policy_my_custom_policy
-pip install -e .
-
-# Or install from PyPI if published
-pip install lerobot_policy_my_custom_policy
-```
-
-### Use Your Policy
-
-Once installed, your policy automatically integrates with LeRobot's training and evaluation tools:
-
-```bash
-lerobot-train \
-    --policy.type my_custom_policy \
-    --env.type pusht \
-    --steps 200000
-```
-
-## Examples and Community Contributions
-
-Check out these example policy implementations:
-
-- [DiTFlow Policy](https://github.com/danielsanjosepro/lerobot_policy_ditflow) - Diffusion Transformer policy with flow-matching objective. Try it out in this example: [DiTFlow Example](https://github.com/danielsanjosepro/test_lerobot_policy_ditflow)
-
-Share your policy implementations with the community! 🤗

docs/source/earthrover_mini_plus.mdx

@@ -1,225 +0,0 @@
-# EarthRover Mini Plus
-
-The EarthRover Mini Plus is a fully open source mobile robot that connects through the cloud using the Frodobots SDK. This lets you control the robot and record datasets for training AI models.
-
-## What You Need
-
-### Hardware
-
-- EarthRover Mini robot
-- Computer with Python 3.10 or newer
-- Internet connection
-
-### Setting Up the Frodobots SDK
-
-The robot needs the [Frodobots SDK](https://github.com/frodobots-org/earth-rovers-sdk) running on your computer. Here's how:
-
-1. Download and install the SDK:
-
-```bash
-git clone https://github.com/frodobots-org/earth-rovers-sdk.git
-cd earth-rovers-sdk
-pip install -r requirements.txt
-```
-
-2. Save Credentials:
-
-Write your .env variables with the SDK API key and bot name provided by the Frodobots team.
-
-```bash
-SDK_API_TOKEN=your_sdk_api_token_here
-BOT_SLUG=your_bot_slug_here
-CHROME_EXECUTABLE_PATH=/path/to/chrome_or_chromium
-# Default value is MAP_ZOOM_LEVEL=18 https://wiki.openstreetmap.org/wiki/Zoom_levels
-MAP_ZOOM_LEVEL=18
-MISSION_SLUG=your_mission_slug_here
-# Image quality between 0.1 and 1.0 (default: 0.8)
-# Recommended: 0.8 for better performance
-IMAGE_QUALITY=0.8
-# Image format: jpeg, png or webp (default: png)
-# Recommended: jpeg for better performance and lower bandwidth usage
-IMAGE_FORMAT=jpeg
-```
-
-3. Start the SDK:
-
-```bash
-hypercorn main:app --reload
-```
-
-4. Open your web browser and go to `http://localhost:8000`, then click "Join"
-
-The SDK gives you:
-
-- Live video from front and rear cameras
-
-> [!IMPORTANT]
-> The SDK must be running before you can use the robot.
-
-## Install LeRobot
-
-Follow our [Installation Guide](./installation) to install LeRobot.
-
-In addition to the base installation, install the EarthRover Mini dependencies:
-
-```bash
-pip install -e .
-```
-
-## How It Works
-
-The robot uses the internet to communicate:
-
-- **Movement commands**: Sent through the SDK
-- **Camera video**: Received from the SDK
-- **Robot info**: Battery, location, speed from the SDK
-
-You don't need to plug anything in - it all works through the SDK.
-
-## Calibration
-
-No calibration needed! The robot is ready to use as soon as the SDK is running.
-
-## Controlling the Robot
-
-You control the robot using your keyboard - just like playing a video game with WASD keys.
-
-### Keyboard Controls
-
-| Key | Action                           |
-| --- | -------------------------------- |
-| W   | Move forward                     |
-| S   | Move backward                    |
-| A   | Turn left (with forward motion)  |
-| D   | Turn right (with forward motion) |
-| Q   | Rotate left in place             |
-| E   | Rotate right in place            |
-| X   | Stop all movement                |
-| +/= | Increase speed                   |
-| -   | Decrease speed                   |
-| ESC | Disconnect                       |
-
-### Speed Settings
-
-You can adjust how fast the robot moves:
-
-- **Forward/backward speed**: Default is full speed (1.0)
-- **Turning speed**: Default is full speed (1.0)
-- **Speed changes**: Use +/- keys to adjust by 0.1 each time
-
-### Try It Out
-
-Test driving the robot before recording data:
-
-```python
-from lerobot.robots.earthrover_mini_plus import EarthRoverMiniPlus, EarthRoverMiniPlusConfig
-from lerobot.teleoperators.keyboard import KeyboardRoverTeleop, KeyboardRoverTeleopConfig
-
-# Initialize robot
-robot_config = EarthRoverMiniPlusConfig()
-robot = EarthRoverMiniPlus(robot_config)
-
-# Initialize teleoperator
-teleop_config = KeyboardRoverTeleopConfig(
-    linear_speed=1.0,
-    angular_speed=1.0,
-    speed_increment=0.1
-)
-teleop = KeyboardRoverTeleop(teleop_config)
-
-# Connect
-robot.connect()
-teleop.connect()
-
-# Teleoperate (use keyboard controls)
-try:
-    while True:
-        action = teleop.get_action()
-        robot.send_action(action)
-except KeyboardInterrupt:
-    pass
-finally:
-    robot.disconnect()
-    teleop.disconnect()
-```
-
-> [!TIP]
-> If you're using a Mac, you might need to give Terminal permission to access your keyboard for teleoperation. Go to System Preferences > Security & Privacy > Input Monitoring and check the box for Terminal.
-
-## Recording Data
-
-Once you can drive the robot well, you can start recording data to train AI models. The system records:
-
-- **What you do**: How you move the robot (forward, backward, turning)
-- **What the robot sees**:
-  - Videos from both cameras
-  - Robot speed and direction
-  - Battery level and location
-  - GPS position and signal
-  - Other sensor data
-- **When it happened**: Timestamps for everything
-
-### Setting Up Hugging Face
-
-We use Hugging Face to store your data online. First, log in with your token from [Hugging Face settings](https://huggingface.co/settings/tokens):
-
-```bash
-huggingface-cli login --token ${HUGGINGFACE_TOKEN} --add-to-git-credential
-```
-
-Store your Hugging Face username:
-
-```bash
-HF_USER=$(huggingface-cli whoami | head -n 1)
-echo $HF_USER
-```
-
-### Start Recording
-
-Use the standard recording command:
-
-```bash
-python src/lerobot/scripts/lerobot_record.py \
-    --robot.type=earthrover_mini_plus \
-    --teleop.type=keyboard_rover \
-    --dataset.repo_id=your_username/dataset_name \
-    --dataset.num_episodes=2 \
-    --dataset.fps=10 \
-    --dataset.single_task="Navigate around obstacles" \
-    --display_data=true
-```
-
-Replace `your_username/dataset_name` with your Hugging Face username and a name for your dataset.
-
-### What Gets Saved
-
-Your dataset includes:
-
-**Your Actions (2 things)**:
-
-- How much you moved forward/backward
-- How much you turned left/right
-
-**Robot Observations (12 things)**:
-
-- Front camera video
-- Rear camera video
-- Current speed
-- Battery level
-- Which way the robot is facing
-- GPS location (latitude, longitude, signal strength)
-- Network signal strength
-- Vibration level
-- Lamp status (on/off)
-
-### Where Your Data Goes
-
-On your computer: `~/.cache/huggingface/lerobot/{repo-id}`
-
-After recording, your data automatically uploads to your Hugging Face page:
-
-```bash
-echo https://huggingface.co/datasets/${HF_USER}/earthrover-navigation
-```
-
-Your dataset will be tagged with `LeRobot` for community discovery.

docs/source/envhub.mdx

@@ -2,32 +2,14 @@
 
 The **EnvHub** feature allows you to load simulation environments directly from the Hugging Face Hub with a single line of code. This unlocks a powerful new model for collaboration: instead of environments being locked away inside monolithic libraries, anyone can publish custom environments and share them with the community.
 
-## What is EnvHub?
+## Overview
 
-EnvHub lets you create custom robotics simulation environments with your own robot models and scenarios, and make them easily usable by anyone through the LeRobot framework.
+With EnvHub, you can:
 
-EnvHub packages are stored on the Hugging Face Hub, and can be seamlessly pulled and used in your AI robotics projects through LeRobot with a single line of code.
-
-Thanks to EnvHub, you can:
-
-1. **Create and publish environments** to the Hugging Face Hub as Git repositories, and distribute complex physics simulations without packaging hassles
-2. **Load environments** dynamically, without installing them as packages
-3. **Version and track** environment changes using Git semantics
-4. **Discover** new simulation tasks shared by the community
-
-This design means you can go from discovering an interesting environment on the Hub to running experiments in seconds, or create your own custom robot and environment without worrying about dependency conflicts or complex installation procedures.
-
-When you create an EnvHub package, you can build anything you want inside it and use any simulation tool you like: this is your own space to play with. The only requirement is that the package contains an `env.py` file that defines the environment and allows LeRobot to load and use your EnvHub package.
-
-This `env.py` file needs to expose a small API so LeRobot can load and run it. In particular, you must provide a `make_env(n_envs: int = 1, use_async_envs: bool = False)` or `make_env(n_envs: int = 1, use_async_envs: bool = False, cfg: EnvConfig)` function, which is the main entry point for LeRobot. It should return one of:
-
-- A `gym.vector.VectorEnv` (most common)
-- A single `gym.Env` (will be automatically wrapped)
-- A dict mapping `{suite_name: {task_id: VectorEnv}}` (for multi-task benchmarks)
-
-You can also pass an `EnvConfig` object to `make_env` to configure the environment (e.g. the number of environments, task, camera name, initial states, control mode, episode length, etc.).
-
-Finally, your environment must implement the standard `gym.vector.VectorEnv` interface so it works with LeRobot, including methods like `reset` and `step`.
+- Load environments from the Hub instantly
+- Share your custom simulation tasks with the community
+- Version control your environments using Git
+- Distribute complex physics simulations without packaging hassles
 
 ## Quick Start
 
@@ -47,6 +29,17 @@ env = make_env("lerobot/cartpole-env", trust_remote_code=True)
   hash for reproducibility and security.
 </Tip>
 
+## What is EnvHub?
+
+EnvHub is a framework that allows researchers and developers to:
+
+1. **Publish environments** to the Hugging Face Hub as Git repositories
+2. **Load environments** dynamically without installing them as packages
+3. **Version and track** environment changes using Git semantics
+4. **Discover** new simulation tasks shared by the community
+
+This design means you can go from discovering an interesting environment on the Hub to running experiments in seconds, without worrying about dependency conflicts or complex installation procedures.
+
 ## Repository Structure
 
 To make your environment loadable from the Hub, your repository must contain at minimum:

docs/source/envhub_isaaclab_arena.mdx

@@ -1,510 +0,0 @@
-# NVIDIA IsaacLab Arena & LeRobot
-
-LeRobot EnvHub now supports **GPU-accelerated simulation** with IsaacLab Arena for policy evaluation at scale.
-Train and evaluate imitation learning policies with high-fidelity simulation — all integrated into the LeRobot ecosystem.
-
-<img
-  src="https://huggingface.co/nvidia/isaaclab-arena-envs/resolve/main/assets/Gr1OpenMicrowaveEnvironment.png"
-  alt="IsaacLab Arena - GR1 Microwave Environment"
-  style={{ maxWidth: "100%", borderRadius: "8px", marginBottom: "1rem" }}
-/>
-
-[IsaacLab Arena](https://github.com/isaac-sim/IsaacLab-Arena) integrates with NVIDIA IsaacLab to provide:
-
-- 🤖 **Humanoid embodiments**: GR1, G1, Galileo with various configurations
-- 🎯 **Manipulation & loco-manipulation tasks**: Door opening, pick-and-place, button pressing, and more
-- ⚡ **GPU-accelerated rollouts**: Parallel environment execution on NVIDIA GPUs
-- 🖼️ **RTX Rendering**: Evaluate vision-based policies with realistic rendering, reflections and refractions
-- 📦 **LeRobot-compatible datasets**: Ready for training with GR00T N1x, PI0, SmolVLA, ACT, and Diffusion policies
-- 🔄 **EnvHub integration**: Load environments from HuggingFace EnvHub with one line
-
-## Installation
-
-### Prerequisites
-
-Hardware requirements are shared with Isaac Sim, and are detailed in [Isaac Sim Requirements](https://docs.isaacsim.omniverse.nvidia.com/5.1.0/installation/requirements.html).
-
-- NVIDIA GPU with CUDA support
-- NVIDIA driver compatible with IsaacSim 5.1.0
-- Linux (Ubuntu 22.04 / 24.04)
-
-### Setup
-
-```bash
-# 1. Create conda environment
-conda create -y -n lerobot-arena python=3.11
-conda activate lerobot-arena
-conda install -y -c conda-forge ffmpeg=7.1.1
-
-# 2. Install Isaac Sim 5.1.0
-pip install "isaacsim[all,extscache]==5.1.0" --extra-index-url https://pypi.nvidia.com
-
-# Accept NVIDIA EULA (required)
-export ACCEPT_EULA=Y
-export PRIVACY_CONSENT=Y
-
-# 3. Install IsaacLab 2.3.0
-git clone https://github.com/isaac-sim/IsaacLab.git
-cd IsaacLab
-git checkout v2.3.0
-./isaaclab.sh -i
-cd ..
-
-# 4. Install IsaacLab Arena
-git clone https://github.com/isaac-sim/IsaacLab-Arena.git
-cd IsaacLab-Arena
-git checkout release/0.1.1
-pip install -e .
-cd ..
-
-
-# 5. Install LeRobot
-git clone https://github.com/huggingface/lerobot.git
-cd lerobot
-pip install -e .
-cd ..
-
-
-# 6. Install additional dependencies
-pip install onnxruntime==1.23.2 lightwheel-sdk==1.0.1 vuer[all]==0.0.70 qpsolvers==4.8.1
-pip install numpy==1.26.0 # Isaac Sim 5.1 depends on numpy==1.26.0, this will be fixed in next release
-```
-
-## Evaluating Policies
-
-### Pre-trained Policies
-
-The following trained policies are available:
-
-| Policy                      | Architecture | Task          | Link                                                                     |
-| :-------------------------- | :----------- | :------------ | :----------------------------------------------------------------------- |
-| pi05-arena-gr1-microwave    | PI0.5        | GR1 Microwave | [HuggingFace](https://huggingface.co/nvidia/pi05-arena-gr1-microwave)    |
-| smolvla-arena-gr1-microwave | SmolVLA      | GR1 Microwave | [HuggingFace](https://huggingface.co/nvidia/smolvla-arena-gr1-microwave) |
-
-### Evaluate SmolVLA
-
-```bash
-pip install -e ".[smolvla]"
-pip install numpy==1.26.0 # revert numpy to version 1.26
-```
-
-```bash
-lerobot-eval \
-    --policy.path=nvidia/smolvla-arena-gr1-microwave \
-    --env.type=isaaclab_arena \
-    --env.hub_path=nvidia/isaaclab-arena-envs \
-    --rename_map='{"observation.images.robot_pov_cam_rgb": "observation.images.robot_pov_cam"}' \
-    --policy.device=cuda \
-    --env.environment=gr1_microwave \
-    --env.embodiment=gr1_pink \
-    --env.object=mustard_bottle \
-    --env.headless=false \
-    --env.enable_cameras=true \
-    --env.video=true \
-    --env.video_length=10 \
-    --env.video_interval=15 \
-    --env.state_keys=robot_joint_pos \
-    --env.camera_keys=robot_pov_cam_rgb \
-    --trust_remote_code=True \
-    --eval.batch_size=1
-```
-
-### Evaluate PI0.5
-
-```bash
-pip install -e ".[pi]"
-pip install numpy==1.26.0 # revert numpy to version 1.26
-```
-
-<Tip>PI0.5 requires disabling torch compile for evaluation:</Tip>
-
-```bash
-TORCH_COMPILE_DISABLE=1 TORCHINDUCTOR_DISABLE=1 lerobot-eval \
-    --policy.path=nvidia/pi05-arena-gr1-microwave \
-    --env.type=isaaclab_arena \
-    --env.hub_path=nvidia/isaaclab-arena-envs \
-    --rename_map='{"observation.images.robot_pov_cam_rgb": "observation.images.robot_pov_cam"}' \
-    --policy.device=cuda \
-    --env.environment=gr1_microwave \
-    --env.embodiment=gr1_pink \
-    --env.object=mustard_bottle \
-    --env.headless=false \
-    --env.enable_cameras=true \
-    --env.video=true \
-    --env.video_length=15 \
-    --env.video_interval=15 \
-    --env.state_keys=robot_joint_pos \
-    --env.camera_keys=robot_pov_cam_rgb \
-    --trust_remote_code=True \
-    --eval.batch_size=1
-```
-
-<Tip>
-  To change the number of parallel environments, use the ```--eval.batch_size```
-  flag.
-</Tip>
-
-### What to Expect
-
-During evaluation, you will see a progress bar showing the running success rate:
-
-```
-Stepping through eval batches:   8%|██████▍    | 4/50 [00:45<08:06, 10.58s/it, running_success_rate=25.0%]
-```
-
-### Video Recording
-
-To enable video recording during evaluation, add the following flags to your command:
-
-```bash
---env.video=true \
---env.video_length=15 \
---env.video_interval=15
-```
-
-For more details on video recording, see the [IsaacLab Recording Documentation](https://isaac-sim.github.io/IsaacLab/main/source/how-to/record_video.html).
-
-<Tip>
-When running headless with `--env.headless=true`, you must also enable cameras explicitly for camera enabled environments:
-
-```bash
---env.headless=true --env.enable_cameras=true
-```
-
-</Tip>
-
-### Output Directory
-
-Evaluation videos are saved to the output directory with the following structure:
-
-```
-outputs/eval/<date>/<timestamp>_<env>_<policy>/videos/<task>_<env_id>/eval_episode_<n>.mp4
-```
-
-For example:
-
-```
-outputs/eval/2026-01-02/14-38-01_isaaclab_arena_smolvla/videos/gr1_microwave_0/eval_episode_0.mp4
-```
-
-## Training Policies
-
-To learn more about training policies with LeRobot, please refer to the training documentation:
-
-- [SmolVLA](./smolvla)
-- [Pi0.5](./pi05)
-- [GR00T N1.5](./groot)
-
-Sample IsaacLab Arena datasets are available on HuggingFace Hub for experimentation:
-
-| Dataset                                                                                                   | Description                | Frames |
-| :-------------------------------------------------------------------------------------------------------- | :------------------------- | :----- |
-| [Arena-GR1-Manipulation-Task](https://huggingface.co/datasets/nvidia/Arena-GR1-Manipulation-Task-v3)      | GR1 microwave manipulation | ~4K    |
-| [Arena-G1-Loco-Manipulation-Task](https://huggingface.co/datasets/nvidia/Arena-G1-Loco-Manipulation-Task) | G1 loco-manipulation       | ~4K    |
-
-## Environment Configuration
-
-### Full Configuration Options
-
-```python
-from lerobot.envs.configs import IsaaclabArenaEnv
-
-config = IsaaclabArenaEnv(
-    # Environment selection
-    environment="gr1_microwave",      # Task environment
-    embodiment="gr1_pink",            # Robot embodiment
-    object="power_drill",             # Object to manipulate
-
-    # Simulation settings
-    episode_length=300,               # Max steps per episode
-    headless=True,                    # Run without GUI
-    device="cuda:0",                  # GPU device
-    seed=42,                          # Random seed
-
-    # Observation configuration
-    state_keys="robot_joint_pos",     # State observation keys (comma-separated)
-    camera_keys="robot_pov_cam_rgb",  # Camera observation keys (comma-separated)
-    state_dim=54,                     # Expected state dimension
-    action_dim=36,                    # Expected action dimension
-    camera_height=512,                # Camera image height
-    camera_width=512,                 # Camera image width
-    enable_cameras=True,              # Enable camera observations
-
-    # Video recording
-    video=False,                      # Enable video recording
-    video_length=100,                 # Frames per video
-    video_interval=200,               # Steps between recordings
-
-    # Advanced
-    mimic=False,                      # Enable mimic mode
-    teleop_device=None,               # Teleoperation device
-    disable_fabric=False,             # Disable fabric optimization
-    enable_pinocchio=True,            # Enable Pinocchio for IK
-)
-```
-
-### Using Environment Hub directly for advanced usage
-
-Create a file called `test_env_load_arena.py` or [download from the EnvHub](https://huggingface.co/nvidia/isaaclab-arena-envs/blob/main/tests/test_env_load_arena.py):
-
-```python
-import logging
-from dataclasses import asdict
-from pprint import pformat
-import torch
-import tqdm
-from lerobot.configs import parser
-from lerobot.configs.eval import EvalPipelineConfig
-
-
-@parser.wrap()
-def main(cfg: EvalPipelineConfig):
-    """Run random action rollout for IsaacLab Arena environment."""
-    logging.info(pformat(asdict(cfg)))
-
-    from lerobot.envs.factory import make_env
-
-    env_dict = make_env(
-        cfg.env,
-        n_envs=cfg.env.num_envs,
-        trust_remote_code=True,
-    )
-    env = next(iter(env_dict.values()))[0]
-    env.reset()
-    for _ in tqdm.tqdm(range(cfg.env.episode_length)):
-        with torch.inference_mode():
-            actions = env.action_space.sample()
-            obs, rewards, terminated, truncated, info = env.step(actions)
-            if terminated.any() or truncated.any():
-                obs, info = env.reset()
-    env.close()
-
-
-if __name__ == "__main__":
-    main()
-```
-
-Run with:
-
-```bash
-python test_env_load_arena.py \
-    --env.environment=g1_locomanip_pnp \
-    --env.embodiment=gr1_pink \
-    --env.object=cracker_box \
-    --env.num_envs=4 \
-    --env.enable_cameras=true \
-    --env.seed=1000 \
-    --env.video=true \
-    --env.video_length=10 \
-    --env.video_interval=15 \
-    --env.headless=false \
-    --env.hub_path=nvidia/isaaclab-arena-envs \
-    --env.type=isaaclab_arena
-```
-
-## Creating New Environments
-
-First create a new IsaacLab Arena environment by following the [IsaacLab Arena Documentation](https://isaac-sim.github.io/IsaacLab-Arena/release/0.1.1/index.html).
-
-Clone our EnvHub repo:
-
-```bash
-git clone https://huggingface.co/nvidia/isaaclab-arena-envs
-```
-
-Modify the `example_envs.yaml` file based on your new environment.
-[Upload](./envhub#step-3-upload-to-the-hub) your modified repo to HuggingFace EnvHub.
-
-<Tip>
-  Your IsaacLab Arena environment code must be locally available during
-  evaluation. Users can clone your environment repository separately, or you can
-  bundle the environment code and assets directly in your EnvHub repo.
-</Tip>
-
-Then, when evaluating, use your new environment:
-
-```bash
-lerobot-eval \
-    --env.hub_path=<your-env-hub-path>/isaaclab-arena-envs \
-    --env.environment=<your new environment> \
-    ...other flags...
-```
-
-We look forward to your contributions!
-
-## Troubleshooting
-
-### CUDA out of memory
-
-Reduce `batch_size` or use a GPU with more VRAM:
-
-```bash
---eval.batch_size=1
-```
-
-### EULA not accepted
-
-Set environment variables before running:
-
-```bash
-export ACCEPT_EULA=Y
-export PRIVACY_CONSENT=Y
-```
-
-### Video recording not working
-
-Enable cameras when running headless:
-
-```bash
---env.video=true --env.enable_cameras=true --env.headless=true
-```
-
-### Policy output dimension mismatch
-
-Ensure `action_dim` matches your policy:
-
-```bash
---env.action_dim=36
-```
-
-### libGLU.so.1 Errors during Isaac Sim initialization
-
-Ensure you have the following dependencies installed, this is likely to happen on headless machines.
-
-```bash
-sudo apt update && sudo apt install -y libglu1-mesa libxt6
-```
-
-## See Also
-
-- [EnvHub Documentation](./envhub.mdx) - General EnvHub usage
-- [IsaacLab Arena GitHub](https://github.com/isaac-sim/IsaacLab-Arena)
-- [IsaacLab Documentation](https://isaac-sim.github.io/IsaacLab/)
-
-## Lightwheel LW-BenchHub
-
-[Lightwheel](https://www.lightwheel.ai) is bringing `Lightwheel-Libero-Tasks` and `Lightwheel-RoboCasa-Tasks` with 268 tasks to the LeRobot ecosystem.
-LW-BenchHub collects and generates large-scale datasets via teleoperation that comply with the LeRobot specification, enabling out-of-the-box training and evaluation workflows.
-With the unified interface provided by EnvHub, developers can quickly build end-to-end experimental pipelines.
-
-### Install
-
-Assuming you followed the [Installation](#installation) steps, you can install LW-BenchHub with:
-
-```bash
-conda install pinocchio -c conda-forge -y
-pip install numpy==1.26.0 # revert numpy to version 1.26
-
-sudo apt-get install git-lfs && git lfs install
-
-git clone https://github.com/LightwheelAI/lw_benchhub
-git lfs pull # Ensure LFS files (e.g., .usd assets) are downloaded
-
-cd lw_benchhub
-pip install -e .
-```
-
-For more detailed instructions, please refer to the [LW-BenchHub Documentation](https://docs.lightwheel.net/lw_benchhub/usage/Installation).
-
-### Lightwheel Tasks Dataset
-
-LW-BenchHub datasets are available on HuggingFace Hub:
-
-| Dataset                                                                                                       | Description             | Tasks | Frames |
-| :------------------------------------------------------------------------------------------------------------ | :---------------------- | :---- | :----- |
-| [Lightwheel-Tasks-X7S](https://huggingface.co/datasets/LightwheelAI/Lightwheel-Tasks-X7S)                     | X7S LIBERO and RoboCasa | 117   | ~10.3M |
-| [Lightwheel-Tasks-Double-Piper](https://huggingface.co/datasets/LightwheelAI/Lightwheel-Tasks-Double-Piper)   | Double-Piper LIBERO     | 130   | ~6.0M  |
-| [Lightwheel-Tasks-G1-Controller](https://huggingface.co/datasets/LightwheelAI/Lightwheel-Tasks-G1-Controller) | G1-Controller LIBERO    | 62    | ~2.7M  |
-| [Lightwheel-Tasks-G1-WBC](https://huggingface.co/datasets/LightwheelAI/Lightwheel-Tasks-G1-WBC)               | G1-WBC RoboCasa         | 32    | ~1.5M  |
-
-For training policies, refer to the [Training Policies](#training-policies) section.
-
-### Evaluating Policies
-
-#### Pre-trained Policies
-
-The following trained policies are available:
-
-| Policy                   | Architecture | Task                           | Layout     | Robot           | Link                                                                                  |
-| :----------------------- | :----------- | :----------------------------- | :--------- | :-------------- | :------------------------------------------------------------------------------------ |
-| smolvla-double-piper-pnp | SmolVLA      | L90K1PutTheBlackBowlOnThePlate | libero-1-1 | DoublePiper-Abs | [HuggingFace](https://huggingface.co/LightwheelAI/smolvla-double-piper-pnp/tree/main) |
-
-#### Evaluate SmolVLA
-
-```bash
-lerobot-eval \
-  --policy.path=LightwheelAI/smolvla-double-piper-pnp \
-  --env.type=isaaclab_arena \
-  --rename_map='{"observation.images.left_hand_camera_rgb": "observation.images.left_hand", "observation.images.right_hand_camera_rgb": "observation.images.right_hand", "observation.images.first_person_camera_rgb": "observation.images.first_person"}' \
-  --env.hub_path=LightwheelAI/lw_benchhub_env \
-  --env.kwargs='{"config_path": "configs/envhub/example.yml"}' \
-  --trust_remote_code=true \
-  --env.state_keys=joint_pos \
-  --env.action_dim=12 \
-  --env.camera_keys=left_hand_camera_rgb,right_hand_camera_rgb,first_person_camera_rgb \
-  --policy.device=cuda \
-  --eval.batch_size=10 \
-  --eval.n_episodes=100
-```
-
-### Environment Configuration
-
-Evaluation can be quickly launched by modifying the `robot`, `task`, and `layout` settings in the configuration file.
-
-#### Full Configuration Options
-
-```yml
-# =========================
-# Basic Settings
-# =========================
-disable_fabric: false
-device: cuda:0
-sensitivity: 1.0
-step_hz: 50
-enable_cameras: true
-execute_mode: eval
-episode_length_s: 20.0 # Episode length in seconds, increase if episodes timeout during eval
-
-# =========================
-# Robot Settings
-# =========================
-robot: DoublePiper-Abs # Robot type, DoublePiper-Abs, X7S-Abs, G1-Controller or G1-Controller-DecoupledWBC
-robot_scale: 1.0
-
-# =========================
-# Task & Scene Settings
-# =========================
-task: L90K1PutTheBlackBowlOnThePlate # Task name
-scene_backend: robocasa
-task_backend: robocasa
-debug_assets: null
-layout: libero-1-1 # Layout and style ID
-sources:
-  - objaverse
-  - lightwheel
-  - aigen_objs
-object_projects: []
-usd_simplify: false
-seed: 42
-
-# =========================
-# Object Placement Retry Settings
-# =========================
-max_scene_retry: 4
-max_object_placement_retry: 3
-
-resample_objects_placement_on_reset: true
-resample_robot_placement_on_reset: true
-
-# =========================
-# Replay Configuration Settings
-# =========================
-replay_cfgs:
-  add_camera_to_observation: true
-  render_resolution: [640, 480]
-```
-
-### See Also
-
-- [LW-BenchHub GitHub](https://github.com/LightwheelAI/LW-BenchHub)
-- [LW-BenchHub Documentation](https://docs.lightwheel.net/lw_benchhub/)

docs/source/envhub_leisaac.mdx

@@ -137,8 +137,7 @@ from lerobot.teleoperators import (  # noqa: F401
     Teleoperator,
     TeleoperatorConfig,
     make_teleoperator_from_config,
-    so_leader,
-    bi_so_leader,
+    so101_leader,
 )
 from lerobot.utils.robot_utils import precise_sleep
 from lerobot.utils.utils import init_logging
@@ -197,7 +196,7 @@ def teleop_loop(teleop: Teleoperator, env: gym.Env, fps: int):
             obs, info = env.reset()
 
         dt_s = time.perf_counter() - loop_start
-        precise_sleep(max(1 / fps - dt_s, 0.0))
+        precise_sleep(1 / fps - dt_s)
         loop_s = time.perf_counter() - loop_start
         print(f"\ntime: {loop_s * 1e3:.2f}ms ({1 / loop_s:.0f} Hz)")
 
@@ -223,7 +222,7 @@ def teleoperate(cfg: TeleoperateConfig):
 
 def main():
     teleoperate(TeleoperateConfig(
-        teleop=so_leader.SO101LeaderConfig(
+        teleop=so101_leader.SO101LeaderConfig(
             port="/dev/ttyACM0",
             id='leader',
             use_degrees=False,

docs/source/groot.mdx

@@ -12,12 +12,6 @@ Developers and researchers can post-train GR00T N1.5 with their own real or synt
 
 GR00T N1.5 (specifically the GR00T-N1.5-3B model) is built using pre-trained vision and language encoders. It utilizes a flow matching action transformer to model a chunk of actions, conditioned on vision, language, and proprioception.
 
-<img
-  src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/lerobot/lerobot-groot-paper1%20(1).png"
-  alt="An overview of GR00T"
-  width="80%"
-/>
-
 Its strong performance comes from being trained on an expansive and diverse humanoid dataset, which includes:
 
 - Real captured data from robots.
@@ -109,7 +103,7 @@ Once you have trained your model using your parameters you can run inference in
 
 ```bash
 lerobot-record \
-  --robot.type=bi_so_follower \
+  --robot.type=bi_so100_follower \
   --robot.left_arm_port=/dev/ttyACM1 \
   --robot.right_arm_port=/dev/ttyACM0 \
   --robot.id=bimanual_follower \

docs/source/il_robots.mdx

@@ -58,8 +58,8 @@ lerobot-teleoperate \
 
 <!-- prettier-ignore-start -->
 ```python
-from lerobot.teleoperators.so_leader import SO101LeaderConfig, SO101Leader
-from lerobot.robots.so_follower import SO101FollowerConfig, SO101Follower
+from lerobot.teleoperators.so101_leader import SO101LeaderConfig, SO101Leader
+from lerobot.robots.so101_follower import SO101FollowerConfig, SO101Follower
 
 robot_config = SO101FollowerConfig(
     port="/dev/tty.usbmodem58760431541",
@@ -195,14 +195,13 @@ lerobot-record \
 from lerobot.cameras.opencv.configuration_opencv import OpenCVCameraConfig
 from lerobot.datasets.lerobot_dataset import LeRobotDataset
 from lerobot.datasets.utils import hw_to_dataset_features
-from lerobot.robots.so_follower import SO100Follower, SO100FollowerConfig
-from lerobot.teleoperators.so_leader.config_so100_leader import SO100LeaderConfig
-from lerobot.teleoperators.so_leader.so100_leader import SO100Leader
+from lerobot.robots.so100_follower import SO100Follower, SO100FollowerConfig
+from lerobot.teleoperators.so100_leader.config_so100_leader import SO100LeaderConfig
+from lerobot.teleoperators.so100_leader.so100_leader import SO100Leader
 from lerobot.utils.control_utils import init_keyboard_listener
 from lerobot.utils.utils import log_say
 from lerobot.utils.visualization_utils import init_rerun
-from lerobot.scripts.lerobot_record import record_loop
-from lerobot.processor import make_default_processors
+from lerobot.record import record_loop
 
 NUM_EPISODES = 5
 FPS = 30
@@ -210,19 +209,12 @@ EPISODE_TIME_SEC = 60
 RESET_TIME_SEC = 10
 TASK_DESCRIPTION = "My task description"
 
-# Create robot configuration
+# Create the robot and teleoperator configurations
+camera_config = {"front": OpenCVCameraConfig(index_or_path=0, width=640, height=480, fps=FPS)}
 robot_config = SO100FollowerConfig(
-    id="my_awesome_follower_arm",
-    cameras={
-        "front": OpenCVCameraConfig(index_or_path=0, width=640, height=480, fps=FPS) # Optional: fourcc="MJPG" for troubleshooting OpenCV async error.
-    },
-    port="/dev/tty.usbmodem58760434471",
-)
-
-teleop_config = SO100LeaderConfig(
-    id="my_awesome_leader_arm",
-    port="/dev/tty.usbmodem585A0077581",
+    port="/dev/tty.usbmodem58760434471", id="my_awesome_follower_arm", cameras=camera_config
 )
+teleop_config = SO100LeaderConfig(port="/dev/tty.usbmodem585A0077581", id="my_awesome_leader_arm")
 
 # Initialize the robot and teleoperator
 robot = SO100Follower(robot_config)
@@ -251,9 +243,6 @@ init_rerun(session_name="recording")
 robot.connect()
 teleop.connect()
 
-# Create the required processors
-teleop_action_processor, robot_action_processor, robot_observation_processor = make_default_processors()
-
 episode_idx = 0
 while episode_idx < NUM_EPISODES and not events["stop_recording"]:
     log_say(f"Recording episode {episode_idx + 1} of {NUM_EPISODES}")
@@ -262,9 +251,6 @@ while episode_idx < NUM_EPISODES and not events["stop_recording"]:
         robot=robot,
         events=events,
         fps=FPS,
-        teleop_action_processor=teleop_action_processor,
-        robot_action_processor=robot_action_processor,
-        robot_observation_processor=robot_observation_processor,
         teleop=teleop,
         dataset=dataset,
         control_time_s=EPISODE_TIME_SEC,
@@ -279,9 +265,6 @@ while episode_idx < NUM_EPISODES and not events["stop_recording"]:
             robot=robot,
             events=events,
             fps=FPS,
-            teleop_action_processor=teleop_action_processor,
-            robot_action_processor=robot_action_processor,
-            robot_observation_processor=robot_observation_processor,
             teleop=teleop,
             control_time_s=RESET_TIME_SEC,
             single_task=TASK_DESCRIPTION,
@@ -408,8 +391,8 @@ lerobot-replay \
 import time
 
 from lerobot.datasets.lerobot_dataset import LeRobotDataset
-from lerobot.robots.so_follower.config_so100_follower import SO100FollowerConfig
-from lerobot.robots.so_follower.so100_follower import SO100Follower
+from lerobot.robots.so100_follower.config_so100_follower import SO100FollowerConfig
+from lerobot.robots.so100_follower.so100_follower import SO100Follower
 from lerobot.utils.robot_utils import precise_sleep
 from lerobot.utils.utils import log_say
 
@@ -432,7 +415,7 @@ for idx in range(dataset.num_frames):
     }
     robot.send_action(action)
 
-    precise_sleep(max(1.0 / dataset.fps - (time.perf_counter() - t0), 0.0))
+    precise_sleep(1.0 / dataset.fps - (time.perf_counter() - t0))
 
 robot.disconnect()
 ```
@@ -445,7 +428,7 @@ Your robot should replicate movements similar to those you recorded. For example
 
 ## Train a policy
 
-To train a policy to control your robot, use the [`lerobot-train`](https://github.com/huggingface/lerobot/blob/main/src/lerobot/scripts/lerobot_train.py) script. A few arguments are required. Here is an example command:
+To train a policy to control your robot, use the [`lerobot-train`](https://github.com/huggingface/lerobot/blob/main/src/lerobot/scripts/train.py) script. A few arguments are required. Here is an example command:
 
 ```bash
 lerobot-train \
@@ -502,7 +485,7 @@ huggingface-cli upload ${HF_USER}/act_so101_test${CKPT} \
 
 ## Run inference and evaluate your policy
 
-You can use the `record` script from [`lerobot-record`](https://github.com/huggingface/lerobot/blob/main/src/lerobot/scripts/lerobot_record.py) with a policy checkpoint as input, to run inference and evaluate your policy. For instance, run this command or API example to run inference and record 10 evaluation episodes:
+You can use the `record` script from [`lerobot/record.py`](https://github.com/huggingface/lerobot/blob/main/src/lerobot/record.py) with a policy checkpoint as input, to run inference and evaluate your policy. For instance, run this command or API example to run inference and record 10 evaluation episodes:
 
 <hfoptions id="eval">
 <hfoption id="Command">
@@ -531,8 +514,8 @@ from lerobot.datasets.lerobot_dataset import LeRobotDataset
 from lerobot.datasets.utils import hw_to_dataset_features
 from lerobot.policies.act.modeling_act import ACTPolicy
 from lerobot.policies.factory import make_pre_post_processors
-from lerobot.robots.so_follower.config_so100_follower import SO100FollowerConfig
-from lerobot.robots.so_follower.so100_follower import SO100Follower
+from lerobot.robots.so100_follower.config_so100_follower import SO100FollowerConfig
+from lerobot.robots.so100_follower.so100_follower import SO100Follower
 from lerobot.scripts.lerobot_record import record_loop
 from lerobot.utils.control_utils import init_keyboard_listener
 from lerobot.utils.utils import log_say

docs/source/installation.mdx

@@ -90,7 +90,7 @@ If you encounter build errors, you may need to install additional dependencies:
 To install these for linux run:
 
 ```bash
-sudo apt-get install cmake build-essential python3-dev pkg-config libavformat-dev libavcodec-dev libavdevice-dev libavutil-dev libswscale-dev libswresample-dev libavfilter-dev
+sudo apt-get install cmake build-essential python-dev pkg-config libavformat-dev libavcodec-dev libavdevice-dev libavutil-dev libswscale-dev libswresample-dev libavfilter-dev pkg-config
 ```
 
 For other systems, see: [Compiling PyAV](https://pyav.org/docs/develop/overview/installation.html#bring-your-own-ffmpeg)

docs/source/integrate_hardware.mdx

@@ -18,7 +18,7 @@ If you're using Feetech or Dynamixel motors, LeRobot provides built-in bus inter
 - [`DynamixelMotorsBus`](https://github.com/huggingface/lerobot/blob/main/src/lerobot/motors/dynamixel/dynamixel.py) – for controlling Dynamixel servos
 
 Please refer to the [`MotorsBus`](https://github.com/huggingface/lerobot/blob/main/src/lerobot/motors/motors_bus.py) abstract class to learn about its API.
-For a good example of how it can be used, you can have a look at our own [SO101 follower implementation](https://github.com/huggingface/lerobot/blob/main/src/lerobot/robots/so_follower/so101_follower/so101_follower.py)
+For a good example of how it can be used, you can have a look at our own [SO101 follower implementation](https://github.com/huggingface/lerobot/blob/main/src/lerobot/robots/so101_follower/so101_follower.py)
 
 Use these if compatible. Otherwise, you'll need to find or write a Python interface (not covered in this tutorial):
 

docs/source/lekiwi.mdx

@@ -204,7 +204,7 @@ lerobot-calibrate \
 
 <!-- prettier-ignore-start -->
 ```python
-from lerobot.teleoperators.so_leader import SO100LeaderConfig, SO100Leader
+from lerobot.teleoperators.so100_leader import SO100LeaderConfig, SO100Leader
 
 config = SO100LeaderConfig(
     port="/dev/tty.usbmodem58760431551",

docs/source/libero.mdx

@@ -62,11 +62,6 @@ lerobot-eval \
 
 - Pass a comma-separated list to `--env.task` for multi-suite evaluation.
 
-### Control Mode
-
-LIBERO now supports two control modes: relative and absolute. This matters because different VLA checkpoints are trained with different mode of action to output hence control parameterizations.
-You can switch them with: `env.control_mode = "relative"` and `env.control_mode = "absolute"`
-
 ### Policy inputs and outputs
 
 When using LIBERO through LeRobot, policies interact with the environment via **observations** and **actions**:

docs/source/peft_training.mdx

@@ -1,62 +0,0 @@
-# Parameter efficient fine-tuning with 🤗 PEFT
-
-[🤗 PEFT](https://github.com/huggingface/peft) (Parameter-Efficient Fine-Tuning) is a library for efficiently adapting
-large pretrained models such as pre-trained policies (e.g., SmolVLA, π₀, ...) to new tasks without training all
-of the model's parameters while yielding comparable performance.
-
-Install the `lerobot[peft]` optional package to enable PEFT support.
-
-To read about all the possible methods of adaption, please refer to the [🤗 PEFT docs](https://huggingface.co/docs/peft/index).
-
-## Training SmolVLA
-
-In this section we'll show you how to train a pre-trained SmolVLA policy with PEFT on the libero dataset.
-For brevity we're only training on the `libero_spatial` subset. We will use `lerobot/smolvla_base` as the model
-to parameter efficiently fine-tune:
-
-```
-lerobot-train \
- --policy.path=lerobot/smolvla_base \
- --policy.repo_id=your_hub_name/my_libero_smolvla \
- --dataset.repo_id=HuggingFaceVLA/libero \
- --policy.output_features=null \
- --policy.input_features=null \
- --policy.optimizer_lr=1e-3 \
- --policy.scheduler_decay_lr=1e-4 \
- --env.type=libero \
- --env.task=libero_spatial \
- --steps=100000 \
- --batch_size=32 \
- --peft.method_type=LORA \
- --peft.r=64
-```
-
-Note the `--peft.method_type` parameter that let's you select which PEFT method to use. Here we use
-[LoRA](https://huggingface.co/docs/peft/main/en/package_reference/lora) (Low-Rank Adapter) which is probably the most
-popular fine-tuning method to date. Low-rank adaption means that we only fine-tune a matrix with comparably low rank
-instead of the full weight matrix. This rank can be specified using the `--peft.r` parameter. The higher the rank
-the closer you get to full fine-tuning
-
-There are more complex methods that have more parameters. These are not yet supported, feel free to raise an issue
-if you want to see a specific PEFT method supported.
-
-By default, PEFT will target the `q_proj` and `v_proj` layers of the LM expert in SmolVLA. It will also target the
-state and action projection matrices as they are most likely task-dependent. If you need to target different layers
-you can use `--peft.target_modules` to specify which layers to target. You can refer to the respective PEFT method's
-documentation to see what inputs are supported, (e.g., [LoRA's target_modules documentation](https://huggingface.co/docs/peft/main/en/package_reference/lora#peft.LoraConfig.target_modules)).
-Usually a list of suffixes or a regex are supported. For example, to target the MLPs of the `lm_expert` instead of
-the `q` and `v` projections, use:
-
-```
---peft.target_modules='(model\.vlm_with_expert\.lm_expert\..*\.(down|gate|up)_proj|.*\.(state_proj|action_in_proj|action_out_proj|action_time_mlp_in|action_time_mlp_out))'
-```
-
-In case you need to fully fine-tune a layer instead of just adapting it, you can supply a list of layer suffixes
-to the `--peft.full_training_modules` parameter:
-
-```
---peft.full_training_modules=["state_proj"]
-```
-
-The learning rate and the scheduled target learning rate can usually be scaled by a factor of 10 compared to the
-learning rate used for full fine-tuning (e.g., 1e-4 normal, so 1e-3 using LoRA).

docs/source/phone_teleop.mdx

@@ -44,7 +44,7 @@ Modify the examples to use `PhoneOS.IOS` or `PhoneOS.ANDROID` in `PhoneConfig`.
 
 Teleoperation example:
 
-```python
+```36:43:examples/phone_so100_teleop.py
 from lerobot.teleoperators.phone.config_phone import PhoneConfig, PhoneOS
 
 teleop_config = PhoneConfig(phone_os=PhoneOS.IOS)  # or PhoneOS.ANDROID
@@ -103,7 +103,7 @@ Additionally you can customize mapping or safety limits by editing the processor
 
 - Kinematics are used in multiple steps. We use [Placo](https://github.com/Rhoban/placo) which is a wrapper around Pinocchio for handling our kinematics. We construct the kinematics object by passing the robot's URDF and target frame. We set `target_frame_name` to the gripper frame.
 
-  ```python
+  ```examples/phone_to_so100/teleoperate.py
   kinematics_solver = RobotKinematics(
     urdf_path="./SO101/so101_new_calib.urdf",
     target_frame_name="gripper_frame_link",
@@ -114,7 +114,7 @@ Additionally you can customize mapping or safety limits by editing the processor
 
 - The `MapPhoneActionToRobotAction` step converts the calibrated phone pose and inputs into target deltas and gripper commands, below is shown what the step outputs.
 
-  ```python
+  ```src/lerobot/teleoperators/phone/phone_processor.py
   action["enabled"] = enabled
         action["target_x"] = -pos[1] if enabled else 0.0
         action["target_y"] = pos[0] if enabled else 0.0
@@ -127,7 +127,7 @@ Additionally you can customize mapping or safety limits by editing the processor
 
 - The `EEReferenceAndDelta` step converts target deltas to an absolute desired EE pose, storing a reference on enable, the `end_effector_step_sizes` are the step sizes for the EE pose and can be modified to change the motion speed.
 
-  ```python
+  ```examples/phone_to_so100/teleoperate.py
   EEReferenceAndDelta(
       kinematics=kinematics_solver,
       end_effector_step_sizes={"x": 0.5, "y": 0.5, "z": 0.5},
@@ -138,7 +138,7 @@ Additionally you can customize mapping or safety limits by editing the processor
 
 - The `EEBoundsAndSafety` step clamps EE motion to a workspace and checks for large ee step jumps to ensure safety. The `end_effector_bounds` are the bounds for the EE pose and can be modified to change the workspace. The `max_ee_step_m` are the step limits for the EE pose and can be modified to change the safety limits.
 
-  ```python
+  ```examples/phone_to_so100/teleoperate.py
   EEBoundsAndSafety(
       end_effector_bounds={"min": [-1.0, -1.0, -1.0], "max": [1.0, 1.0, 1.0]},
       max_ee_step_m=0.10,
@@ -147,7 +147,7 @@ Additionally you can customize mapping or safety limits by editing the processor
 
 - The `GripperVelocityToJoint` step turns a velocity‑like gripper input into absolute gripper position using the current measured state. The `speed_factor` is the factor by which the velocity is multiplied.
 
-  ```python
+  ```examples/phone_to_so100/teleoperate.py
   GripperVelocityToJoint(speed_factor=20.0)
   ```
 
@@ -157,7 +157,7 @@ We use different IK initial guesses in the kinematic steps. As initial guess eit
 
 - Closed loop (used in record/eval): sets `initial_guess_current_joints=True` so IK starts from the measured joints each frame.
 
-  ```python
+  ```examples/phone_to_so100/record.py
   InverseKinematicsEEToJoints(
       kinematics=kinematics_solver,
       motor_names=list(robot.bus.motors.keys()),
@@ -167,7 +167,7 @@ We use different IK initial guesses in the kinematic steps. As initial guess eit
 
 - Open loop (used in replay): sets `initial_guess_current_joints=False` so IK continues from the previous IK solution rather than the measured state. This preserves action stability when we replay without feedback.
 
-  ```python
+  ```examples/phone_to_so100/replay.py
   InverseKinematicsEEToJoints(
       kinematics=kinematics_solver,
       motor_names=list(robot.bus.motors.keys()),

docs/source/pi0.mdx

@@ -6,12 +6,6 @@
 
 π₀ represents a breakthrough in robotics as the first general-purpose robot foundation model developed by [Physical Intelligence](https://www.physicalintelligence.company/blog/pi0). Unlike traditional robot programs that are narrow specialists programmed for repetitive motions, π₀ is designed to be a generalist policy that can understand visual inputs, interpret natural language instructions, and control a variety of different robots across diverse tasks.
 
-<img
-  src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/lerobot/lerobot-pi0%20(1).png"
-  alt="An overview of Pi0"
-  width="85%"
-/>
-
 ### The Vision for Physical Intelligence
 
 As described by Physical Intelligence, while AI has achieved remarkable success in digital domains, from chess-playing to drug discovery, human intelligence still dramatically outpaces AI in the physical world. To paraphrase Moravec's paradox, winning a game of chess represents an "easy" problem for AI, but folding a shirt or cleaning up a table requires solving some of the most difficult engineering problems ever conceived. π₀ represents a first step toward developing artificial physical intelligence that enables users to simply ask robots to perform any task they want, just like they can with large language models.
@@ -70,8 +64,6 @@ python src/lerobot/scripts/lerobot_train.py \
     --policy.compile_model=true \
     --policy.gradient_checkpointing=true \
     --policy.dtype=bfloat16 \
-    --policy.freeze_vision_encoder=false \
-    --policy.train_expert_only=false \
     --steps=3000 \
     --policy.device=cuda \
     --batch_size=32
@@ -87,15 +79,6 @@ python src/lerobot/scripts/lerobot_train.py \
   - [lerobot/pi0_base](https://huggingface.co/lerobot/pi0_base)
   - [lerobot/pi0_libero](https://huggingface.co/lerobot/pi0_libero) (specifically trained on the Libero dataset)
 
-### Training Parameters Explained
-
-| Parameter               | Default | Description                                 |
-| ----------------------- | ------- | ------------------------------------------- |
-| `freeze_vision_encoder` | `false` | Do not freeze the vision encoder            |
-| `train_expert_only`     | `false` | Do not freeze the VLM, train all parameters |
-
-**💡 Tip**: Setting `train_expert_only=true` freezes the VLM and trains only the action expert and projections, allowing finetuning with reduced memory usage.
-
 ## License
 
 This model follows the **Apache 2.0 License**, consistent with the original [OpenPI repository](https://github.com/Physical-Intelligence/openpi).

docs/source/pi05.mdx

@@ -67,8 +67,6 @@ python src/lerobot/scripts/lerobot_train.py\
     --policy.gradient_checkpointing=true \
     --wandb.enable=true \
     --policy.dtype=bfloat16 \
-    --policy.freeze_vision_encoder=false \
-    --policy.train_expert_only=false \
     --steps=3000 \
     --policy.device=cuda \
     --batch_size=32
@@ -84,15 +82,6 @@ python src/lerobot/scripts/lerobot_train.py\
   - [lerobot/pi05_base](https://huggingface.co/lerobot/pi05_base)
   - [lerobot/pi05_libero](https://huggingface.co/lerobot/pi05_libero) (specifically trained on the Libero dataset)
 
-### Training Parameters Explained
-
-| Parameter               | Default | Description                                 |
-| ----------------------- | ------- | ------------------------------------------- |
-| `freeze_vision_encoder` | `false` | Do not freeze the vision encoder            |
-| `train_expert_only`     | `false` | Do not freeze the VLM, train all parameters |
-
-**💡 Tip**: Setting `train_expert_only=true` freezes the VLM and trains only the action expert and projections, allowing finetuning with reduced memory usage.
-
 If your dataset is not converted with `quantiles`, you can convert it with the following command:
 
 ```bash

docs/source/pi0fast.mdx

@@ -1,246 +0,0 @@
-# π₀-FAST (Pi0-FAST)
-
-π₀-FAST is a **Vision-Language-Action model for general robot control** that uses autoregressive next-token prediction to model continuous robot actions.
-
-## Model Overview
-
-π₀-FAST combines the power of Vision-Language Models with a novel action tokenization approach called **FAST (Frequency-space Action Sequence Tokenization)**. This enables training autoregressive VLAs on highly dexterous tasks that are impossible with standard binning-based discretization, while training **up to 5x faster** than diffusion-based approaches like π₀.
-
-<img
-  src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/lerobot/lerobot-pifast.png"
-  alt="An overview of Pi0-FAST"
-  width="85%"
-/>
-
-### Why FAST?
-
-Standard approaches for robot action tokenization use simple per-dimension, per-timestep binning schemes. While passable for simple behaviors, this rapidly breaks down for complex and dexterous skills that require precision and high-frequency control.
-
-FAST solves this by compressing action sequences using signal processing techniques, resulting in a dense sequence of action tokens that can be predicted autoregressively—just like language tokens.
-
-### How FAST Tokenization Works
-
-The FAST tokenizer compresses action sequences through the following steps:
-
-1. **Normalize**: Take a continuous action chunk of shape `(H, D)` where `H` is the horizon and `D` is the action dimension. Normalize using one of the supported normalization methods (Quantiles recommended to handle outliers).
-
-2. **Discrete Cosine Transform (DCT)**: Apply DCT (via scipy) to each action dimension separately. DCT is a compression algorithm commonly used in image and audio codecs (JPEG, MP3).
-
-3. **Quantization**: Round and remove insignificant coefficients for each action dimension, producing a sparse frequency matrix.
-
-4. **Flatten**: Flatten the matrix into a 1D vector, with low-frequency components first.
-
-5. **Byte Pair Encoding (BPE)**: Train a BPE tokenizer to compress the DCT coefficients into dense action tokens, typically achieving **10x compression** over prior tokenization approaches.
-
-This approach can transform **any existing VLM** into a VLA by training it to predict these FAST tokens.
-
-## Installation Requirements
-
-1. Install LeRobot by following our [Installation Guide](./installation).
-2. Install π₀-FAST dependencies by running:
-
-   ```bash
-   pip install -e ".[pi]"
-   ```
-
-   > [!NOTE]
-   > For lerobot 0.4.0, if you want to install the pi tag, you will have to do: `pip install "lerobot[pi]@git+https://github.com/huggingface/lerobot.git"`.
-   >
-   > This will be solved in the next patch release
-
-## Training a Custom FAST Tokenizer
-
-You have two options for the FAST tokenizer:
-
-1. **Use the pre-trained tokenizer**: The `physical-intelligence/fast` tokenizer was trained on 1M+ real robot action sequences and works as a general-purpose tokenizer.
-
-2. **Train your own tokenizer**: For maximum performance on your specific dataset, you can finetune the tokenizer on your own data.
-
-### Training Your Own Tokenizer
-
-```bash
-lerobot-train-tokenizer \
-    --repo_id "user/my-lerobot-dataset" \
-    --action_horizon 10 \
-    --encoded_dims "0:6" \
-    --vocab_size 1024 \
-    --scale 10.0 \
-    --normalization_mode QUANTILES \
-    --output_dir "./my_fast_tokenizer" \
-    --push_to_hub \
-    --hub_repo_id "username/my-action-tokenizer"
-```
-
-### Key Tokenizer Parameters
-
-| Parameter              | Description                                                                       | Default      |
-| ---------------------- | --------------------------------------------------------------------------------- | ------------ |
-| `--repo_id`            | LeRobot dataset repository ID                                                     | Required     |
-| `--action_horizon`     | Number of future actions in each chunk                                            | `10`         |
-| `--encoded_dims`       | Comma-separated dimension ranges to encode (e.g., `"0:6,7:23"`)                   | `"0:6,7:23"` |
-| `--vocab_size`         | BPE vocabulary size                                                               | `1024`       |
-| `--scale`              | DCT scaling factor for quantization                                               | `10.0`       |
-| `--normalization_mode` | Normalization mode (`MEAN_STD`, `MIN_MAX`, `QUANTILES`, `QUANTILE10`, `IDENTITY`) | `QUANTILES`  |
-| `--sample_fraction`    | Fraction of chunks to sample per episode                                          | `0.1`        |
-
-## Usage
-
-To use π₀-FAST in LeRobot, specify the policy type as:
-
-```python
-policy.type=pi0_fast
-```
-
-## Training
-
-For training π₀-FAST, you can use the LeRobot training script:
-
-```bash
-lerobot-train \
-    --dataset.repo_id=your_dataset \
-    --policy.type=pi0_fast \
-    --output_dir=./outputs/pi0fast_training \
-    --job_name=pi0fast_training \
-    --policy.pretrained_path=lerobot/pi0_fast_base \
-    --policy.dtype=bfloat16 \
-    --policy.gradient_checkpointing=true \
-    --policy.chunk_size=10 \
-    --policy.n_action_steps=10 \
-    --policy.max_action_tokens=256 \
-    --steps=100000 \
-    --batch_size=4 \
-    --policy.device=cuda
-```
-
-### Key Training Parameters
-
-| Parameter                              | Description                                        | Default                      |
-| -------------------------------------- | -------------------------------------------------- | ---------------------------- |
-| `--policy.gradient_checkpointing=true` | Reduces memory usage significantly during training | `false`                      |
-| `--policy.dtype=bfloat16`              | Use mixed precision training for efficiency        | `float32`                    |
-| `--policy.chunk_size`                  | Number of action steps to predict (action horizon) | `50`                         |
-| `--policy.n_action_steps`              | Number of action steps to execute                  | `50`                         |
-| `--policy.max_action_tokens`           | Maximum number of FAST tokens per action chunk     | `256`                        |
-| `--policy.action_tokenizer_name`       | FAST tokenizer to use                              | `physical-intelligence/fast` |
-| `--policy.compile_model=true`          | Enable torch.compile for faster training           | `false`                      |
-
-## Inference
-
-### KV-Caching for Fast Inference
-
-π₀-FAST supports **KV-caching**, a widely used optimization in LLM inference. This caches the key-value pairs from the attention mechanism, avoiding redundant computation during autoregressive decoding.
-
-```python
-# KV-caching is enabled by default
-policy.use_kv_cache=true
-```
-
-### Inference Example
-
-```python
-from lerobot.policies.pi0_fast import PI0FastPolicy, PI0FastConfig
-
-# Load the policy
-policy = PI0FastPolicy.from_pretrained("your-model-path")
-
-# During inference
-actions = policy.predict_action_chunk(batch)
-```
-
-## Model Architecture
-
-π₀-FAST uses a PaliGemma-based architecture:
-
-- **Vision Encoder**: SigLIP vision tower for image understanding
-- **Language Model**: Gemma 2B for processing language instructions and predicting action tokens
-
-The model takes images, text instructions, and robot state as input, and outputs discrete FAST tokens that are decoded back to continuous actions.
-
-## Configuration Options
-
-| Parameter            | Description                                     | Default    |
-| -------------------- | ----------------------------------------------- | ---------- |
-| `paligemma_variant`  | VLM backbone variant (`gemma_300m`, `gemma_2b`) | `gemma_2b` |
-| `max_state_dim`      | Maximum state vector dimension (padded)         | `32`       |
-| `max_action_dim`     | Maximum action vector dimension (padded)        | `32`       |
-| `temperature`        | Sampling temperature (0.0 for greedy)           | `0.0`      |
-| `max_decoding_steps` | Maximum decoding steps                          | `256`      |
-| `use_kv_cache`       | Enable KV caching for faster inference          | `true`     |
-
-## Comparison with π₀
-
-| Feature               | π₀                        | π₀-FAST                      |
-| --------------------- | ------------------------- | ---------------------------- |
-| Action Representation | Flow Matching (Diffusion) | Autoregressive Tokens (FAST) |
-| Training Speed        | 1x                        | **5x faster**                |
-| Dexterity             | High                      | High                         |
-| Inference Method      | Iterative Denoising       | Autoregressive Decoding      |
-| KV-Caching            | N/A                       | Supported                    |
-
-## Reproducing π₀Fast results
-
-We reproduce the results of π₀Fast on the LIBERO benchmark using the LeRobot implementation. We take the LeRobot PiFast base model [lerobot/pi0fast-base](https://huggingface.co/lerobot/pi0fast-base) and finetune for an additional 40kk steps in bfloat16, with batch size of 256 on 8 H100 GPUs using the [HuggingFace LIBERO dataset](https://huggingface.co/datasets/HuggingFaceVLA/libero).
-
-The finetuned model can be found here:
-
-- **π₀Fast LIBERO**: [lerobot/pi0fast-libero](https://huggingface.co/lerobot/pi0fast-libero)
-
-With the following training command:
-
-```bash
-lerobot-train \
-  --dataset.repo_id=lerobot/libero \
-  --output_dir=outputs/libero_pi0fast \
-  --job_name=libero_pi0fast \
-  --policy.path=lerobot/pi0fast_base \
-  --policy.dtype=bfloat16 \
-  --steps=100000 \
-  --save_freq=20000 \
-  --batch_size=4 \
-  --policy.device=cuda \
-  --policy.scheduler_warmup_steps=4000 \
-  --policy.scheduler_decay_steps=100000 \
-  --policy.scheduler_decay_lr=1e-5 \
-  --policy.gradient_checkpointing=true \
-  --policy.chunk_size=10 \
-  --policy.n_action_steps=10 \
-  --policy.max_action_tokens=256 \
-  --policy.empty_cameras=1 \
-```
-
-We then evaluate the finetuned model using the LeRobot LIBERO implementation, by running the following command:
-
-```bash
-tasks="libero_object,libero_spatial,libero_goal,libero_10"
-lerobot-eval \
-  --policy.path=lerobot/pi0fast-libero \
-  --policy.max_action_tokens=256 \
-  --env.type=libero \
-  --policy.gradient_checkpointing=false \
-  --env.task=${tasks} \
-  --eval.batch_size=1 \
-  --eval.n_episodes=1 \
-  --rename_map='{"observation.images.image":"observation.images.base_0_rgb","observation.images.image2":"observation.images.left_wrist_0_rgb"}'
-```
-
-**Note:** We set `n_action_steps=10`, similar to the original OpenPI implementation.
-
-### Results
-
-We obtain the following results on the LIBERO benchmark:
-
-| Model       | LIBERO Spatial | LIBERO Object | LIBERO Goal | LIBERO 10 | Average  |
-| ----------- | -------------- | ------------- | ----------- | --------- | -------- |
-| **π₀-fast** | 70.0           | 100.0         | 100.0       | 60.0      | **82.5** |
-
-The full evaluation output folder, including videos, is available [here](https://drive.google.com/drive/folders/1HXpwPTRm4hx6g1sF2P7OOqGG0TwPU7LQ?usp=sharing)
-
-## License
-
-This model follows the **Apache 2.0 License**, consistent with the original [OpenPI repository](https://github.com/Physical-Intelligence/openpi).
-
-## References
-
-- [FAST: Efficient Robot Action Tokenization](https://www.physicalintelligence.company/research/fast) - Physical Intelligence Blog
-- [OpenPI Repository](https://github.com/Physical-Intelligence/openpi) - Original implementation
-- [FAST Tokenizer on Hugging Face](https://huggingface.co/physical-intelligence/fast) - Pre-trained tokenizer

docs/source/policy_walloss_README.md

@@ -1,45 +0,0 @@
-# WALL-OSS
-
-This repository contains the Hugging Face port of [**WALL-OSS**](https://x2robot.com/en/research/68bc2cde8497d7f238dde690), a Vision-Language-Action model for cross-embodiment robotic control based on Qwen2.5-VL with flow matching/FAST action prediction.
-
----
-
-## Model Overview
-
-| Feature            | Description                                           |
-| ------------------ | ----------------------------------------------------- |
-| Base Model         | Qwen2.5-VL (Vision-Language Model)                    |
-| Action Prediction  | Flow Matching (diffusion) or FAST (discrete tokens)   |
-| Architecture       | Mixture of Experts (MoE) with action-specific routing |
-| Multi-Modal Inputs | Vision (images/videos), Language, Proprioception      |
-
----
-
-## Additional Resources
-
-Paper: https://arxiv.org/pdf/2509.11766
-
-Official Repository: https://github.com/X-Square-Robot/wall-x
-
-Hugging Face: https://huggingface.co/x-square-robot
-
----
-
-## Citation
-
-If you use this work, please cite:
-
-```bibtex
-@article{zhai2025igniting,
-    title   = {Igniting VLMs Toward the Embodied Space},
-    author  = {Zhai, Andy and Liu, Brae and Fang, Bruno and Cai, Chalse and Ma, Ellie and Yin, Ethan and Wang, Hao and Zhou, Hugo and Wang, James and Shi, Lights and Liang, Lucy and Wang, Make and Wang, Qian and Gan, Roy and Yu, Ryan and Li, Shalfun and Liu, Starrick and Chen, Sylas and Chen, Vincent and Xu, Zach},
-    journal = {arXiv preprint arXiv:2509.11766},
-    year    = {2025}
-}
-```
-
----
-
-## License
-
-This model follows the **Apache 2.0 License**, consistent with the original [WallX repository](https://github.com/X-Square-Robot/wall-x).

docs/source/processors_robots_teleop.mdx

@@ -30,7 +30,7 @@ Each of these pipelines handle different conversions between different action an
 
 Below is an example of the three pipelines that we use in the phone to SO-100 follower examples:
 
-```python
+```69:90:examples/phone_so100_record.py
 phone_to_robot_ee_pose_processor = RobotProcessorPipeline[RobotAction, RobotAction]( # teleop -> dataset action
     steps=[
         MapPhoneActionToRobotAction(platform=teleop_config.phone_os),
@@ -84,7 +84,7 @@ Dataset features are determined by the keys saved in the dataset. Each step can
 
 Below is and example of how we declare features with the `transform_features` method in the phone to SO-100 follower examples:
 
-```python
+```src/lerobot/robots/so100_follower/robot_kinematic_processor.py
     def transform_features(
         self, features: dict[PipelineFeatureType, dict[str, PolicyFeature]]
     ) -> dict[PipelineFeatureType, dict[str, PolicyFeature]]:
@@ -103,7 +103,7 @@ Here we declare what PolicyFeatures we modify in this step, so we know what feat
 
 Below is an example of how we aggregate and merge features in the phone to SO-100 record example:
 
-```python
+```121:145:examples/phone_so100_record.py
 features=combine_feature_dicts(
         # Run the feature contract of the pipelines
         # This tells you how the features would look like after the pipeline steps

docs/source/reachy2.mdx

@@ -38,7 +38,6 @@ docker run --rm -it \
   start_rviz:=true start_sdk_server:=true mujoco:=true
 ```
 
-> [!NOTE]
 > If MuJoCo runs slowly (low simulation frequency), append `-e LD_LIBRARY_PATH="/opt/host-libs:$LD_LIBRARY_PATH" \` to the previous command to improve performance:
 >
 > ```
@@ -142,7 +141,7 @@ If you choose this option but still want to use the VR teleoperation application
 First add reachy2 and reachy2_teleoperator to the imports of the record script. Then you can use the following command:
 
 ```bash
-lerobot-record \
+python -m lerobot.record \
     --robot.type=reachy2 \
     --robot.ip_address=192.168.0.200 \
     --robot.id=r2-0000 \
@@ -151,7 +150,6 @@ lerobot-record \
     --teleop.type=reachy2_teleoperator \
     --teleop.ip_address=192.168.0.200 \
     --teleop.with_mobile_base=false \
-    --robot.with_torso_camera=true \
     --dataset.repo_id=pollen_robotics/record_test \
     --dataset.single_task="Reachy 2 recording test" \
     --dataset.num_episodes=1 \
@@ -167,7 +165,7 @@ lerobot-record \
 **Extended setup overview (all options included):**
 
 ```bash
-lerobot-record \
+python -m lerobot.record \
     --robot.type=reachy2 \
     --robot.ip_address=192.168.0.200 \
     --robot.use_external_commands=true \
@@ -179,8 +177,6 @@ lerobot-record \
     --robot.with_left_teleop_camera=true \
     --robot.with_right_teleop_camera=true \
     --robot.with_torso_camera=false \
-    --robot.camera_width=640 \
-    --robot.camera_height=480 \
     --robot.disable_torque_on_disconnect=false \
     --robot.max_relative_target=5.0 \
     --teleop.type=reachy2_teleoperator \
@@ -216,10 +212,9 @@ Must be set to true if a compliant Reachy 2 is used to control another one.
 From our initial tests, recording **all** joints when only some are moving can reduce model quality with certain policies.
 To avoid this, you can exclude specific parts from recording and replay using:
 
-```bash
+````
 --robot.with_<part>=false
-```
-
+```,
 with `<part>` being one of : `mobile_base`, `l_arm`, `r_arm", `neck`, `antennas`.
 It determine whether the corresponding part is recorded in the observations. True if not set.
 
@@ -227,60 +222,49 @@ By default, **all parts are recorded**.
 
 The same per-part mechanism is available in `reachy2_teleoperator` as well.
 
-```bash
---teleop.with\_<part>
-```
+````
 
+--teleop.with\_<part>
+
+```
 with `<part>` being one of : `mobile_base`, `l_arm`, `r_arm", `neck`, `antennas`.
 Determine whether the corresponding part is recorded in the actions. True if not set.
 
 > **Important:** In a given session, the **enabled parts must match** on both the robot and the teleoperator.
-> For example, if the robot runs with `--robot.with_mobile_base=false`, the teleoperator must disable the same part `--teleoperator.with_mobile_base=false`.
+For example, if the robot runs with `--robot.with_mobile_base=false`, the teleoperator must disable the same part `--teleoperator.with_mobile_base=false`.
 
 ##### Use the relevant cameras
 
-You can do the same for **cameras**. Enable or disable each camera with default parameters using:
+You can do the same for **cameras**. By default, only the **teleoperation cameras** are recorded (both `left_teleop_camera` and `right_teleop_camera`). Enable or disable each camera with:
 
-```bash
---robot.with_left_teleop_camera=<true|false> \
---robot.with_right_teleop_camera=<true|false> \
+```
+
+--robot.with_left_teleop_camera=<true|false>
+--robot.with_right_teleop_camera=<true|false>
 --robot.with_torso_camera=<true|false>
-```
 
-By default, no camera is recorded, all camera arguments are set to `false`.
-If you want to, you can use custom `width` and `height` parameters for Reachy 2's cameras using the `--robot.camera_width` & `--robot.camera_height` argument:
+````
 
-```bash
---robot.camera_width=1920 \
---robot.camera_height=1080
-```
-
-This will change the resolution of all 3 default robot cameras (enabled by the above bool arguments).
-
-If you want, you can add additional cameras other than the ones in the robot as usual with:
-
-```bash
---robot.cameras="{ extra: {type: opencv, index_or_path: 42, width: 640, height: 480, fps: 30}}" \
-```
 
 ## Step 2: Replay
 
 Make sure the robot is configured with the same parts as the dataset:
 
 ```bash
-lerobot-replay \
+python -m lerobot.replay \
     --robot.type=reachy2 \
     --robot.ip_address=192.168.0.200 \
     --robot.use_external_commands=false \
     --robot.with_mobile_base=false \
     --dataset.repo_id=pollen_robotics/record_test \
     --dataset.episode=0
-```
+    --display_data=true
+````
 
 ## Step 3: Train
 
 ```bash
-lerobot-train \
+python -m lerobot.scripts.train \
   --dataset.repo_id=pollen_robotics/record_test \
   --policy.type=act \
   --output_dir=outputs/train/reachy2_test \
@@ -293,9 +277,10 @@ lerobot-train \
 ## Step 4: Evaluate
 
 ```bash
-lerobot-eval \
+python -m lerobot.record \
   --robot.type=reachy2 \
   --robot.ip_address=192.168.0.200 \
+  --display_data=false \
   --dataset.repo_id=pollen_robotics/eval_record_test \
   --dataset.single_task="Evaluate reachy2 policy" \
   --dataset.num_episodes=10 \

docs/source/sarm.mdx

@@ -1,592 +0,0 @@
-# SARM: Stage-Aware Reward Modeling
-
-SARM (Stage-Aware Reward Modeling) is a video-based reward modeling framework for long-horizon robot manipulation tasks. This guide covers how to train SARM reward models and optionally use them with Reward-Aligned Behavior Cloning (RA-BC).
-
-**Paper**: [SARM: Stage-Aware Reward Modeling for Long Horizon Robot Manipulation](https://arxiv.org/abs/2509.25358)
-
-<img
-  src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/lerobot/lerobot-sarm.png"
-  alt="An overview of SARM"
-  width="80%"
-/>
-
-## Why Reward Models?
-
-Standard behavior cloning treats all demonstration frames equally, but real-world robot datasets are messy. They contain hesitations, corrections, and variable-quality trajectories. Reward models solve this by learning a generalizable notion of **task progress** from demonstrations: given video frames and a task description, they predict how close the robot is to completing the task (0→1). This learned "progress signal" can be used in multiple ways, two promising applications are: (1) **weighted imitation learning** (RA-BC), where high-progress frames receive more weight during policy training, and (2) **reinforcement learning**, where the reward model provides dense rewards for online or offline policy improvement.
-
-## Overview
-
-SARM has following features:
-
-1. **Stage-aware architecture**: Jointly predicts the high-level task stage and fine-grained progress within each stage
-2. **Subtask annotations**: Uses natural language subtask annotations to derive consistent progress labels
-3. **Temporal proportions**: Computes dataset-level priors (α̅\_k) for each subtask to normalize progress across variable-length demonstrations
-
-SARM trains on a compact **stage+tau** target for each frame:
-
-- **stage**: integer stage index `k ∈ {0, ..., K-1}`
-- **τ (tau)**: within-stage progress `τ ∈ [0, 1]`
-- **target encoding**: `y = k + τ` (this is what the dataset processor produces)
-
-At inference time (and in downstream RA-BC), SARM converts the raw `k + τ` value into a **normalized progress** in `[0, 1]` using dataset-level **temporal proportions** `α̅_k` (stored in `meta/temporal_proportions_*.json`).
-
-This matches **Formula (2)** from the paper:
-
-```
-progress_t = P_{k-1} + α̅_k × τ_t
-```
-
-Where:
-
-- `τ_t = (t - s_k) / (e_k - s_k)` is within-subtask normalized time
-- `P_{k-1}` is cumulative prior (sum of previous subtask proportions)
-- `α̅_k` is the temporal proportion for subtask k
-
-This ensures identical task states map to consistent progress values, even across demonstrations of different lengths.
-
-## Inputs and Targets (What the new code expects)
-
-SARM is trained through its processor (`src/lerobot/policies/sarm/processor_sarm.py`), which:
-
-- **Encodes** images and task text with CLIP (ViT-B/32) into `video_features` and `text_features`
-- **Pads/truncates** robot state into `state_features` (up to `max_state_dim`)
-- **Builds targets** as `sparse_targets` (and `dense_targets` in `dense_only`/`dual`) using the stage+tau encoding `y = k + τ`
-- **Masks rewind frames** using a per-sample `lengths` tensor (rewind is a training-time augmentation)
-
-At minimum, each training sample needs:
-
-- `task` (string): task description
-- `policy.image_key` images and `policy.state_key` states from the dataset
-
----
-
-## Annotation Modes
-
-You can choose from **3 annotation modes** that determine how progress labels are computed:
-
-| Mode           | Annotations Required | Heads                        | Use Case                                                     |
-| -------------- | -------------------- | ---------------------------- | ------------------------------------------------------------ |
-| `single_stage` | None                 | Sparse only                  | Simple tasks, quick experiments, no VLM needed               |
-| `dense_only`   | Dense (VLM)          | Dual (sparse auto-generated) | Detailed subtask tracking without defining high-level stages |
-| `dual`         | Sparse + Dense (VLM) | Dual                         | Full SARM paper setup with both granularities                |
-
-### Mode Details
-
-<hfoptions id="mode_explanation">
-<hfoption id="single_stage">
-
-**No annotations required.** The entire episode is treated as a single stage called `"task"`, and progress is linear from 0 to 1 over the episode duration.
-
-- **Sparse head**: 1 stage ("task"), linear progress
-- **Dense head**: Not used
-- **Best for**: Simple tasks, quick experiments, or when VLM annotation is not available
-
-## Set Up Your Environment
-
-1. Install LeRobot by following our [Installation Guide](./installation).
-2. Install SARM dependencies by running:
-
-```bash
-pip install -e ".[sarm]"
-```
-
-Workflow:
-
-```
-1. Train SARM → 2. Visualize predictions → 3. (Optional) Train policy with RA-BC
-```
-
-</hfoption>
-<hfoption id="dense_only">
-
-**Only dense (fine-grained) annotations from a VLM.** The sparse head automatically uses a single `"task"` stage covering the full episode, while the dense head learns detailed subtask progression.
-
-- **Sparse head**: 1 stage ("task"), linear progress (auto-generated)
-- **Dense head**: Multiple fine-grained stages from VLM annotations
-- **Best for**: When you want detailed subtask tracking but don't need to define high-level stages
-
-Workflow:
-
-```
-1. Annotate (dense) → 2. Verify → 3. Train SARM → 4. Visualize → 5. (Optional) Train policy with RA-BC
-```
-
-</hfoption>
-<hfoption id="dual">
-
-**Both sparse and dense annotations from VLM.** Full dual-head mode as described in the SARM paper, with both high-level (sparse) and fine-grained (dense) stage predictions.
-
-- **Sparse head**: High-level stages from VLM annotations
-- **Dense head**: Fine-grained stages from VLM annotations
-- **Best for**: Complex multi-stage tasks where both granularities are useful
-
-Workflow:
-
-```
-1. Annotate (sparse+dense) → 2. Verify → 3. Train SARM → 4. Visualize → 5. (Optional) Train policy with RA-BC
-```
-
-</hfoption>
-</hfoptions>
-
----
-
-## Step 1: Subtask Annotation
-
-<hfoptions id="annotation_mode">
-<hfoption id="single_stage">
-
-**No annotation required!** Skip this step entirely. The model will use the episode's task description and compute linear progress automatically.
-
-</hfoption>
-<hfoption id="dense_only">
-
-Generate **dense (fine-grained) annotations only** using a VLM. The sparse stage will be auto-generated.
-
-```bash
-python src/lerobot/data_processing/sarm_annotations/subtask_annotation.py \
-  --repo-id your-username/your-dataset \
-  --dense-only \
-  --dense-subtasks "Bring robot arms up from starting position,Grab near side and do 1st fold,Grab side and do 2nd fold,Grab side and do 3rd fold to finish folding" \
-  --video-key observation.images.base \
-  --num-workers 4 \
-  --push-to-hub
-```
-
-**What gets saved:**
-
-- `meta/temporal_proportions_sparse.json` - Auto-generated sparse proportions (`{"task": 1.0}`)
-- `meta/temporal_proportions_dense.json` - Dense temporal proportions
-- Per-episode columns in `episodes/*.parquet`:
-  - `dense_subtask_names`, `dense_subtask_start_frames`, `dense_subtask_end_frames`
-  - (also time-based columns: `dense_subtask_start_times`, `dense_subtask_end_times`)
-
-</hfoption>
-<hfoption id="dual">
-
-Generate **both sparse (high-level) and dense (fine-grained) annotations** using a VLM.
-
-```bash
-python src/lerobot/data_processing/sarm_annotations/subtask_annotation.py \
-  --repo-id your-username/your-dataset \
-  --sparse-subtasks "Bring arms up from starting position,Fold the towel (3 folds in total)" \
-  --dense-subtasks "Bring robot arms up from starting position,Grab near side and do 1st fold,Grab side and do 2nd fold,Grab side and do 3rd fold to finish folding" \
-  --video-key observation.images.base \
-  --num-workers 4 \
-  --push-to-hub
-```
-
-**What gets saved:**
-
-- `meta/temporal_proportions_sparse.json` - Sparse temporal proportions
-- `meta/temporal_proportions_dense.json` - Dense temporal proportions
-- Per-episode columns in `episodes/*.parquet`:
-  - `sparse_subtask_names`, `sparse_subtask_start_frames`, `sparse_subtask_end_frames`
-  - `dense_subtask_names`, `dense_subtask_start_frames`, `dense_subtask_end_frames`
-  - (also time-based columns: `*_subtask_start_times`, `*_subtask_end_times`)
-
-</hfoption>
-</hfoptions>
-
-### Annotation Arguments
-
-| Argument               | Description                                                                     |
-| ---------------------- | ------------------------------------------------------------------------------- |
-| `--repo-id`            | HuggingFace dataset repository ID                                               |
-| `--sparse-subtasks`    | Comma-separated list of high-level subtask names                                |
-| `--dense-subtasks`     | Comma-separated list of fine-grained subtask names                              |
-| `--dense-only`         | Generate only dense annotations (auto-creates sparse "task" stage)              |
-| `--video-key`          | Camera/video key to use (e.g., `observation.images.top`)                        |
-| `--num-workers`        | Number of parallel GPU workers (default: 1)                                     |
-| `--episodes`           | Specific episode indices to annotate (default: all)                             |
-| `--skip-existing`      | Skip episodes that already have annotations                                     |
-| `--model`              | VLM model (default: `Qwen/Qwen3-VL-30B-A3B-Instruct`)                           |
-| `--num-visualizations` | Number of episodes to visualize after annotation (default: 5, set to 0 to skip) |
-
-> **Note**: After annotation completes, 5 episodes are automatically visualized by default. Use `--num-visualizations 0` to skip this step.
-
----
-
-## Step 2: Verify Annotations
-
-<hfoptions id="verify_mode">
-<hfoption id="single_stage">
-
-**No verification needed!** Skip this step.
-
-</hfoption>
-<hfoption id="dense_only">
-
-Visualize annotations using the `--visualize-only` flag:
-
-```bash
-python src/lerobot/data_processing/sarm_annotations/subtask_annotation.py \
-  --repo-id your-username/your-dataset \
-  --visualize-only \
-  --visualize-type dense \
-  --num-visualizations 5 \
-  --video-key observation.images.base \
-  --output-dir ./subtask_viz
-```
-
-</hfoption>
-<hfoption id="dual">
-
-Visualize annotations using the `--visualize-only` flag:
-
-```bash
-python src/lerobot/data_processing/sarm_annotations/subtask_annotation.py \
-  --repo-id your-username/your-dataset \
-  --visualize-only \
-  --visualize-type both \
-  --num-visualizations 5 \
-  --video-key observation.images.base \
-  --output-dir ./subtask_viz
-```
-
-</hfoption>
-</hfoptions>
-
-This generates visualizations showing video frames with subtask boundaries overlaid and timeline of subtasks.
-
-### Visualization Arguments
-
-| Argument               | Description                                                    |
-| ---------------------- | -------------------------------------------------------------- |
-| `--visualize-only`     | Only visualize existing annotations (no generation)            |
-| `--num-visualizations` | Number of episodes to visualize (default: 5)                   |
-| `--visualize-type`     | Type of annotations to visualize: `sparse`, `dense`, or `both` |
-
-**Tip**: If annotations are inaccurate, adjust your subtask descriptions to be more specific and re-run.
-
----
-
-## Step 3: Train SARM
-
-<hfoptions id="train_mode">
-<hfoption id="single_stage">
-
-Train with **no annotations** - uses linear progress from 0 to 1:
-
-```bash
-python src/lerobot/scripts/lerobot_train.py \
-  --dataset.repo_id=your-username/your-dataset \
-  --policy.type=sarm \
-  --policy.annotation_mode=single_stage \
-  --policy.image_key=observation.images.base \
-  --output_dir=outputs/train/sarm_single \
-  --batch_size=32 \
-  --steps=5000 \
-  --wandb.enable=true \
-  --wandb.project=sarm \
-  --policy.repo_id=your-username/your-model-name
-```
-
-</hfoption>
-<hfoption id="dense_only">
-
-Train with **dense annotations only** (sparse auto-generated):
-
-```bash
-python src/lerobot/scripts/lerobot_train.py \
-  --dataset.repo_id=your-username/your-dataset \
-  --policy.type=sarm \
-  --policy.annotation_mode=dense_only \
-  --policy.image_key=observation.images.base \
-  --output_dir=outputs/train/sarm_dense \
-  --batch_size=32 \
-  --steps=5000 \
-  --wandb.enable=true \
-  --wandb.project=sarm \
-  --policy.repo_id=your-username/your-model-name
-```
-
-</hfoption>
-<hfoption id="dual">
-
-Train with **both sparse and dense annotations**:
-
-```bash
-python src/lerobot/scripts/lerobot_train.py \
-  --dataset.repo_id=your-username/your-dataset \
-  --policy.type=sarm \
-  --policy.annotation_mode=dual \
-  --policy.image_key=observation.images.base \
-  --output_dir=outputs/train/sarm_dual \
-  --batch_size=32 \
-  --steps=5000 \
-  --wandb.enable=true \
-  --wandb.project=sarm \
-  --policy.repo_id=your-username/your-model-name
-```
-
-</hfoption>
-</hfoptions>
-
-### Multi-GPU Training
-
-Add `accelerate launch --multi_gpu --num_processes=4` to use multiple GPUs for training.
-
-### Training Arguments
-
-| Argument                   | Description                                                       | Default                  |
-| -------------------------- | ----------------------------------------------------------------- | ------------------------ |
-| `--policy.annotation_mode` | `single_stage`, `dense_only`, or `dual`                           | `single_stage`           |
-| `--policy.image_key`       | Camera key for images                                             | `observation.images.top` |
-| `--policy.state_key`       | Key for joint states                                              | `observation.state`      |
-| `--policy.n_obs_steps`     | Observation history steps (total obs frames = `n_obs_steps + 1`)  | `8`                      |
-| `--policy.frame_gap`       | Gap (in frames) between sampled observations (at 30 fps: 30 ≈ 1s) | `30`                     |
-
----
-
-## Step 4: Visualize Predictions
-
-Use `compute_rabc_weights.py` with `--visualize-only` to visualize model predictions (and, if available, annotation-derived targets) without writing a parquet file.
-
-<hfoptions id="viz_mode">
-<hfoption id="single_stage">
-
-```bash
-python src/lerobot/policies/sarm/compute_rabc_weights.py \
-  --dataset-repo-id your-username/your-dataset \
-  --reward-model-path your-username/sarm-model \
-  --visualize-only \
-  --num-visualizations 5 \
-  --head-mode sparse \
-  --output-dir ./sarm_viz
-```
-
-</hfoption>
-<hfoption id="dense_only">
-
-```bash
-python src/lerobot/policies/sarm/compute_rabc_weights.py \
-  --dataset-repo-id your-username/your-dataset \
-  --reward-model-path your-username/sarm-model \
-  --visualize-only \
-  --num-visualizations 5 \
-  --head-mode dense \
-  --output-dir ./sarm_viz
-```
-
-</hfoption>
-<hfoption id="dual">
-
-```bash
-python src/lerobot/policies/sarm/compute_rabc_weights.py \
-  --dataset-repo-id your-username/your-dataset \
-  --reward-model-path your-username/sarm-model \
-  --visualize-only \
-  --num-visualizations 5 \
-  --head-mode both \
-  --output-dir ./sarm_viz
-```
-
-</hfoption>
-</hfoptions>
-
-The visualization shows:
-
-- **Progress plot**: Predicted progress (and optional annotation-derived “GT” when available and `--stride 1`)
-- **Stage probabilities**: Stacked area plot of predicted stage probabilities
-- **Sample frames**: Key frames from the episode with progress/stage labels
-
-### Visualization Arguments
-
-| Argument               | Description                                               |
-| ---------------------- | --------------------------------------------------------- |
-| `--visualize-only`     | Only visualize predictions (no RABC computation)          |
-| `--num-visualizations` | Number of episodes to visualize (default: 5)              |
-| `--head-mode`          | SARM head to use: `sparse`, `dense`, or `both`            |
-| `--stride`             | Compute every N frames, interpolate the rest (default: 1) |
-
----
-
-## Step 5 (Optional): Train Policy with RA-BC
-
-Reward-Aligned Behavior Cloning (RA-BC) uses the trained SARM model to weight training samples based on predicted progress improvement. This requires two steps:
-
-1. **Precompute progress values** for all frames using the trained SARM model
-2. **Train policy** with RA-BC weighting using the precomputed values
-
-### How RA-BC Works
-
-For each training sample, RA-BC computes the progress delta:
-
-```
-r_i = φ(o_{t+Δ}) - φ(o_t)
-```
-
-Where `φ` is the SARM progress prediction and `Δ` is the policy's `chunk_size`. Samples with positive progress (good demonstrations) get higher weights, while samples with negative or zero progress get down-weighted.
-
-The weighting follows **Equations 8-9** from the paper:
-
-- **Soft weight**: `w̃_i = clip((r_i − (μ − 2σ)) / (4σ + ε), 0, 1)`
-- **Final weight**: `w_i = 𝟙{r_i > κ} + 𝟙{0 ≤ r_i ≤ κ} × w̃_i`
-
-### Step 5a: Compute SARM Progress Values
-
-First, run the SARM model on all frames in your dataset to compute progress values:
-
-```bash
-python src/lerobot/policies/sarm/compute_rabc_weights.py \
-  --dataset-repo-id your-username/your-dataset \
-  --reward-model-path your-username/sarm-model \
-  --head-mode sparse \
-  --num-visualizations 5 \
-  --push-to-hub
-```
-
-This script:
-
-- Processes all frames and computes progress values
-- Saves progress values to a parquet file next to the dataset on disk (defaults to `<dataset_root>/sarm_progress.parquet`)
-- Generates visualizations of the first N episodes (default: 5)
-
-**Arguments:**
-
-| Argument               | Description                                                    | Default    |
-| ---------------------- | -------------------------------------------------------------- | ---------- |
-| `--reward-model-path`  | Path to trained SARM model                                     | (required) |
-| `--head-mode`          | SARM head to use: `sparse`, `dense`, or `both`                 | `sparse`   |
-| `--device`             | Device for inference                                           | `cuda`     |
-| `--visualize-only`     | Only visualize predictions (no RA-BC computation)              | `false`    |
-| `--num-visualizations` | Number of episodes to visualize (default: 5, set to 0 to skip) | `5`        |
-
-**Output format** (`sarm_progress.parquet`):
-
-| Column            | Description                                    |
-| ----------------- | ---------------------------------------------- |
-| `index`           | Global frame index in dataset                  |
-| `episode_index`   | Episode number                                 |
-| `frame_index`     | Local frame index within episode               |
-| `progress_sparse` | Sparse head progress value [0, 1]              |
-| `progress_dense`  | Dense head progress value [0, 1] (if computed) |
-
-### Step 5b: Train Policy with RA-BC
-
-Once you have the progress file, train your policy with RA-BC weighting. The progress file is auto-detected from the dataset path (`sarm_progress.parquet`). Currently PI0, PI0.5 and SmolVLA are supported with RA-BC:
-
-```bash
-python src/lerobot/scripts/lerobot_train.py \
-  --dataset.repo_id=your-username/your-dataset \
-  --policy.type=pi0 \
-  --use_rabc=true \
-  --rabc_head_mode=sparse \
-  --rabc_kappa=0.01 \
-  --output_dir=outputs/train/policy_rabc \
-  --batch_size=32 \
-  --steps=40000
-```
-
-The training script automatically:
-
-- Loads the precomputed progress values from the parquet file
-- Uses the policy's `chunk_size` to compute progress deltas (Δ)
-- Computes sample weights based on progress improvement
-- Applies weighted loss during training
-
-**RA-BC Arguments:**
-
-| Argument               | Description                                                | Default                            |
-| ---------------------- | ---------------------------------------------------------- | ---------------------------------- |
-| `--use_rabc`           | Enable RA-BC sample weighting                              | `false`                            |
-| `--rabc_progress_path` | Path to progress parquet file (auto-detected from dataset) | `sarm_progress.parquet` in dataset |
-| `--rabc_head_mode`     | Which SARM head's progress to use: `sparse` or `dense`     | `sparse`                           |
-| `--rabc_kappa`         | Threshold κ for high-quality samples                       | `0.01`                             |
-
-### Tuning RA-BC Kappa
-
-The `kappa` parameter is the threshold that determines which samples get full weight (w=1). Understanding how to tune it is critical for RA-BC to work effectively.
-
-**How the weighting works:**
-
-| Condition           | Weight                  |
-| ------------------- | ----------------------- |
-| `delta > kappa`     | 1.0 (hard threshold)    |
-| `0 ≤ delta ≤ kappa` | Soft weight from Eq. 8  |
-| `delta < 0`         | 0.0 (negative progress) |
-
-**Diagnosing kappa issues:**
-
-Monitor these WandB metrics during training:
-
-| Metric             | Healthy Range | Problem Indicator         |
-| ------------------ | ------------- | ------------------------- |
-| `rabc_mean_weight` | 0.3 - 0.8     | ≈ 1.0 means kappa too low |
-| `rabc_delta_mean`  | > 0           | Should be positive        |
-| `rabc_delta_std`   | > 0           | Variance in data quality  |
-
-**If `rabc_mean_weight ≈ 1.0`:** Your kappa is too low. Most samples have `delta > kappa` and bypass the soft-weighting entirely. RA-BC becomes equivalent to vanilla BC.
-
-**Setting kappa based on your data:**
-
-The default `kappa=0.01` was tuned for the paper's T-shirt folding task (~90s episodes at 30fps). For your dataset, check the logged `rabc_delta_mean` and `rabc_delta_std`:
-
-```
-# If delta_mean ≈ 0.03 and delta_std ≈ 0.02:
-# Most deltas fall in range [0.01, 0.05]
-
-# Option 1: Set kappa = delta_mean (medium selectivity)
---rabc_kappa=0.03
-
-# Option 2: Set kappa = delta_mean + delta_std (high selectivity)
---rabc_kappa=0.05
-
-# Option 3: Set kappa = delta_mean + 2*delta_std (very selective)
---rabc_kappa=0.07
-```
-
-**When RA-BC may not help:**
-
-If your dataset is already high quality (consistent progress across all demonstrations), RA-BC won't provide much benefit since there's nothing to filter.
-
-### Multi-GPU Training with RA-BC
-
-```bash
-accelerate launch \
-  --multi_gpu \
-  --num_processes=4 \
-  src/lerobot/scripts/lerobot_train.py \
-  --dataset.repo_id=your-username/your-dataset \
-  --policy.type=pi0 \
-  --use_rabc=true \
-  --rabc_kappa=0.01 \
-  --output_dir=outputs/train/policy_rabc \
-  --batch_size=32 \
-  --steps=40000
-```
-
----
-
-## Tips & Best Practices
-
-### Choosing a Mode
-
-- **Start with `single_stage`** for quick experiments - no annotation overhead
-- Use **`dense_only`** when you want detailed progress tracking but tasks don't have clear high-level stages
-- Use **`dual`** for complex tasks where both coarse and fine-grained progress is meaningful
-
-### Annotation Quality
-
-1. **Be specific with subtask names**: Instead of "fold", use "grab near side and fold toward center"
-2. **Verify with visualization**: Always check a few episodes before training
-3. **Consistent naming**: Use the same subtask names across all episodes
-
-### RA-BC
-
-1. **Train SARM first**: RA-BC quality depends entirely on SARM quality
-2. **Monitor `rabc_mean_weight`**: If it's ≈ 1.0, increase kappa (see [Tuning RA-BC Kappa](#tuning-ra-bc-kappa))
-
----
-
-## Citation
-
-```bibtex
-@article{chen2025sarm,
-  title={SARM: Stage-Aware Reward Modeling for Long Horizon Robot Manipulation},
-  author={Chen, Qianzhong and Yu, Justin and Schwager, Mac and Abbeel, Pieter and Shentu, Yide and Wu, Philipp},
-  journal={arXiv preprint arXiv:2509.25358},
-  year={2025}
-}
-```

docs/source/so100.mdx

@@ -103,7 +103,7 @@ lerobot-setup-motors \
 
 <!-- prettier-ignore-start -->
 ```python
-from lerobot.robots.so_follower import SO100Follower, SO100FollowerConfig
+from lerobot.robots.so100_follower import SO100Follower, SO100FollowerConfig
 
 config = SO100FollowerConfig(
     port="/dev/tty.usbmodem585A0076841",
@@ -177,7 +177,7 @@ lerobot-setup-motors \
 
 <!-- prettier-ignore-start -->
 ```python
-from lerobot.teleoperators.so_leader import SO100Leader, SO100LeaderConfig
+from lerobot.teleoperators.so100_leader import SO100Leader, SO100LeaderConfig
 
 config = SO100LeaderConfig(
     port="/dev/tty.usbmodem585A0076841",
@@ -579,7 +579,7 @@ lerobot-calibrate \
 
 <!-- prettier-ignore-start -->
 ```python
-from lerobot.robots.so_follower import SO100FollowerConfig, SO100Follower
+from lerobot.robots.so100_follower import SO100FollowerConfig, SO100Follower
 
 config = SO100FollowerConfig(
     port="/dev/tty.usbmodem585A0076891",
@@ -617,7 +617,7 @@ lerobot-calibrate \
 
 <!-- prettier-ignore-start -->
 ```python
-from lerobot.teleoperators.so_leader import SO100LeaderConfig, SO100Leader
+from lerobot.teleoperators.so100_leader import SO100LeaderConfig, SO100Leader
 
 config = SO100LeaderConfig(
     port="/dev/tty.usbmodem58760431551",

docs/source/so101.mdx

@@ -30,191 +30,6 @@ The follower arm uses 6x STS3215 motors with 1/345 gearing. The leader, however,
 | Wrist Roll          |   5   |  1 / 147   |
 | Gripper             |   6   |  1 / 147   |
 
-## Configure the motors
-
-### 1. Find the USB ports associated with each arm
-
-To find the port for each bus servo adapter, connect MotorBus to your computer via USB and power. Run the following script and disconnect the MotorBus when prompted:
-
-```bash
-lerobot-find-port
-```
-
-<hfoptions id="example">
-<hfoption id="Mac">
-
-Example output:
-
-```
-Finding all available ports for the MotorBus.
-['/dev/tty.usbmodem575E0032081', '/dev/tty.usbmodem575E0031751']
-Remove the USB cable from your MotorsBus and press Enter when done.
-
-[...Disconnect corresponding leader or follower arm and press Enter...]
-
-The port of this MotorsBus is /dev/tty.usbmodem575E0032081
-Reconnect the USB cable.
-```
-
-Where the found port is: `/dev/tty.usbmodem575E0032081` corresponding to your leader or follower arm.
-
-</hfoption>
-<hfoption id="Linux">
-
-On Linux, you might need to give access to the USB ports by running:
-
-```bash
-sudo chmod 666 /dev/ttyACM0
-sudo chmod 666 /dev/ttyACM1
-```
-
-Example output:
-
-```
-Finding all available ports for the MotorBus.
-['/dev/ttyACM0', '/dev/ttyACM1']
-Remove the usb cable from your MotorsBus and press Enter when done.
-
-[...Disconnect corresponding leader or follower arm and press Enter...]
-
-The port of this MotorsBus is /dev/ttyACM1
-Reconnect the USB cable.
-```
-
-Where the found port is: `/dev/ttyACM1` corresponding to your leader or follower arm.
-
-</hfoption>
-</hfoptions>
-
-### 2. Set the motors ids and baudrates
-
-Each motor is identified by a unique id on the bus. When brand new, motors usually come with a default id of `1`. For the communication to work properly between the motors and the controller, we first need to set a unique, different id to each motor. Additionally, the speed at which data is transmitted on the bus is determined by the baudrate. In order to talk to each other, the controller and all the motors need to be configured with the same baudrate.
-
-To that end, we first need to connect to each motor individually with the controller in order to set these. Since we will write these parameters in the non-volatile section of the motors' internal memory (EEPROM), we'll only need to do this once.
-
-If you are repurposing motors from another robot, you will probably also need to perform this step as the ids and baudrate likely won't match.
-
-The video below shows the sequence of steps for setting the motor ids.
-
-##### Setup motors video
-
-<div class="video-container">
-  <video controls width="600">
-    <source
-      src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/lerobot/setup_motors_so101_2.mp4"
-      type="video/mp4"
-    />
-  </video>
-</div>
-
-#### Follower
-
-Connect the usb cable from your computer and the power supply to the follower arm's controller board. Then, run the following command or run the API example with the port you got from the previous step. You'll also need to give your leader arm a name with the `id` parameter.
-
-<hfoptions id="setup_motors">
-<hfoption id="Command">
-
-```bash
-lerobot-setup-motors \
-    --robot.type=so101_follower \
-    --robot.port=/dev/tty.usbmodem585A0076841  # <- paste here the port found at previous step
-```
-
-</hfoption>
-<hfoption id="API example">
-
-<!-- prettier-ignore-start -->
-```python
-from lerobot.robots.so_follower import SO101Follower, SO101FollowerConfig
-
-config = SO101FollowerConfig(
-    port="/dev/tty.usbmodem585A0076841",
-    id="my_awesome_follower_arm",
-)
-follower = SO101Follower(config)
-follower.setup_motors()
-```
-<!-- prettier-ignore-end -->
-
-</hfoption>
-</hfoptions>
-
-You should see the following instruction
-
-```bash
-Connect the controller board to the 'gripper' motor only and press enter.
-```
-
-As instructed, plug the gripper's motor. Make sure it's the only motor connected to the board, and that the motor itself is not yet daisy-chained to any other motor. As you press `[Enter]`, the script will automatically set the id and baudrate for that motor.
-
-<details>
-<summary>Troubleshooting</summary>
-
-If you get an error at that point, check your cables and make sure they are plugged in properly:
-
-<ul>
-  <li>Power supply</li>
-  <li>USB cable between your computer and the controller board</li>
-  <li>The 3-pin cable from the controller board to the motor</li>
-</ul>
-
-If you are using a Waveshare controller board, make sure that the two jumpers are set on the `B` channel (USB).
-
-</details>
-
-You should then see the following message:
-
-```bash
-'gripper' motor id set to 6
-```
-
-Followed by the next instruction:
-
-```bash
-Connect the controller board to the 'wrist_roll' motor only and press enter.
-```
-
-You can disconnect the 3-pin cable from the controller board, but you can leave it connected to the gripper motor on the other end, as it will already be in the right place. Now, plug in another 3-pin cable to the wrist roll motor and connect it to the controller board. As with the previous motor, make sure it is the only motor connected to the board and that the motor itself isn't connected to any other one.
-
-Repeat the operation for each motor as instructed.
-
-> [!TIP]
-> Check your cabling at each step before pressing Enter. For instance, the power supply cable might disconnect as you manipulate the board.
-
-When you are done, the script will simply finish, at which point the motors are ready to be used. You can now plug the 3-pin cable from each motor to the next one, and the cable from the first motor (the 'shoulder pan' with id=1) to the controller board, which can now be attached to the base of the arm.
-
-#### Leader
-
-Do the same steps for the leader arm.
-
-<hfoptions id="setup_motors">
-<hfoption id="Command">
-
-```bash
-lerobot-setup-motors \
-    --teleop.type=so101_leader \
-    --teleop.port=/dev/tty.usbmodem575E0031751  # <- paste here the port found at previous step
-```
-
-</hfoption>
-<hfoption id="API example">
-
-<!-- prettier-ignore-start -->
-```python
-from lerobot.teleoperators.so_leader import SO101Leader, SO101LeaderConfig
-
-config = SO101LeaderConfig(
-    port="/dev/tty.usbmodem585A0076841",
-    id="my_awesome_leader_arm",
-)
-leader = SO101Leader(config)
-leader.setup_motors()
-```
-<!-- prettier-ignore-end -->
-
-</hfoption>
-</hfoptions>
-
 ### Clean Parts
 
 Remove all support material from the 3D-printed parts. The easiest way to do this is using a small screwdriver to get underneath the support material.
@@ -340,6 +155,191 @@ It is advisable to install one 3-pin cable in the motor after placing them befor
 </hfoption>
 </hfoptions>
 
+## Configure the motors
+
+### 1. Find the USB ports associated with each arm
+
+To find the port for each bus servo adapter, connect MotorBus to your computer via USB and power. Run the following script and disconnect the MotorBus when prompted:
+
+```bash
+lerobot-find-port
+```
+
+<hfoptions id="example">
+<hfoption id="Mac">
+
+Example output:
+
+```
+Finding all available ports for the MotorBus.
+['/dev/tty.usbmodem575E0032081', '/dev/tty.usbmodem575E0031751']
+Remove the USB cable from your MotorsBus and press Enter when done.
+
+[...Disconnect corresponding leader or follower arm and press Enter...]
+
+The port of this MotorsBus is /dev/tty.usbmodem575E0032081
+Reconnect the USB cable.
+```
+
+Where the found port is: `/dev/tty.usbmodem575E0032081` corresponding to your leader or follower arm.
+
+</hfoption>
+<hfoption id="Linux">
+
+On Linux, you might need to give access to the USB ports by running:
+
+```bash
+sudo chmod 666 /dev/ttyACM0
+sudo chmod 666 /dev/ttyACM1
+```
+
+Example output:
+
+```
+Finding all available ports for the MotorBus.
+['/dev/ttyACM0', '/dev/ttyACM1']
+Remove the usb cable from your MotorsBus and press Enter when done.
+
+[...Disconnect corresponding leader or follower arm and press Enter...]
+
+The port of this MotorsBus is /dev/ttyACM1
+Reconnect the USB cable.
+```
+
+Where the found port is: `/dev/ttyACM1` corresponding to your leader or follower arm.
+
+</hfoption>
+</hfoptions>
+
+### 2. Set the motors ids and baudrates
+
+Each motor is identified by a unique id on the bus. When brand new, motors usually come with a default id of `1`. For the communication to work properly between the motors and the controller, we first need to set a unique, different id to each motor. Additionally, the speed at which data is transmitted on the bus is determined by the baudrate. In order to talk to each other, the controller and all the motors need to be configured with the same baudrate.
+
+To that end, we first need to connect to each motor individually with the controller in order to set these. Since we will write these parameters in the non-volatile section of the motors' internal memory (EEPROM), we'll only need to do this once.
+
+If you are repurposing motors from another robot, you will probably also need to perform this step as the ids and baudrate likely won't match.
+
+The video below shows the sequence of steps for setting the motor ids.
+
+##### Setup motors video
+
+<div class="video-container">
+  <video controls width="600">
+    <source
+      src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/lerobot/setup_motors_so101_2.mp4"
+      type="video/mp4"
+    />
+  </video>
+</div>
+
+#### Follower
+
+Connect the usb cable from your computer and the power supply to the follower arm's controller board. Then, run the following command or run the API example with the port you got from the previous step. You'll also need to give your leader arm a name with the `id` parameter.
+
+<hfoptions id="setup_motors">
+<hfoption id="Command">
+
+```bash
+lerobot-setup-motors \
+    --robot.type=so101_follower \
+    --robot.port=/dev/tty.usbmodem585A0076841  # <- paste here the port found at previous step
+```
+
+</hfoption>
+<hfoption id="API example">
+
+<!-- prettier-ignore-start -->
+```python
+from lerobot.robots.so101_follower import SO101Follower, SO101FollowerConfig
+
+config = SO101FollowerConfig(
+    port="/dev/tty.usbmodem585A0076841",
+    id="my_awesome_follower_arm",
+)
+follower = SO101Follower(config)
+follower.setup_motors()
+```
+<!-- prettier-ignore-end -->
+
+</hfoption>
+</hfoptions>
+
+You should see the following instruction
+
+```bash
+Connect the controller board to the 'gripper' motor only and press enter.
+```
+
+As instructed, plug the gripper's motor. Make sure it's the only motor connected to the board, and that the motor itself is not yet daisy-chained to any other motor. As you press `[Enter]`, the script will automatically set the id and baudrate for that motor.
+
+<details>
+<summary>Troubleshooting</summary>
+
+If you get an error at that point, check your cables and make sure they are plugged in properly:
+
+<ul>
+  <li>Power supply</li>
+  <li>USB cable between your computer and the controller board</li>
+  <li>The 3-pin cable from the controller board to the motor</li>
+</ul>
+
+If you are using a Waveshare controller board, make sure that the two jumpers are set on the `B` channel (USB).
+
+</details>
+
+You should then see the following message:
+
+```bash
+'gripper' motor id set to 6
+```
+
+Followed by the next instruction:
+
+```bash
+Connect the controller board to the 'wrist_roll' motor only and press enter.
+```
+
+You can disconnect the 3-pin cable from the controller board, but you can leave it connected to the gripper motor on the other end, as it will already be in the right place. Now, plug in another 3-pin cable to the wrist roll motor and connect it to the controller board. As with the previous motor, make sure it is the only motor connected to the board and that the motor itself isn't connected to any other one.
+
+Repeat the operation for each motor as instructed.
+
+> [!TIP]
+> Check your cabling at each step before pressing Enter. For instance, the power supply cable might disconnect as you manipulate the board.
+
+When you are done, the script will simply finish, at which point the motors are ready to be used. You can now plug the 3-pin cable from each motor to the next one, and the cable from the first motor (the 'shoulder pan' with id=1) to the controller board, which can now be attached to the base of the arm.
+
+#### Leader
+
+Do the same steps for the leader arm.
+
+<hfoptions id="setup_motors">
+<hfoption id="Command">
+
+```bash
+lerobot-setup-motors \
+    --teleop.type=so101_leader \
+    --teleop.port=/dev/tty.usbmodem575E0031751  # <- paste here the port found at previous step
+```
+
+</hfoption>
+<hfoption id="API example">
+
+<!-- prettier-ignore-start -->
+```python
+from lerobot.teleoperators.so101_leader import SO101Leader, SO101LeaderConfig
+
+config = SO101LeaderConfig(
+    port="/dev/tty.usbmodem585A0076841",
+    id="my_awesome_leader_arm",
+)
+leader = SO101Leader(config)
+leader.setup_motors()
+```
+<!-- prettier-ignore-end -->
+
+</hfoption>
+</hfoptions>
+
 ## Calibrate
 
 Next, you'll need to calibrate your robot to ensure that the leader and follower arms have the same position values when they are in the same physical position.
@@ -364,7 +364,7 @@ lerobot-calibrate \
 
 <!-- prettier-ignore-start -->
 ```python
-from lerobot.robots.so_follower import SO101FollowerConfig, SO101Follower
+from lerobot.robots.so101_follower import SO101FollowerConfig, SO101Follower
 
 config = SO101FollowerConfig(
     port="/dev/tty.usbmodem585A0076891",
@@ -413,7 +413,7 @@ lerobot-calibrate \
 
 <!-- prettier-ignore-start -->
 ```python
-from lerobot.teleoperators.so_leader import SO101LeaderConfig, SO101Leader
+from lerobot.teleoperators.so101_leader import SO101LeaderConfig, SO101Leader
 
 config = SO101LeaderConfig(
     port="/dev/tty.usbmodem58760431551",

docs/source/torch_accelerators.mdx

@@ -1,42 +0,0 @@
-# PyTorch accelerators
-
-LeRobot supports multiple hardware acceleration options for both training and inference.
-
-These options include:
-
-- **CPU**: CPU executes all computations, no dedicated accelerator is used
-- **CUDA**: acceleration with NVIDIA & AMD GPUs
-- **MPS**: acceleration with Apple Silicon GPUs
-- **XPU**: acceleration with Intel integrated and discrete GPUs
-
-## Getting Started
-
-To use particular accelerator, a suitable version of PyTorch should be installed.
-
-For CPU, CUDA, and MPS backends follow instructions provided on [PyTorch installation page](https://pytorch.org/get-started/locally).
-For XPU backend, follow instructions from [PyTorch documentation](https://docs.pytorch.org/docs/stable/notes/get_start_xpu.html).
-
-### Verifying the installation
-
-After installation, accelerator availability can be verified by running
-
-```python
-import torch
-print(torch.<backend_name>.is_available())  # <backend_name> is cuda, mps, or xpu
-```
-
-## How to run training or evaluation
-
-To select the desired accelerator, use the `--policy.device` flag when running `lerobot-train` or `lerobot-eval`. For example, to use MPS on Apple Silicon, run:
-
-```bash
-lerobot-train
-    --policy.device=mps ...
-```
-
-```bash
-lerobot-eval \
-    --policy.device=mps ...
-```
-
-However, in most cases, presence of an accelerator is detected automatically and `policy.device` parameter can be omitted from CLI commands.

docs/source/unitree_g1.mdx

@@ -1,21 +1,20 @@
-# Unitree G1
+# Unitree G1 Robot Setup and Control
 
 This guide covers the complete setup process for the Unitree G1 humanoid, from initial connection to running gr00t_wbc locomotion.
 
-## About
+## About the Unitree G1
 
-We support both 29 and 23 DOF G1 EDU version. We introduce:
+We offer support for both 29 and 23 DOF G1. In this first PR we introduce:
 
-- **`unitree g1` robot class, handling low level read/write from/to the humanoid**
-- **ZMQ socket bridge** for remote communication and camera streaming, allowing for remote policy deployment over wlan, eth or directly on the robot
-- **Locomotion policies** from NVIDIA gr00t and Amazon FAR Holosoma
-- **Simulation mode** for testing policies without the physical robot in mujoco
+- **`unitree g1` robot class, handling low level communication with the humanoid**
+- **ZMQ socket bridge** for remote communication over WiFi, allowing one to deploy policies remotely instead of over ethernet or directly on the Orin
+- **GR00T locomotion policy** for bipedal walking and balance
 
 ---
 
-## Connection guide
+## Part 1: Connect to Robot over Ethernet
 
-### Step 1: Configure Ethernet Interface
+### Step 1: Configure Your Computer's Ethernet Interface
 
 Set a static IP on the same subnet as the robot:
 
@@ -26,7 +25,7 @@ sudo ip addr add 192.168.123.200/24 dev enp131s0
 sudo ip link set enp131s0 up
 ```
 
-**Note**: The G1's Ethernet IP is fixed at `192.168.123.164`. Your computer must use `192.168.123.x` with x ≠ 164.
+**Note**: The robot's Ethernet IP is fixed at `192.168.123.164`. Your computer must use `192.168.123.x` where x ≠ 164.
 
 ### Step 2: SSH into the Robot
 
@@ -35,24 +34,25 @@ ssh unitree@192.168.123.164
 # Password: 123
 ```
 
-You should now be connected to the G1's Orin.
+You should now be connected to the robot's onboard computer.
 
 ---
 
 ## Part 2: Enable WiFi on the Robot
 
-Wlan0 is disabled by default on the G1. To enable it:
+Once connected via Ethernet, follow these steps to enable WiFi:
 
 ### Step 1: Enable WiFi Hardware
 
 ```bash
+# Unblock WiFi radio
 sudo rfkill unblock wifi
 sudo rfkill unblock all
 
-# Bring up wlan0
+# Bring up WiFi interface
 sudo ip link set wlan0 up
 
-# Enable NetworkManager control of wlan0
+# Enable NetworkManager control
 sudo nmcli radio wifi on
 sudo nmcli device set wlan0 managed yes
 sudo systemctl restart NetworkManager
@@ -72,7 +72,7 @@ sudo iptables -A FORWARD -i wlp132s0f0 -o enp131s0 -m state --state RELATED,ESTA
 sudo iptables -A FORWARD -i enp131s0 -o wlp132s0f0 -j ACCEPT
 ```
 
-**On the G1:**
+**On the robot:**
 
 ```bash
 # Add laptop as default gateway
@@ -110,7 +110,7 @@ ssh unitree@<YOUR_ROBOT_IP>
 # Password: 123
 ```
 
-Replace `<YOUR_ROBOT_IP>` with your robot's actual WiFi IP address.
+Replace `<YOUR_ROBOT_IP>` with your robot's actual WiFi IP address (e.g., `172.18.129.215`).
 
 ---
 
@@ -146,9 +146,9 @@ python src/lerobot/robots/unitree_g1/run_g1_server.py
 
 ---
 
-## Part 4: Controlling the robot
+## Part 4: Running GR00T Locomotion
 
-With the robot server running, you can now control the robot remotely. Let's launch a locomotion policy
+With the robot server running, you can now control the robot from your laptop.
 
 ### Step 1: Install LeRobot on your machine
 
@@ -171,30 +171,30 @@ Edit the config file to match your robot's WiFi IP:
 robot_ip: str = "<YOUR_ROBOT_IP>"  # Replace with your robot's WiFi IP.
 ```
 
+**Note**: When running directly on the G1 (not remotely), set `robot_ip: str = "127.0.0.1"` instead.
+
 ### Step 3: Run the Locomotion Policy
 
 ```bash
 # Run GR00T locomotion controller
 python examples/unitree_g1/gr00t_locomotion.py --repo-id "nepyope/GR00T-WholeBodyControl_g1"
-
-# Run Holosoma locomotion controller
-python examples/unitree_g1/holosoma_locomotion.py
-
 ```
 
+### Step 4: Control with Remote
+
+- **Left stick**: Forward/backward and left/right movement
+- **Right stick**: Rotation
+- **R1 button**: Raise waist height
+- **R2 button**: Lower waist height
+
 Press `Ctrl+C` to stop the policy.
 
 ---
 
-## Running in Simulation Mode (MuJoCo)
-
-You can now test policies before unleashing them on the physical robot using MuJoCo. To do so simply set `is_simulation=True` in config.
-
 ## Additional Resources
 
 - [Unitree SDK Documentation](https://github.com/unitreerobotics/unitree_sdk2_python)
-- [GR00T-WholeBodyControl](https://github.com/NVlabs/GR00T-WholeBodyControl)
-- [Holosoma](https://github.com/amazon-far/holosoma)
+- [GR00T Policy Repository](https://huggingface.co/nepyope/GR00T-WholeBodyControl_g1)
 - [LeRobot Documentation](https://github.com/huggingface/lerobot)
 - [Unitree_IL_Lerobot](https://github.com/unitreerobotics/unitree_IL_lerobot)
 

docs/source/using_dataset_tools.mdx

@@ -11,14 +11,13 @@ LeRobot provides several utilities for manipulating datasets:
 3. **Merge Datasets** - Combine multiple datasets into one. The datasets must have identical features, and episodes are concatenated in the order specified in `repo_ids`
 4. **Add Features** - Add new features to a dataset
 5. **Remove Features** - Remove features from a dataset
-6. **Convert to Video** - Convert image-based datasets to video format for efficient storage
 
 The core implementation is in `lerobot.datasets.dataset_tools`.
 An example script detailing how to use the tools API is available in `examples/dataset/use_dataset_tools.py`.
 
 ## Command-Line Tool: lerobot-edit-dataset
 
-`lerobot-edit-dataset` is a command-line script for editing datasets. It can be used to delete episodes, split datasets, merge datasets, add features, remove features, and convert image datasets to video format.
+`lerobot-edit-dataset` is a command-line script for editing datasets. It can be used to delete episodes, split datasets, merge datasets, add features, and remove features.
 
 Run `lerobot-edit-dataset --help` for more information on the configuration of each operation.
 
@@ -87,78 +86,9 @@ lerobot-edit-dataset \
     --operation.feature_names "['observation.images.top']"
 ```
 
-#### Convert to Video
-
-Convert an image-based dataset to video format, creating a new LeRobotDataset where images are stored as videos. This is useful for reducing storage requirements and improving data loading performance. The new dataset will have the exact same structure as the original, but with images encoded as MP4 videos in the proper LeRobot format.
-
-```bash
-# Local-only: Save to a custom output directory (no hub push)
-lerobot-edit-dataset \
-    --repo_id lerobot/pusht_image \
-    --operation.type convert_image_to_video \
-    --operation.output_dir /path/to/output/pusht_video
-
-# Save with new repo_id (local storage)
-lerobot-edit-dataset \
-    --repo_id lerobot/pusht_image \
-    --new_repo_id lerobot/pusht_video \
-    --operation.type convert_image_to_video
-
-# Convert and push to Hugging Face Hub
-lerobot-edit-dataset \
-    --repo_id lerobot/pusht_image \
-    --new_repo_id lerobot/pusht_video \
-    --operation.type convert_image_to_video \
-    --push_to_hub true
-
-# Convert with custom video codec and quality settings
-lerobot-edit-dataset \
-    --repo_id lerobot/pusht_image \
-    --operation.type convert_image_to_video \
-    --operation.output_dir outputs/pusht_video \
-    --operation.vcodec libsvtav1 \
-    --operation.pix_fmt yuv420p \
-    --operation.g 2 \
-    --operation.crf 30
-
-# Convert only specific episodes
-lerobot-edit-dataset \
-    --repo_id lerobot/pusht_image \
-    --operation.type convert_image_to_video \
-    --operation.output_dir outputs/pusht_video \
-    --operation.episode_indices "[0, 1, 2, 5, 10]"
-
-# Convert with multiple workers for parallel processing
-lerobot-edit-dataset \
-    --repo_id lerobot/pusht_image \
-    --operation.type convert_image_to_video \
-    --operation.output_dir outputs/pusht_video \
-    --operation.num_workers 8
-
-# For memory-constrained systems, users can now specify limits:
-lerobot-edit-dataset \
-    --repo_id lerobot/pusht_image \
-    --operation.type convert_to_video \
-    --operation.max_episodes_per_batch 50 \
-    --operation.max_frames_per_batch 10000
-```
-
-**Parameters:**
-
-- `output_dir`: Custom output directory (optional - by default uses `new_repo_id` or `{repo_id}_video`)
-- `vcodec`: Video codec to use - options: `h264`, `hevc`, `libsvtav1` (default: `libsvtav1`)
-- `pix_fmt`: Pixel format - options: `yuv420p`, `yuv444p` (default: `yuv420p`)
-- `g`: Group of pictures (GOP) size - lower values give better quality but larger files (default: 2)
-- `crf`: Constant rate factor - lower values give better quality but larger files, 0 is lossless (default: 30)
-- `fast_decode`: Fast decode tuning option (default: 0)
-- `episode_indices`: List of specific episodes to convert (default: all episodes)
-- `num_workers`: Number of parallel workers for processing (default: 4)
-
-**Note:** The resulting dataset will be a proper LeRobotDataset with all cameras encoded as videos in the `videos/` directory, with parquet files containing only metadata (no raw image data). All episodes, stats, and tasks are preserved.
-
 ### Push to Hub
 
-Add the `--push_to_hub true` flag to any command to automatically upload the resulting dataset to the Hugging Face Hub:
+Add the `--push_to_hub` flag to any command to automatically upload the resulting dataset to the Hugging Face Hub:
 
 ```bash
 lerobot-edit-dataset \
@@ -166,45 +96,7 @@ lerobot-edit-dataset \
     --new_repo_id lerobot/pusht_after_deletion \
     --operation.type delete_episodes \
     --operation.episode_indices "[0, 2, 5]" \
-    --push_to_hub true
+    --push_to_hub
 ```
 
 There is also a tool for adding features to a dataset that is not yet covered in `lerobot-edit-dataset`.
-
-# Dataset Visualization
-
-## Online Visualization
-
-When you record a dataset using `lerobot`, it automatically uploads to the Hugging Face Hub unless you specify otherwise. To view the dataset online, use our **LeRobot Dataset Visualizer**, available at:
-https://huggingface.co/spaces/lerobot/visualize_dataset
-
-## Local Visualization
-
-You can also visualize episodes from a dataset locally using our command-line tool.
-
-**From the Hugging Face Hub:**
-
-```bash
-lerobot-dataset-viz \
-    --repo-id lerobot/pusht \
-    --episode-index 0
-```
-
-**From a local folder:**
-Add the `--root` option and set `--mode local`. For example, to search in `./my_local_data_dir/lerobot/pusht`:
-
-```bash
-lerobot-dataset-viz \
-    --repo-id lerobot/pusht \
-    --root ./my_local_data_dir \
-    --mode local \
-    --episode-index 0
-```
-
-Once executed, the tool opens `rerun.io` and displays the camera streams, robot states, and actions for the selected episode.
-
-For advanced usage—including visualizing datasets stored on a remote server—run:
-
-```bash
-lerobot-dataset-viz --help
-```

docs/source/walloss.mdx

@@ -1,80 +0,0 @@
-# WALL-OSS
-
-WALL-OSS is an open-source foundation model for embodied intelligence, proposed by the [XSquare Robot](https://x2robot.com/en/research/68bc2cde8497d7f238dde690) team in 2025. The LeRobot implementation is adapted from their open-source [WallX](https://github.com/X-Square-Robot/wall-x) repository.
-
-X Square Robot’s WALL-OSS is now integrated into Hugging Face’s LeRobot ecosystem. This is an exciting collaborative project between the LeRobot and X Square Robot teams. You can now post-train, evaluate, and deploy WALL-OSS directly through LeRobot. With this, we’re aiming to make it easier for the open-source robotics community to customize and deploy WALL-OSS foundation models. Read and explore WALL-OSS [paper](https://arxiv.org/pdf/2509.11766) and [code](https://github.com/X-Square-Robot/wall-x).
-
-## Model Overview
-
-The WALL-OSS team is building the embodied foundation model to capture and compress the world's most valuable data: the continuous, high-fidelity stream of physical interaction. By creating a direct feedback loop between the model's decisions and the body's lived experience, the emergence of a truly generalizable intelligence is enabled—one that understands not just how the world works, but how to act effectively within it.
-
-<img
-  src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/lerobot/walloss-lerobot-paper.png"
-  alt="An overview of WALL-OSS"
-  width="85%"
-/>
-
-Technically, WALL-OSS introduces a tightly coupled multimodal architecture (tightly-coupled MoE structure) that integrates both discrete and continuous action modeling strategies. Through a two-stage training pipeline (Inspiration → Integration), the model gradually unifies semantic reasoning and high-frequency action generation. Its core innovations include:
-
-- **Embodied perception–enhanced multimodal pretraining**: Large-scale training on unified vision–language–action data to strengthen spatial, causal, and manipulation understanding.
-- **Unified Cross-Level Chain-of-Thought (Uni-CoT)**: A single differentiable framework that unifies high-level instruction reasoning, sub-task decomposition, and fine-grained action synthesis, forming a continuous chain from “understanding” to “execution.”
-- **Mixture-of-Experts (MoE) action heads**: Dynamically activating experts depending on the task phase and modeling actions in discrete or continuous space to maintain stable VLM priors.
-- **Two-stage training paradigm**:
-  - **Inspiration stage**: Injecting discrete action priors to strengthen spatial understanding and semantic-action alignment.
-  - **Integration stage**: Using flow matching to achieve high-frequency continuous control.
-
-## Installation Requirements
-
-1. Install LeRobot by following our [Installation Guide](./installation).
-2. Install WallX dependencies by running:
-
-   ```bash
-   pip install -e ".[wallx]"
-   ```
-
-## Usage
-
-To use WallX in LeRobot, specify the policy type as:
-
-```python
-policy.type=wall_x
-```
-
-## Training
-
-For training WallX, you can use the standard LeRobot training script with the appropriate configuration:
-
-```bash
-python src/lerobot/scripts/lerobot_train.py \
-    --dataset.repo_id=your_dataset \
-    --policy.type=wall_x \
-    --output_dir=./outputs/wallx_training \
-    --job_name=wallx_training \
-    --policy.repo_id=your_repo_id \
-    --policy.pretrained_name_or_path=x-square-robot/wall-oss-flow \
-    --policy.prediction_mode=diffusion \
-    --policy.attn_implementation=eager \
-    --steps=3000 \
-    --policy.device=cuda \
-    --batch_size=32
-```
-
-### Training Arguments
-
-| Argument                       | Description                                                                                                                                                   |
-| ------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------- |
-| `--dataset.repo_id`            | The Hugging Face Hub repository ID for your training dataset (e.g., `lerobot/aloha_sim_insertion_human`)                                                      |
-| `--policy.type`                | Specifies using the WallX policy architecture                                                                                                                 |
-| `--output_dir`                 | Local directory where training checkpoints and logs will be saved                                                                                             |
-| `--job_name`                   | A name identifier for this training run (used in logging/tracking)                                                                                            |
-| `--policy.repo_id`             | Your Hugging Face Hub repo ID where the trained model will be pushed                                                                                          |
-| `--policy.pretrained_path`     | Path to pretrained WallX weights to initialize from (the official WALL-OSS checkpoint)                                                                        |
-| `--policy.prediction_mode`     | The action prediction strategy: `diffusion` or `fast` - `diffusion` uses iterative denoising for action generation, `fast` uses next token prediction instead |
-| `--policy.attn_implementation` | Attention implementation backend - `eager` uses standard PyTorch attention (alternatives include `flash_attention_2` or `sdpa`)                               |
-| `--steps`                      | Total number of training steps to run                                                                                                                         |
-| `--policy.device`              | Device to train on (`cuda` for GPU, `cpu` for CPU)                                                                                                            |
-| `--batch_size`                 | Number of samples per training batch                                                                                                                          |
-
-## License
-
-This model follows the **Apache 2.0 License**, consistent with the original [WallX repository](https://github.com/X-Square-Robot/wall-x).

docs/source/xvla.mdx

@@ -1,528 +0,0 @@
-# X-VLA: The First Soft-Prompted Robot Foundation Model for Any Robot, Any Task
-
-## Overview
-
-For years, robotics has aspired to build agents that can follow natural human instructions and operate dexterously across many environments and robot bodies. Recent breakthroughs in LLMs and VLMs suggest a path forward: extend these foundation-model architectures to embodied control by grounding them in actions. This has led to the rise of Vision-Language-Action (VLA) models, with the hope that a single generalist model could combine broad semantic understanding with robust manipulation skills.
-
-But training such models is difficult. Robot data is fragmented across platforms, sensors, embodiments, and collection protocols. Heterogeneity appears everywhere: different arm configurations, different action spaces, different camera setups, different visual domains, and different task distributions. These inconsistencies create major distribution shifts that make pretraining unstable and adaptation unreliable.
-
-Inspired by meta-learning and prompt learning, we ask: **"What if a VLA model could learn the structure of each robot and dataset the same way LLMs learn tasks, through prompts?"**
-
-**X-VLA** is a soft-prompted, flow-matching VLA framework that treats each hardware setup as a "task" and encodes it using a small set of learnable embeddings. These **Soft Prompts** capture embodiment and domain-specific variations, guiding the Transformer from the earliest stages of multimodal fusion. With this mechanism, X-VLA can reconcile diverse robot morphologies, data types, and sensor setups within a single unified architecture.
-
-<p align="center">
-  <img
-    src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/lerobot/xvla-architecture.png"
-    alt="XVLA Architecture"
-    style="max-width: 100%; height: auto; width: 800px;"
-  />
-</p>
-
-Built from pure Transformer encoders, X-VLA scales naturally with model size and dataset diversity. Across 6 simulation benchmarks and 3 real robots, Soft Prompts consistently outperform existing methods in handling hardware and domain differences. X-VLA-0.9B, trained on 290K episodes spanning seven robotic platforms, learns an embodiment-agnostic generalist policy in Phase I, and adapts efficiently to new robots in Phase II simply by learning a new set of prompts, while keeping the backbone frozen.
-
-<p align="center">
-  <img
-    src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/lerobot/xvla-architecture2.png"
-    alt="XVLA Architecture 2"
-    style="width: 60%; height: auto;"
-  />
-</p>
-
-With only 1% of parameters tuned (9M), X-VLA-0.9B achieves near-π₀ performance on LIBERO and Simpler-WidowX, despite using **300× fewer trainable parameters**. It also demonstrates strong real-world dexterity with minimal demonstrations, including folding cloths in under two minutes.
-
-<p align="center">
-  <img
-    src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/lerobot/xvla-fold.png"
-    alt="XVLA fold visualization"
-    style="width: 95%; max-width: 1100px; height: auto;"
-  />
-</p>
-
-X-VLA shows that generalist robot intelligence does not require increasingly complex architectures, only the right way to absorb heterogeneity. Soft Prompts offer a simple, scalable mechanism for unifying diverse robotic data, paving the way toward adaptable, cross-embodiment robot foundation models.
-
-## Installation
-
-After installing LeRobot, install the X-VLA dependencies:
-
-```bash
-pip install -e .[xvla]
-```
-
-After the new release, you'll be able to do:
-
-```bash
-pip install lerobot[xvla]
-```
-
-## Quick Start
-
-### Basic Usage
-
-To use X-VLA in your LeRobot configuration, specify the policy type as:
-
-```bash
-policy.type=xvla
-```
-
-### Evaluating Pre-trained Checkpoints
-
-Example evaluation with LIBERO:
-
-```bash
-lerobot-eval \
-  --policy.path="lerobot/xvla-libero" \
-  --env.type=libero \
-  --env.task=libero_spatial,libero_goal,libero_10 \
-  --env.control_mode=absolute \
-  --eval.batch_size=1 \
-  --eval.n_episodes=1 \
-  --env.episode_length=800 \
-  --seed=142
-```
-
-## Available Checkpoints
-
-### 🎯 Base Model
-
-**[lerobot/xvla-base](https://huggingface.co/lerobot/xvla-base)**
-
-A 0.9B parameter instantiation of X-VLA, trained with a carefully designed data processing and learning recipe. The training pipeline consists of two phases:
-
-- **Phase I: Pretraining** - Pretrained on 290K episodes from Droid, Robomind, and Agibot, spanning seven platforms across five types of robotic arms (single-arm to bi-manual setups). By leveraging soft prompts to absorb embodiment-specific variations, the model learns an embodiment-agnostic generalist policy.
-
-- **Phase II: Domain Adaptation** - Adapted to deployable policies for target domains. A new set of soft prompts is introduced and optimized to encode the hardware configuration of the novel domain, while the pretrained backbone remains frozen.
-
-### Simulation Checkpoints
-
-**[lerobot/xvla-libero](https://huggingface.co/lerobot/xvla-libero)**
-
-Achieves 93% success rate on LIBERO benchmarks. Fine-tuned from the base model for simulation tasks.
-
-**[lerobot/xvla-widowx](https://huggingface.co/lerobot/xvla-widowx)**
-
-Fine-tuned on BridgeData for pick-and-place experiments on compact WidowX platforms. Demonstrates robust manipulation capabilities.
-
-### 🤖 Real-World Checkpoints
-
-**[lerobot/xvla-folding](https://huggingface.co/lerobot/xvla-folding)**
-
-A fine-tuned dexterous manipulation model trained on the high-quality Soft-FOLD cloth folding dataset. Achieves 100% success rate over 2 hours of continuous cloth folding.
-
-**[lerobot/xvla-agibot-world](https://huggingface.co/lerobot/xvla-agibot-world)**
-
-Optimized for AgileX robot dexterous manipulation tasks.
-
-**[lerobot/xvla-google-robot](https://huggingface.co/lerobot/xvla-google-robot)**
-
-Adapted for Google Robot platforms.
-
-## Training X-VLA
-
-### Recommended Training Configuration
-
-When fine-tuning X-VLA for a new embodiment or task, we recommend not freezing the VLM, and also setting the `policy.dtype=bfloat16` to not hit OOM errors.
-
-```bash
-lerobot-train \
-  --dataset.repo_id=YOUR_DATASET \
-  --output_dir=./outputs/xvla_training \
-  --job_name=xvla_training \
-  --policy.path="lerobot/xvla-base" \
-  --policy.repo_id="HF_USER/xvla-your-robot" \
-  --policy.dtype=bfloat16 \
-  --policy.action_mode=auto \
-  --steps=20000 \
-  --policy.device=cuda \
-  --policy.freeze_vision_encoder=false \
-  --policy.freeze_language_encoder=false \
-  --policy.train_policy_transformer=true \
-  --policy.train_soft_prompts=true \
-```
-
-### Training Parameters Explained
-
-| Parameter                  | Default | Description                                    |
-| -------------------------- | ------- | ---------------------------------------------- |
-| `freeze_vision_encoder`    | `false` | Do not freeze the VLM vision encoder weights   |
-| `freeze_language_encoder`  | `false` | Do not freeze the VLM language encoder weights |
-| `train_policy_transformer` | `true`  | Allow policy transformer layers to train       |
-| `train_soft_prompts`       | `true`  | Allow soft prompts to train                    |
-
-**💡 Best Practice**: For Phase II adaptation to new embodiments, do not freeze the VLM encoders and also train the policy transformer and soft prompts.
-
-### Example: Training on Bimanual Robot
-
-```bash
-lerobot-train \
-  --dataset.repo_id=pepijn223/bimanual-so100-handover-cube \
-  --output_dir=./outputs/xvla_bimanual \
-  --job_name=xvla_so101_training \
-  --policy.path="lerobot/xvla-base" \
-  --policy.dtype=bfloat16 \
-  --policy.repo_id="YOUR_USERNAME/xvla-biso101" \
-  --steps=3000 \
-  --policy.device=cuda \
-  --policy.action_mode=so101_bimanual \
-  --policy.freeze_vision_encoder=false \
-  --policy.freeze_language_encoder=false \
-  --policy.train_policy_transformer=true \
-  --policy.train_soft_prompts=true
-```
-
-💡 **Best Performance:** If you have sufficient computational resources and want to achieve best X-VLA finetuning performance, you should follow the official finetuning strategy:
-
-**🔥 Full-finetune all components with a custom learning-rate scheme**
-
-To ensure stable optimization, the Vision-Language Model (VLM) must be trained with only 1/10 of the base learning rate, while all other components use the full LR.
-This LR ratio is crucial for achieving strong and stable finetuning performance. This is already done for you by default.
-❕Note
-
-Completely matching the official reported performance may require an additional warm-up LR schedule for soft-prompts, which can bring minor improvements.
-We encourage implementing this in your customized training pipeline for optimal results.
-
-## Core Concepts
-
-### 1. Action Modes
-
-X-VLA uses an **Action Registry** system to handle different action spaces and embodiments. The `action_mode` parameter defines how actions are processed, what loss functions are used, and how predictions are post-processed.
-
-#### Available Action Modes
-
-| Action Mode      | Action Dim              | Description                                 | Use Case                             |
-| ---------------- | ----------------------- | ------------------------------------------- | ------------------------------------ |
-| `ee6d`           | 20                      | End-effector with xyz, 6D rotation, gripper | Dual-arm setups with spatial control |
-| `joint`          | 14                      | Joint-space with gripper                    | Direct joint control robots          |
-| `agibot_ee6d`    | 20                      | AGI-bot variant with MSE loss               | AGI-bot platforms                    |
-| `so101_bimanual` | 20 (model), 12 (real)   | SO101 bimanual robot                        | Bimanual manipulation tasks          |
-| `auto`           | 20 (model), auto (real) | Auto-detects action dim from dataset        | **Recommended** for new robots       |
-
-#### Why Action Modes Matter
-
-When you have a pretrained checkpoint like `lerobot/xvla-base` trained with `action_dim=20`, and you want to train on a dataset with a different action dimension (e.g., 14 for bimanual arms), you can't simply trim the action dimension. The action mode orchestrates:
-
-1. **Loss Computation**: Different loss functions for different action components (MSE for joints, BCE for grippers, etc.)
-2. **Preprocessing**: Zeroing out gripper channels, padding dimensions
-3. **Postprocessing**: Applying sigmoid to gripper logits, trimming padding
-
-#### Example: BimanualSO101 Action Space
-
-The `so101_bimanual` action mode handles the mismatch between model output (20D) and real robot control (12D):
-
-```python
-# Model outputs 20 dimensions for compatibility
-dim_action = 20
-
-# Real robot only needs 12 dimensions
-# [left_arm (6), right_arm (6)] = [joints (5) + gripper (1)] × 2
-REAL_DIM = 12
-
-# Preprocessing: Pad 12D actions to 20D for training
-# Postprocessing: Trim 20D predictions to 12D for deployment
-```
-
-See the [action_hub.py](/home/jade_choghari/robot/lerobot/src/lerobot/policies/xvla/action_hub.py) implementation for details.
-
-#### Auto Action Mode (Recommended)
-
-The `auto` action mode is the easiest way to use X-VLA with any robot. It automatically detects your dataset's action dimension and handles padding/trimming:
-
-```bash
-lerobot-train \
-  --policy.path="lerobot/xvla-base" \
-  --policy.action_mode=auto \
-  --policy.max_action_dim=20 \
-  ...
-```
-
-**How it works:**
-
-- Reads `action_feature.shape[-1]` from your dataset (e.g., 7 for Franka)
-- Model outputs `max_action_dim` (default 20) for pretrained compatibility
-- Loss is computed **only on the real dimensions**: `MSE(pred[:,:,:real_dim], target[:,:,:real_dim])`
-- Postprocess trims output back to `real_dim` for robot control
-
-This eliminates the need to create custom action modes for most robots.
-
-### 2. Domain IDs
-
-Domain IDs are learnable identifiers for different robot configurations and camera setups. They allow X-VLA to distinguish between:
-
-- Different robots (Robot 1 vs Robot 2)
-- Different camera configurations (cam1 vs cam2)
-- Different combinations (Robot1-cam1-cam2 vs Robot1-cam1 vs Robot2-cam1)
-
-#### Setting Domain IDs
-
-**During Training**: By default, domain_id is set to 0 for general training.
-
-**During Evaluation**: Specify the domain_id that matches your checkpoint's training configuration.
-
-```python
-# Example: LIBERO checkpoint uses domain_id=3
-domain_id = 3
-```
-
-The domain_id is automatically added to observations by the `XVLAAddDomainIdProcessorStep` in the preprocessing pipeline.
-
-The `lerobot/xvla-base` model has been trained on the following domain IDs. It is recommended to choose one that most resembles your robot/configuration:
-
-#### Fine-tuning Datasets
-
-| Dataset Name     | Domain ID |
-| ---------------- | --------- |
-| Bridge           | 0         |
-| RT1              | 1         |
-| Calvin           | 2         |
-| libero           | 3         |
-| widowx-air       | 4         |
-| AIR-AGILEX-HQ    | 5         |
-| robotwin2_abs_ee | 6         |
-| robotwin2_clean  | 6         |
-| robocasa-human   | 7         |
-| VLABench         | 8         |
-| AGIBOT-challenge | 9         |
-| AIR-AGILEX       | 10        |
-| AIRBOT           | 18        |
-
-### 3. Processor Steps
-
-X-VLA requires specific preprocessing and postprocessing steps for proper operation.
-
-#### Required Preprocessing Steps
-
-1. **XVLAImageToFloatProcessorStep**: Converts images from [0, 255] to [0, 1] range
-2. **XVLAImageNetNormalizeProcessorStep**: Applies ImageNet normalization (required for VLM backbone)
-3. **XVLAAddDomainIdProcessorStep**: Adds domain_id to observations
-
-#### Example Custom Processor
-
-For LIBERO environments, a custom processor handles the specific observation format:
-
-```python
-from lerobot.policies.xvla.processor_xvla import LiberoProcessorStep
-
-processor = LiberoProcessorStep()
-# Handles robot_state dictionary, converts rotation matrices to 6D representation
-# Applies 180° image rotation for camera convention
-```
-
-### 4. Configuration Parameters
-
-Key configuration parameters for X-VLA:
-
-```python
-# Observation and action
-n_obs_steps: int = 1          # Number of observation timesteps
-chunk_size: int = 32           # Action sequence length
-n_action_steps: int = 32       # Number of action steps to execute
-
-# Model architecture
-hidden_size: int = 1024        # Transformer hidden dimension
-depth: int = 24                # Number of transformer layers
-num_heads: int = 16            # Number of attention heads
-num_domains: int = 30          # Maximum number of domain IDs
-len_soft_prompts: int = 32     # Length of soft prompt embeddings
-
-# Action space
-action_mode: str = "ee6d"      # Action space type (use "auto" for auto-detection)
-use_proprio: bool = True       # Use proprioceptive state
-max_state_dim: int = 32        # Maximum state dimension
-max_action_dim: int = 20       # Max action dim for padding (used by "auto" mode)
-
-# Vision
-num_image_views: int | None    # Number of camera views
-resize_imgs_with_padding: tuple[int, int] | None  # Target image size with padding
-
-# Training
-num_denoising_steps: int = 10  # Flow matching denoising steps
-```
-
-## Creating Custom Action Modes
-
-If your robot has a unique action space, you can create a custom action mode:
-
-### Step 1: Define Your Action Space
-
-```python
-from lerobot.policies.xvla.action_hub import BaseActionSpace, register_action
-import torch.nn as nn
-
-@register_action("my_custom_robot")
-class MyCustomActionSpace(BaseActionSpace):
-    """Custom action space for my robot."""
-
-    dim_action = 15  # Your robot's action dimension
-    gripper_idx = (7, 14)  # Gripper channel indices
-
-    def __init__(self):
-        super().__init__()
-        self.mse = nn.MSELoss()
-        self.bce = nn.BCEWithLogitsLoss()
-
-    def compute_loss(self, pred, target):
-        """Define your loss computation."""
-        # Example: MSE for joints, BCE for grippers
-        joints_loss = self.mse(pred[:, :, :7], target[:, :, :7])
-        gripper_loss = self.bce(pred[:, :, self.gripper_idx],
-                                target[:, :, self.gripper_idx])
-
-        return {
-            "joints_loss": joints_loss,
-            "gripper_loss": gripper_loss,
-        }
-
-    def preprocess(self, proprio, action, mode="train"):
-        """Preprocess actions before training."""
-        # Example: Zero out grippers in proprioception
-        proprio_m = proprio.clone()
-        action_m = action.clone() if action is not None else None
-        proprio_m[..., self.gripper_idx] = 0.0
-        if action_m is not None:
-            action_m[..., self.gripper_idx] = 0.0
-        return proprio_m, action_m
-
-    def postprocess(self, action):
-        """Post-process predictions for deployment."""
-        # Example: Apply sigmoid to gripper logits
-        action[..., self.gripper_idx] = torch.sigmoid(action[..., self.gripper_idx])
-        return action
-```
-
-### Step 2: Use Your Custom Action Mode
-
-```bash
-lerobot-train \
-  --policy.action_mode=my_custom_robot \
-  --dataset.repo_id=YOUR_DATASET \
-  --policy.path="lerobot/xvla-base" \
-  ...
-```
-
-## Advanced Topics
-
-### Multi-Camera Support
-
-X-VLA supports multiple camera views through the `num_image_views` parameter:
-
-```python
-# Configure for 3 camera views
-policy.num_image_views=3
-
-# Add empty cameras if you have fewer physical cameras
-policy.empty_cameras=1  # Adds 1 zero-padded camera view
-```
-
-### Custom Preprocessing Pipeline
-
-Create a custom preprocessing pipeline for your environment:
-
-```python
-from lerobot.processor import PolicyProcessorPipeline
-from lerobot.policies.xvla.processor_xvla import (
-    XVLAImageToFloatProcessorStep,
-    XVLAImageNetNormalizeProcessorStep,
-    XVLAAddDomainIdProcessorStep,
-)
-
-# Build custom pipeline
-preprocessor = PolicyProcessorPipeline(
-    steps=[
-        YourCustomProcessorStep(),  # Your custom processing
-        XVLAImageToFloatProcessorStep(),  # Required: convert to float
-        XVLAImageNetNormalizeProcessorStep(),  # Required: ImageNet norm
-        XVLAAddDomainIdProcessorStep(domain_id=5),  # Your domain ID
-    ]
-)
-```
-
-### Handling Different Action Dimensions
-
-When your dataset has fewer action dimensions than the pretrained model:
-
-**Option 1 (Recommended)**: Use `auto` action mode
-
-```bash
-# Automatically detects your dataset's action dimension
-# Works with any robot without custom code
-policy.action_mode=auto
-policy.max_action_dim=20  # Match pretrained model
-```
-
-**Option 2**: Use a predefined action mode with built-in padding
-
-```python
-# Model expects 20D, dataset has 12D
-# Action mode handles padding internally
-action_mode = "so101_bimanual"  # Pads 12 → 20
-```
-
-**Option 2**: Create a custom action mode that maps dimensions explicitly
-
-```python
-@register_action("my_mapped_action")
-class MappedActionSpace(BaseActionSpace):
-    dim_action = 20
-    REAL_DIM = 12
-
-    def _pad_to_model_dim(self, x):
-        # Custom padding logic
-        ...
-```
-
-## Troubleshooting
-
-### Common Issues
-
-**Issue**: "Action dimension mismatch"
-
-- **Solution**: Check that your `action_mode` matches your robot's action space. Create a custom action mode if needed.
-
-**Issue**: "Image values outside [0, 1] range"
-
-- **Solution**: Ensure images are preprocessed with `XVLAImageToFloatProcessorStep` before normalization.
-
-**Issue**: "Domain ID not found"
-
-- **Solution**: Make sure `XVLAAddDomainIdProcessorStep` is in your preprocessing pipeline with the correct domain_id.
-
-**Issue**: "Low success rate on new embodiment"
-
-- **Solution**:
-  1. Verify your action_mode is correct
-  2. Check that soft prompts are being trained (`train_soft_prompts=True`)
-  3. Ensure proper preprocessing (ImageNet normalization, domain_id)
-  4. Consider increasing training steps
-
-**Issue**: "Out of memory during training"
-
-- **Solution**:
-  1. Reduce `chunk_size` (e.g., from 32 to 16)
-  2. Enable gradient checkpointing
-  3. Reduce batch size
-  4. Freeze more components
-
-## Citation
-
-If you use X-VLA in your research, please cite:
-
-```bibtex
-@article{zheng2025x,
-  title   = {X-VLA: Soft-Prompted Transformer as Scalable Cross-Embodiment Vision-Language-Action Model},
-  author  = {Zheng, Jinliang and Li, Jianxiong and Wang, Zhihao and Liu, Dongxiu and Kang, Xirui
-             and Feng, Yuchun and Zheng, Yinan and Zou, Jiayin and Chen, Yilun and Zeng, Jia and others},
-  journal = {arXiv preprint arXiv:2510.10274},
-  year    = {2025}
-}
-```
-
-## Additional Resources
-
-- [X-VLA Paper](https://arxiv.org/pdf/2510.10274)
-- [LeRobot Documentation](https://github.com/huggingface/lerobot)
-- [Action Registry Implementation](https://github.com/huggingface/lerobot/src/lerobot/policies/xvla/action_hub.py)
-- [Processor Implementation](https://github.com/huggingface/lerobot/src/lerobot/policies/xvla/processor_xvla.py)
-- [Model Configuration](https://github.com/huggingface/lerobot/src/lerobot/policies/xvla/configuration_xvla.py)
-
-## Contributing
-
-We welcome contributions! If you've implemented a new action mode or processor for your robot, please consider submitting a PR to help the community.

examples/backward_compatibility/replay.py

@@ -41,7 +41,8 @@ from lerobot.robots import (  # noqa: F401
     RobotConfig,
     koch_follower,
     make_robot_from_config,
-    so_follower,
+    so100_follower,
+    so101_follower,
 )
 from lerobot.utils.constants import ACTION
 from lerobot.utils.robot_utils import precise_sleep
@@ -96,7 +97,7 @@ def replay(cfg: ReplayConfig):
         robot.send_action(action)
 
         dt_s = time.perf_counter() - start_episode_t
-        precise_sleep(max(1 / dataset.fps - dt_s, 0.0))
+        precise_sleep(1 / dataset.fps - dt_s)
 
     robot.disconnect()
 

examples/dataset/aggregate_egodex.py

@@ -0,0 +1,245 @@
+#!/usr/bin/env python
+
+# Copyright 2025 The HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""
+Aggregate EgoDex shards into a single dataset.
+
+After distributed processing creates multiple shards, this script combines
+them into a single unified dataset.
+
+Reference: https://arxiv.org/abs/2505.11709, https://github.com/apple/ml-egodex
+"""
+
+import argparse
+import logging
+from pathlib import Path
+
+from datatrove.executor import LocalPipelineExecutor
+from datatrove.executor.slurm import SlurmPipelineExecutor
+from datatrove.pipeline.base import PipelineStep
+
+
+class AggregateEgoDexDatasets(PipelineStep):
+    """Datatrove pipeline step for aggregating EgoDex shards."""
+
+    def __init__(
+        self,
+        repo_ids: list[str],
+        aggregated_repo_id: str,
+        local_dir: Path | str | None = None,
+        push_to_hub: bool = False,
+        hf_repo_id: str | None = None,
+    ):
+        super().__init__()
+        self.repo_ids = repo_ids
+        self.aggr_repo_id = aggregated_repo_id
+        self.local_dir = Path(local_dir) if local_dir else None
+        self.push_to_hub = push_to_hub
+        self.hf_repo_id = hf_repo_id if hf_repo_id else aggregated_repo_id
+
+    def run(self, data=None, rank: int = 0, world_size: int = 1):
+        import logging
+
+        from lerobot.datasets.aggregate import aggregate_datasets
+        from lerobot.datasets.lerobot_dataset import LeRobotDataset
+        from lerobot.utils.utils import init_logging
+
+        init_logging()
+
+        # Only worker 0 performs aggregation (aggregate_datasets handles parallelism internally)
+        if rank == 0:
+            logging.info(f"Starting aggregation of {len(self.repo_ids)} shards into {self.aggr_repo_id}")
+
+            # Build roots list if local_dir is specified
+            roots = None
+            if self.local_dir:
+                roots = [self.local_dir / repo_id for repo_id in self.repo_ids]
+                # Filter to only existing directories
+                existing_roots = [r for r in roots if r.exists()]
+                if len(existing_roots) != len(self.repo_ids):
+                    logging.warning(
+                        f"Only {len(existing_roots)} of {len(self.repo_ids)} shard directories found. "
+                        "Missing shards will be skipped."
+                    )
+                # Update repo_ids to match existing roots
+                existing_repo_ids = [
+                    repo_id for repo_id, r in zip(self.repo_ids, roots, strict=False) if r.exists()
+                ]
+                roots = existing_roots
+                self.repo_ids = existing_repo_ids
+
+            if len(self.repo_ids) == 0:
+                logging.error("No shard directories found. Nothing to aggregate.")
+                return
+
+            aggr_root = self.local_dir / self.aggr_repo_id if self.local_dir else None
+
+            aggregate_datasets(
+                repo_ids=self.repo_ids,
+                aggr_repo_id=self.aggr_repo_id,
+                roots=roots,
+                aggr_root=aggr_root,
+            )
+            logging.info("Aggregation complete!")
+
+            # Push to Hugging Face Hub if requested
+            if self.push_to_hub:
+                logging.info(f"Pushing to Hugging Face Hub as {self.hf_repo_id}...")
+                dataset = LeRobotDataset(
+                    repo_id=self.aggr_repo_id,
+                    root=aggr_root,
+                )
+                # Update repo_id for pushing to different HF account if specified
+                dataset.repo_id = self.hf_repo_id
+                dataset.push_to_hub(
+                    tags=["egodex", "hand", "dexterous", "lerobot"],
+                    license="cc-by-nc-nd-4.0",
+                )
+                logging.info("Push to hub complete!")
+        else:
+            logging.info(f"Worker {rank} skipping - only worker 0 performs aggregation")
+
+
+def make_aggregate_executor(
+    repo_id,
+    num_shards,
+    job_name,
+    logs_dir,
+    partition,
+    cpus_per_task,
+    mem_per_cpu,
+    local_dir,
+    push_to_hub,
+    hf_repo_id,
+    slurm=True,
+):
+    """Create executor for aggregating EgoDex shards."""
+    # Generate repo IDs for all shards
+    repo_ids = [f"{repo_id}_world_{num_shards}_rank_{rank}" for rank in range(num_shards)]
+
+    kwargs = {
+        "pipeline": [
+            AggregateEgoDexDatasets(repo_ids, repo_id, local_dir, push_to_hub, hf_repo_id),
+        ],
+        "logging_dir": str(logs_dir / job_name),
+    }
+
+    if slurm:
+        kwargs.update(
+            {
+                "job_name": job_name,
+                "tasks": 1,  # Only need 1 task for aggregation
+                "workers": 1,  # Only need 1 worker
+                "time": "24:00:00",  # 24 hours for aggregation
+                "partition": partition,
+                "cpus_per_task": cpus_per_task,
+                "sbatch_args": {"mem-per-cpu": mem_per_cpu},
+            }
+        )
+        executor = SlurmPipelineExecutor(**kwargs)
+    else:
+        kwargs.update(
+            {
+                "tasks": 1,
+                "workers": 1,
+            }
+        )
+        executor = LocalPipelineExecutor(**kwargs)
+
+    return executor
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Aggregate EgoDex dataset shards into a single unified dataset."
+    )
+
+    parser.add_argument(
+        "--repo-id",
+        type=str,
+        required=True,
+        help="Repository identifier (base name without shard suffix, e.g., pepijn/egodex-test)",
+    )
+    parser.add_argument(
+        "--num-shards",
+        type=int,
+        required=True,
+        help="Number of shards to aggregate (must match --workers from slurm_port_egodex.py)",
+    )
+    parser.add_argument(
+        "--logs-dir",
+        type=Path,
+        default=Path("logs"),
+        help="Path to logs directory for datatrove",
+    )
+    parser.add_argument(
+        "--job-name",
+        type=str,
+        default="aggr_egodex",
+        help="Job name used in SLURM",
+    )
+    parser.add_argument(
+        "--slurm",
+        type=int,
+        default=1,
+        help="Launch over SLURM. Use --slurm 0 to launch locally (for debugging)",
+    )
+    parser.add_argument(
+        "--partition",
+        type=str,
+        help="SLURM partition (ideally CPU partition)",
+    )
+    parser.add_argument(
+        "--cpus-per-task",
+        type=int,
+        default=16,
+        help="Number of CPUs for aggregation task",
+    )
+    parser.add_argument(
+        "--mem-per-cpu",
+        type=str,
+        default="8G",
+        help="Memory per CPU for aggregation",
+    )
+    parser.add_argument(
+        "--local-dir",
+        type=Path,
+        default=None,
+        help="Local directory where shards are stored. If not specified, uses default HF cache.",
+    )
+    parser.add_argument(
+        "--push-to-hub",
+        action="store_true",
+        help="Push aggregated dataset to Hugging Face Hub after aggregation.",
+    )
+    parser.add_argument(
+        "--hf-repo-id",
+        type=str,
+        default=None,
+        help="Hugging Face repo ID for upload (e.g., username/dataset-name). Defaults to --repo-id.",
+    )
+
+    args = parser.parse_args()
+    kwargs = vars(args)
+    kwargs["slurm"] = kwargs.pop("slurm") == 1
+
+    aggregate_executor = make_aggregate_executor(**kwargs)
+    aggregate_executor.run()
+
+
+if __name__ == "__main__":
+    main()
+

examples/dataset/download_egodex.sh

@@ -0,0 +1,129 @@
+#!/bin/bash
+
+# Download EgoDex dataset
+# Reference: https://arxiv.org/abs/2505.11709, https://github.com/apple/ml-egodex
+#
+# Usage: ./download_egodex.sh [output_dir] [parts...]
+#
+# Examples:
+#   ./download_egodex.sh ./data test           # Download test set only (16 GB)
+#   ./download_egodex.sh ./data part1 part2    # Download training parts 1 and 2
+#   ./download_egodex.sh ./data all            # Download everything (~1.7 TB)
+#
+# Available parts:
+#   test   - Test set (16 GB)
+#   part1  - Training set part 1 (300 GB)
+#   part2  - Training set part 2 (300 GB)
+#   part3  - Training set part 3 (300 GB)
+#   part4  - Training set part 4 (300 GB)
+#   part5  - Training set part 5 (300 GB)
+#   extra  - Additional data (200 GB)
+#   all    - Download all parts (~1.7 TB total)
+
+set -e
+
+BASE_URL="https://ml-site.cdn-apple.com/datasets/egodex"
+
+# Map part names to filenames
+declare -A PART_FILES=(
+    ["test"]="test.zip"
+    ["part1"]="part1.zip"
+    ["part2"]="part2.zip"
+    ["part3"]="part3.zip"
+    ["part4"]="part4.zip"
+    ["part5"]="part5.zip"
+    ["extra"]="extra.zip"
+)
+
+ALL_PARTS=("test" "part1" "part2" "part3" "part4" "part5" "extra")
+
+usage() {
+    echo "Usage: $0 <output_dir> <parts...>"
+    echo ""
+    echo "Examples:"
+    echo "  $0 ./data test           # Download test set only (16 GB)"
+    echo "  $0 ./data part1 part2    # Download training parts 1 and 2"
+    echo "  $0 ./data all            # Download everything (~1.7 TB)"
+    echo ""
+    echo "Available parts: test, part1, part2, part3, part4, part5, extra, all"
+    exit 1
+}
+
+download_part() {
+    local output_dir="$1"
+    local part="$2"
+    local filename="${PART_FILES[$part]}"
+    local url="${BASE_URL}/${filename}"
+    local output_file="${output_dir}/${filename}"
+
+    echo "----------------------------------------"
+    echo "Downloading: ${part} (${filename})"
+    echo "URL: ${url}"
+    echo "Output: ${output_file}"
+    echo "----------------------------------------"
+
+    # Download with curl, showing progress
+    curl -L --progress-bar "${url}" -o "${output_file}"
+
+    # Unzip
+    echo "Extracting ${filename}..."
+    unzip -q "${output_file}" -d "${output_dir}"
+
+    # Optionally remove zip file to save space
+    # Uncomment the next line if you want to delete zips after extraction
+    # rm "${output_file}"
+
+    echo "Done: ${part}"
+    echo ""
+}
+
+# Check arguments
+if [ $# -lt 2 ]; then
+    usage
+fi
+
+OUTPUT_DIR="$1"
+shift
+
+# Create output directory
+mkdir -p "${OUTPUT_DIR}"
+
+# Determine which parts to download
+PARTS_TO_DOWNLOAD=()
+
+for arg in "$@"; do
+    if [ "$arg" == "all" ]; then
+        PARTS_TO_DOWNLOAD=("${ALL_PARTS[@]}")
+        break
+    elif [ -n "${PART_FILES[$arg]}" ]; then
+        PARTS_TO_DOWNLOAD+=("$arg")
+    else
+        echo "Error: Unknown part '${arg}'"
+        echo "Available parts: test, part1, part2, part3, part4, part5, extra, all"
+        exit 1
+    fi
+done
+
+if [ ${#PARTS_TO_DOWNLOAD[@]} -eq 0 ]; then
+    echo "Error: No valid parts specified"
+    usage
+fi
+
+echo "========================================"
+echo "EgoDex Dataset Download"
+echo "========================================"
+echo "Output directory: ${OUTPUT_DIR}"
+echo "Parts to download: ${PARTS_TO_DOWNLOAD[*]}"
+echo "========================================"
+echo ""
+
+# Download each part
+for part in "${PARTS_TO_DOWNLOAD[@]}"; do
+    download_part "${OUTPUT_DIR}" "${part}"
+done
+
+echo "========================================"
+echo "Download complete!"
+echo "Data saved to: ${OUTPUT_DIR}"
+echo "========================================"
+

examples/dataset/slurm_port_egodex.py

@@ -0,0 +1,443 @@
+#!/usr/bin/env python
+
+# Copyright 2025 The HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""
+Distributed EgoDex dataset porting using SLURM and datatrove.
+
+EgoDex is a large-scale dataset for egocentric dexterous manipulation collected
+with ARKit on Apple Vision Pro. This script converts EgoDex data to LeRobot format.
+
+Reference: https://arxiv.org/abs/2505.11709, https://github.com/apple/ml-egodex 
+"""
+
+import argparse
+from pathlib import Path
+
+import cv2
+import h5py
+import mediapy as mpy
+import numpy as np
+from datatrove.executor import LocalPipelineExecutor
+from datatrove.executor.slurm import SlurmPipelineExecutor
+from datatrove.pipeline.base import PipelineStep
+
+from lerobot.datasets.lerobot_dataset import LeRobotDataset
+
+# Image dimensions
+DEFAULT_IMAGE_HEIGHT = 1080
+DEFAULT_IMAGE_WIDTH = 1920
+
+class PortEgoDexShards(PipelineStep):
+    def __init__(
+        self,
+        raw_dir: Path | str,
+        repo_id: str,
+        local_dir: Path | str = None,
+        percentage: float = 100.0,
+    ):
+        super().__init__()
+        self.raw_dir = Path(raw_dir)
+        self.repo_id = repo_id
+        self.local_dir = Path(local_dir) if local_dir else Path("data/local_datasets")
+        self.percentage = percentage
+
+    def run(self, data=None, rank: int = 0, world_size: int = 1):
+        from pathlib import Path
+
+        import cv2
+        import h5py
+        import mediapy as mpy
+        import numpy as np
+
+        from lerobot.datasets.lerobot_dataset import LeRobotDataset
+        from lerobot.utils.utils import init_logging
+
+        def _get_state_for_single_frame(transforms_group, frame_idx):
+            """
+            Construct 48D hand state representation from EgoDex.
+            
+            State vector composition (per hand = 24D, total = 48D):
+            - Wrist 3D position (3)
+            - Wrist orientation in 6D representation (6)
+            - 5 fingertip 3D positions (15)
+            """
+            state_vector = []
+            fingertip_joints = {
+                "left": [
+                    "leftThumbTip",
+                    "leftIndexFingerTip",
+                    "leftMiddleFingerTip",
+                    "leftRingFingerTip",
+                    "leftLittleFingerTip",
+                ],
+                "right": [
+                    "rightThumbTip",
+                    "rightIndexFingerTip",
+                    "rightMiddleFingerTip",
+                    "rightRingFingerTip",
+                    "rightLittleFingerTip",
+                ],
+            }
+
+            for hand_side in ["left", "right"]:
+                hand_key = f"{hand_side}Hand"
+                hand_transform = transforms_group[hand_key][frame_idx]
+
+                # 1. Wrist 3D position
+                hand_position = hand_transform[:3, 3]
+                state_vector.extend(hand_position)
+
+                # 2. Wrist orientation in compact 6D representation
+                rotation_matrix = hand_transform[:3, :3]
+                rotation_6d = np.concatenate([rotation_matrix[:, 0], rotation_matrix[:, 1]])
+                state_vector.extend(rotation_6d)
+
+                # 3. 3D positions of 5 fingertips
+                for fingertip in fingertip_joints[hand_side]:
+                    fingertip_transform = transforms_group[fingertip][frame_idx]
+                    fingertip_pos = fingertip_transform[:3, 3]
+                    state_vector.extend(fingertip_pos)
+
+            # Also return camera extrinsics for optional coordinate frame transformations
+            return np.array(state_vector, dtype=np.float32), transforms_group["camera"][frame_idx]
+
+        def get_state_and_action_from_egodex_annotations(demo):
+            """
+            Convert EgoDex demo annotations into states and actions.
+            
+            The "action" is the state at time t+1 (next-pose prediction).
+            """
+            transforms_group = demo["transforms"]
+            total_frames = list(transforms_group.values())[0].shape[0]
+
+            states_list, extrinsics_list = [], []
+            for frame_idx in range(total_frames):
+                state_vector, extrinsics = _get_state_for_single_frame(transforms_group, frame_idx)
+                states_list.append(state_vector)
+                extrinsics_list.append(extrinsics.flatten())  # Flatten 4x4 to 16D
+
+            state = np.array(states_list, dtype=np.float32)
+            extrinsics = np.array(extrinsics_list, dtype=np.float32)
+
+            # Shift by 1 timestep to convert state to action
+            action = np.roll(state, -1, axis=0)
+
+            return state, action, extrinsics
+
+        def process_demo(hdf5_file_path, video_path):
+            """Process a single EgoDex demo and return frames for LeRobot."""
+            video = mpy.read_video(str(video_path))
+            video = np.asarray(video)
+            num_frames = video.shape[0]
+            frames = []
+
+            with h5py.File(hdf5_file_path, "r") as demo:
+                state, action, extrinsics = get_state_and_action_from_egodex_annotations(demo)
+
+                # Get natural language task description
+                if demo.attrs.get("llm_type") == "reversible":
+                    direction = demo.attrs.get("which_llm_description", "1")
+                    lang_instruction = demo.attrs.get(
+                        "llm_description" if direction == "1" else "llm_description2",
+                        "manipulation task",
+                    )
+                else:
+                    lang_instruction = demo.attrs.get("llm_description", "manipulation task")
+
+                for step_idx in range(num_frames):
+                    # Resize image to default dimensions
+                    image_resized = cv2.resize(
+                        video[step_idx],
+                        (DEFAULT_IMAGE_WIDTH, DEFAULT_IMAGE_HEIGHT),
+                        interpolation=cv2.INTER_AREA,
+                    )
+                    frame = {
+                        "task": lang_instruction,
+                        "observation.image": image_resized,
+                        "observation.state": state[step_idx],
+                        "observation.extrinsics": extrinsics[step_idx],
+                        "action": action[step_idx],
+                    }
+                    frames.append(frame)
+
+            return frames
+
+        init_logging()
+
+        # Define EgoDex features
+        EGODEX_FEATURES = {
+            "observation.image": {
+                "dtype": "video",
+                "shape": (DEFAULT_IMAGE_HEIGHT, DEFAULT_IMAGE_WIDTH, 3),
+                "names": ["height", "width", "rgb"],
+            },
+            "observation.state": {
+                "dtype": "float32",
+                "shape": (48,),
+                "names": [
+                    # Left hand wrist position (3)
+                    "left_wrist_x",
+                    "left_wrist_y",
+                    "left_wrist_z",
+                    # Left hand wrist rotation 6D (6)
+                    "left_rot_0",
+                    "left_rot_1",
+                    "left_rot_2",
+                    "left_rot_3",
+                    "left_rot_4",
+                    "left_rot_5",
+                    # Left fingertips (15)
+                    "left_thumb_x",
+                    "left_thumb_y",
+                    "left_thumb_z",
+                    "left_index_x",
+                    "left_index_y",
+                    "left_index_z",
+                    "left_middle_x",
+                    "left_middle_y",
+                    "left_middle_z",
+                    "left_ring_x",
+                    "left_ring_y",
+                    "left_ring_z",
+                    "left_little_x",
+                    "left_little_y",
+                    "left_little_z",
+                    # Right hand wrist position (3)
+                    "right_wrist_x",
+                    "right_wrist_y",
+                    "right_wrist_z",
+                    # Right hand wrist rotation 6D (6)
+                    "right_rot_0",
+                    "right_rot_1",
+                    "right_rot_2",
+                    "right_rot_3",
+                    "right_rot_4",
+                    "right_rot_5",
+                    # Right fingertips (15)
+                    "right_thumb_x",
+                    "right_thumb_y",
+                    "right_thumb_z",
+                    "right_index_x",
+                    "right_index_y",
+                    "right_index_z",
+                    "right_middle_x",
+                    "right_middle_y",
+                    "right_middle_z",
+                    "right_ring_x",
+                    "right_ring_y",
+                    "right_ring_z",
+                    "right_little_x",
+                    "right_little_y",
+                    "right_little_z",
+                ],
+            },
+            "observation.extrinsics": {
+                "dtype": "float32",
+                "shape": (16,),
+                "names": [f"extrinsic_{i}" for i in range(16)],
+            },
+            "action": {
+                "dtype": "float32",
+                "shape": (48,),
+                "names": [f"action_{i}" for i in range(48)],
+            },
+        }
+
+        # 1. Discover all HDF5 files
+        files = sorted(list(self.raw_dir.rglob("*.hdf5")))
+        if not files:
+            print(f"No HDF5 files found in {self.raw_dir}")
+            return
+
+        # 2. Apply percentage filter
+        if self.percentage < 100:
+            num_files = max(1, int(len(files) * self.percentage / 100))
+            files = files[:num_files]
+            print(f"Processing {self.percentage}% of dataset: {num_files} files")
+
+        # 3. Assign files to this worker
+        my_files = files[rank::world_size]
+        if not my_files:
+            print(f"Rank {rank} has no files to process.")
+            return
+
+        print(f"Rank {rank} processing {len(my_files)} files out of {len(files)} total.")
+
+        # 4. Create a LeRobot dataset for this shard
+        shard_repo_id = f"{self.repo_id}_world_{world_size}_rank_{rank}"
+        shard_root = self.local_dir / shard_repo_id if self.local_dir else None
+
+        dataset = LeRobotDataset.create(
+            repo_id=shard_repo_id,
+            fps=30,
+            robot_type="hand",
+            features=EGODEX_FEATURES,
+            root=shard_root,
+        )
+
+        # 5. Process each file
+        for input_h5 in my_files:
+            try:
+                # Derive corresponding video path
+                video_path = input_h5.with_suffix(".mp4")
+                if not video_path.exists():
+                    print(f"Warning: Video file not found for {input_h5}, skipping.")
+                    continue
+
+                # Process demo and add frames
+                frames = process_demo(input_h5, video_path)
+                for frame in frames:
+                    dataset.add_frame(frame)
+                dataset.save_episode()
+
+                # Clean up to avoid OOM
+                del frames
+
+            except Exception as e:
+                print(f"Error processing {input_h5}: {e}")
+                continue
+
+        # 6. Finalize the dataset
+        dataset.finalize()
+
+
+def make_port_executor(
+    raw_dir,
+    repo_id,
+    job_name,
+    logs_dir,
+    workers,
+    partition,
+    cpus_per_task,
+    mem_per_cpu,
+    local_dir,
+    percentage,
+    slurm=True,
+):
+    kwargs = {
+        "pipeline": [
+            PortEgoDexShards(raw_dir, repo_id, local_dir, percentage),
+        ],
+        "logging_dir": str(logs_dir / job_name),
+    }
+
+    if slurm:
+        kwargs.update(
+            {
+                "job_name": job_name,
+                "tasks": workers,
+                "workers": workers,
+                "time": "10:00:00",  # EgoDex is large, allow more time
+                "partition": partition,
+                "cpus_per_task": cpus_per_task,
+                "sbatch_args": {"mem-per-cpu": mem_per_cpu},
+            }
+        )
+        executor = SlurmPipelineExecutor(**kwargs)
+    else:
+        kwargs.update(
+            {
+                "tasks": workers,
+                "workers": 1,  # Run locally sequentially for debugging
+            }
+        )
+        executor = LocalPipelineExecutor(**kwargs)
+
+    return executor
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Convert EgoDex dataset to LeRobot format using SLURM."
+    )
+
+    parser.add_argument(
+        "--raw-dir",
+        type=Path,
+        required=True,
+        help="Directory containing input EgoDex data (HDF5 + MP4 files).",
+    )
+    parser.add_argument(
+        "--repo-id",
+        type=str,
+        required=True,
+        help="Repository identifier (e.g., user/egodex-lerobot).",
+    )
+    parser.add_argument(
+        "--logs-dir",
+        type=Path,
+        default=Path("logs"),
+        help="Path to logs directory.",
+    )
+    parser.add_argument(
+        "--job-name",
+        type=str,
+        default="port_egodex",
+        help="Job name used in SLURM.",
+    )
+    parser.add_argument(
+        "--slurm",
+        type=int,
+        default=1,
+        help="Launch over SLURM. Use --slurm 0 to launch sequentially (useful for debugging).",
+    )
+    parser.add_argument(
+        "--workers",
+        type=int,
+        default=50,
+        help="Number of SLURM workers.",
+    )
+    parser.add_argument(
+        "--partition",
+        type=str,
+        help="SLURM partition.",
+    )
+    parser.add_argument(
+        "--cpus-per-task",
+        type=int,
+        default=4,
+        help="Number of CPUs per worker.",
+    )
+    parser.add_argument(
+        "--mem-per-cpu",
+        type=str,
+        default="4G",
+        help="Memory per CPU.",
+    )
+    parser.add_argument(
+        "--percentage",
+        type=float,
+        default=100.0,
+        help="Percentage of dataset to process (e.g., 1.0 for 1%%). Useful for testing.",
+    )
+    parser.add_argument(
+        "--local-dir",
+        type=Path,
+        default=None,
+        help="Local directory to save the LeRobot dataset. Defaults to data/local_datasets.",
+    )
+
+    args = parser.parse_args()
+    kwargs = vars(args)
+    kwargs["slurm"] = kwargs.pop("slurm") == 1
+
+    port_executor = make_port_executor(**kwargs)
+    port_executor.run()
+
+
+if __name__ == "__main__":
+    main()
+

examples/lekiwi/record.py

@@ -21,7 +21,7 @@ from lerobot.robots.lekiwi.config_lekiwi import LeKiwiClientConfig
 from lerobot.robots.lekiwi.lekiwi_client import LeKiwiClient
 from lerobot.scripts.lerobot_record import record_loop
 from lerobot.teleoperators.keyboard import KeyboardTeleop, KeyboardTeleopConfig
-from lerobot.teleoperators.so_leader import SO100Leader, SO100LeaderConfig
+from lerobot.teleoperators.so100_leader import SO100Leader, SO100LeaderConfig
 from lerobot.utils.constants import ACTION, OBS_STR
 from lerobot.utils.control_utils import init_keyboard_listener
 from lerobot.utils.utils import log_say

examples/lekiwi/teleoperate.py

@@ -18,7 +18,7 @@ import time
 
 from lerobot.robots.lekiwi import LeKiwiClient, LeKiwiClientConfig
 from lerobot.teleoperators.keyboard.teleop_keyboard import KeyboardTeleop, KeyboardTeleopConfig
-from lerobot.teleoperators.so_leader import SO100Leader, SO100LeaderConfig
+from lerobot.teleoperators.so100_leader import SO100Leader, SO100LeaderConfig
 from lerobot.utils.robot_utils import precise_sleep
 from lerobot.utils.visualization_utils import init_rerun, log_rerun_data
 

examples/phone_to_so100/evaluate.py

@@ -34,11 +34,12 @@ from lerobot.processor.converters import (
     transition_to_observation,
     transition_to_robot_action,
 )
-from lerobot.robots.so_follower import SO100Follower, SO100FollowerConfig
-from lerobot.robots.so_follower.robot_kinematic_processor import (
+from lerobot.robots.so100_follower.config_so100_follower import SO100FollowerConfig
+from lerobot.robots.so100_follower.robot_kinematic_processor import (
     ForwardKinematicsJointsToEE,
     InverseKinematicsEEToJoints,
 )
+from lerobot.robots.so100_follower.so100_follower import SO100Follower
 from lerobot.scripts.lerobot_record import record_loop
 from lerobot.utils.control_utils import init_keyboard_listener
 from lerobot.utils.utils import log_say

examples/phone_to_so100/record.py

@@ -26,14 +26,15 @@ from lerobot.processor.converters import (
     transition_to_observation,
     transition_to_robot_action,
 )
-from lerobot.robots.so_follower import SO100Follower, SO100FollowerConfig
-from lerobot.robots.so_follower.robot_kinematic_processor import (
+from lerobot.robots.so100_follower.config_so100_follower import SO100FollowerConfig
+from lerobot.robots.so100_follower.robot_kinematic_processor import (
     EEBoundsAndSafety,
     EEReferenceAndDelta,
     ForwardKinematicsJointsToEE,
     GripperVelocityToJoint,
     InverseKinematicsEEToJoints,
 )
+from lerobot.robots.so100_follower.so100_follower import SO100Follower
 from lerobot.scripts.lerobot_record import record_loop
 from lerobot.teleoperators.phone.config_phone import PhoneConfig, PhoneOS
 from lerobot.teleoperators.phone.phone_processor import MapPhoneActionToRobotAction

examples/phone_to_so100/replay.py

@@ -23,10 +23,11 @@ from lerobot.processor.converters import (
     robot_action_observation_to_transition,
     transition_to_robot_action,
 )
-from lerobot.robots.so_follower import SO100Follower, SO100FollowerConfig
-from lerobot.robots.so_follower.robot_kinematic_processor import (
+from lerobot.robots.so100_follower.config_so100_follower import SO100FollowerConfig
+from lerobot.robots.so100_follower.robot_kinematic_processor import (
     InverseKinematicsEEToJoints,
 )
+from lerobot.robots.so100_follower.so100_follower import SO100Follower
 from lerobot.utils.constants import ACTION
 from lerobot.utils.robot_utils import precise_sleep
 from lerobot.utils.utils import log_say
@@ -95,7 +96,7 @@ def main():
         # Send action to robot
         _ = robot.send_action(joint_action)
 
-        precise_sleep(max(1.0 / dataset.fps - (time.perf_counter() - t0), 0.0))
+        precise_sleep(1.0 / dataset.fps - (time.perf_counter() - t0))
 
     # Clean up
     robot.disconnect()

examples/phone_to_so100/teleoperate.py

@@ -21,13 +21,14 @@ from lerobot.processor.converters import (
     robot_action_observation_to_transition,
     transition_to_robot_action,
 )
-from lerobot.robots.so_follower import SO100Follower, SO100FollowerConfig
-from lerobot.robots.so_follower.robot_kinematic_processor import (
+from lerobot.robots.so100_follower.config_so100_follower import SO100FollowerConfig
+from lerobot.robots.so100_follower.robot_kinematic_processor import (
     EEBoundsAndSafety,
     EEReferenceAndDelta,
     GripperVelocityToJoint,
     InverseKinematicsEEToJoints,
 )
+from lerobot.robots.so100_follower.so100_follower import SO100Follower
 from lerobot.teleoperators.phone.config_phone import PhoneConfig, PhoneOS
 from lerobot.teleoperators.phone.phone_processor import MapPhoneActionToRobotAction
 from lerobot.teleoperators.phone.teleop_phone import Phone

examples/rtc/eval_with_real_robot.py

@@ -94,9 +94,9 @@ from lerobot.rl.process import ProcessSignalHandler
 from lerobot.robots import (  # noqa: F401
     Robot,
     RobotConfig,
-    bi_so_follower,
     koch_follower,
-    so_follower,
+    so100_follower,
+    so101_follower,
 )
 from lerobot.robots.utils import make_robot_from_config
 from lerobot.utils.constants import OBS_IMAGES
@@ -455,18 +455,7 @@ def demo_cli(cfg: RTCDemoConfig):
     if cfg.policy.type == "pi05" or cfg.policy.type == "pi0":
         config.compile_model = cfg.use_torch_compile
 
-    if config.use_peft:
-        from peft import PeftConfig, PeftModel
-
-        peft_pretrained_path = cfg.policy.pretrained_path
-        peft_config = PeftConfig.from_pretrained(peft_pretrained_path)
-
-        policy = policy_class.from_pretrained(
-            pretrained_name_or_path=peft_config.base_model_name_or_path, config=config
-        )
-        policy = PeftModel.from_pretrained(policy, peft_pretrained_path, config=peft_config)
-    else:
-        policy = policy_class.from_pretrained(cfg.policy.pretrained_path, config=config)
+    policy = policy_class.from_pretrained(cfg.policy.pretrained_path, config=config)
 
     # Turn on RTC
     policy.config.rtc_config = cfg.rtc

examples/so100_to_so100_EE/evaluate.py

@@ -34,11 +34,12 @@ from lerobot.processor.converters import (
     transition_to_observation,
     transition_to_robot_action,
 )
-from lerobot.robots.so_follower import SO100Follower, SO100FollowerConfig
-from lerobot.robots.so_follower.robot_kinematic_processor import (
+from lerobot.robots.so100_follower.config_so100_follower import SO100FollowerConfig
+from lerobot.robots.so100_follower.robot_kinematic_processor import (
     ForwardKinematicsJointsToEE,
     InverseKinematicsEEToJoints,
 )
+from lerobot.robots.so100_follower.so100_follower import SO100Follower
 from lerobot.scripts.lerobot_record import record_loop
 from lerobot.utils.control_utils import init_keyboard_listener
 from lerobot.utils.utils import log_say

examples/so100_to_so100_EE/record.py

@@ -27,14 +27,16 @@ from lerobot.processor.converters import (
     transition_to_observation,
     transition_to_robot_action,
 )
-from lerobot.robots.so_follower import SO100Follower, SO100FollowerConfig
-from lerobot.robots.so_follower.robot_kinematic_processor import (
+from lerobot.robots.so100_follower.config_so100_follower import SO100FollowerConfig
+from lerobot.robots.so100_follower.robot_kinematic_processor import (
     EEBoundsAndSafety,
     ForwardKinematicsJointsToEE,
     InverseKinematicsEEToJoints,
 )
+from lerobot.robots.so100_follower.so100_follower import SO100Follower
 from lerobot.scripts.lerobot_record import record_loop
-from lerobot.teleoperators.so_leader import SO100Leader, SO100LeaderConfig
+from lerobot.teleoperators.so100_leader.config_so100_leader import SO100LeaderConfig
+from lerobot.teleoperators.so100_leader.so100_leader import SO100Leader
 from lerobot.utils.control_utils import init_keyboard_listener
 from lerobot.utils.utils import log_say
 from lerobot.utils.visualization_utils import init_rerun

examples/so100_to_so100_EE/replay.py

@@ -24,10 +24,11 @@ from lerobot.processor.converters import (
     robot_action_observation_to_transition,
     transition_to_robot_action,
 )
-from lerobot.robots.so_follower import SO100Follower, SO100FollowerConfig
-from lerobot.robots.so_follower.robot_kinematic_processor import (
+from lerobot.robots.so100_follower.config_so100_follower import SO100FollowerConfig
+from lerobot.robots.so100_follower.robot_kinematic_processor import (
     InverseKinematicsEEToJoints,
 )
+from lerobot.robots.so100_follower.so100_follower import SO100Follower
 from lerobot.utils.constants import ACTION
 from lerobot.utils.robot_utils import precise_sleep
 from lerobot.utils.utils import log_say
@@ -96,7 +97,7 @@ def main():
         # Send action to robot
         _ = robot.send_action(joint_action)
 
-        precise_sleep(max(1.0 / dataset.fps - (time.perf_counter() - t0), 0.0))
+        precise_sleep(1.0 / dataset.fps - (time.perf_counter() - t0))
 
     # Clean up
     robot.disconnect()

examples/so100_to_so100_EE/teleoperate.py

@@ -23,13 +23,15 @@ from lerobot.processor.converters import (
     robot_action_to_transition,
     transition_to_robot_action,
 )
-from lerobot.robots.so_follower import SO100Follower, SO100FollowerConfig
-from lerobot.robots.so_follower.robot_kinematic_processor import (
+from lerobot.robots.so100_follower.config_so100_follower import SO100FollowerConfig
+from lerobot.robots.so100_follower.robot_kinematic_processor import (
     EEBoundsAndSafety,
     ForwardKinematicsJointsToEE,
     InverseKinematicsEEToJoints,
 )
-from lerobot.teleoperators.so_leader import SO100Leader, SO100LeaderConfig
+from lerobot.robots.so100_follower.so100_follower import SO100Follower
+from lerobot.teleoperators.so100_leader.config_so100_leader import SO100LeaderConfig
+from lerobot.teleoperators.so100_leader.so100_leader import SO100Leader
 from lerobot.utils.robot_utils import precise_sleep
 from lerobot.utils.visualization_utils import init_rerun, log_rerun_data
 

examples/tutorial/act/act_using_example.py

@@ -5,7 +5,8 @@ from lerobot.datasets.lerobot_dataset import LeRobotDatasetMetadata
 from lerobot.policies.act.modeling_act import ACTPolicy
 from lerobot.policies.factory import make_pre_post_processors
 from lerobot.policies.utils import build_inference_frame, make_robot_action
-from lerobot.robots.so_follower import SO100Follower, SO100FollowerConfig
+from lerobot.robots.so100_follower.config_so100_follower import SO100FollowerConfig
+from lerobot.robots.so100_follower.so100_follower import SO100Follower
 
 MAX_EPISODES = 5
 MAX_STEPS_PER_EPISODE = 20

examples/tutorial/async-inf/robot_client.py

@@ -4,7 +4,7 @@ from lerobot.async_inference.configs import RobotClientConfig
 from lerobot.async_inference.helpers import visualize_action_queue_size
 from lerobot.async_inference.robot_client import RobotClient
 from lerobot.cameras.opencv.configuration_opencv import OpenCVCameraConfig
-from lerobot.robots.so_follower import SO100FollowerConfig
+from lerobot.robots.so100_follower import SO100FollowerConfig
 
 
 def main():
@@ -30,7 +30,6 @@ def main():
         robot=robot_cfg,
         server_address=server_address,
         policy_device="mps",
-        client_device="cpu",
         policy_type="act",
         pretrained_name_or_path="<user>/robot_learning_tutorial_act",
         chunk_size_threshold=0.5,  # g

examples/tutorial/diffusion/diffusion_using_example.py

@@ -5,7 +5,8 @@ from lerobot.datasets.lerobot_dataset import LeRobotDatasetMetadata
 from lerobot.policies.diffusion.modeling_diffusion import DiffusionPolicy
 from lerobot.policies.factory import make_pre_post_processors
 from lerobot.policies.utils import build_inference_frame, make_robot_action
-from lerobot.robots.so_follower import SO100Follower, SO100FollowerConfig
+from lerobot.robots.so100_follower.config_so100_follower import SO100FollowerConfig
+from lerobot.robots.so100_follower.so100_follower import SO100Follower
 
 MAX_EPISODES = 5
 MAX_STEPS_PER_EPISODE = 20

examples/tutorial/pi0/using_pi0_example.py

@@ -5,7 +5,8 @@ from lerobot.datasets.utils import hw_to_dataset_features
 from lerobot.policies.factory import make_pre_post_processors
 from lerobot.policies.pi0.modeling_pi0 import PI0Policy
 from lerobot.policies.utils import build_inference_frame, make_robot_action
-from lerobot.robots.so_follower import SO100Follower, SO100FollowerConfig
+from lerobot.robots.so100_follower.config_so100_follower import SO100FollowerConfig
+from lerobot.robots.so100_follower.so100_follower import SO100Follower
 
 MAX_EPISODES = 5
 MAX_STEPS_PER_EPISODE = 20

examples/tutorial/rl/hilserl_example.py

@@ -14,8 +14,8 @@ from lerobot.policies.sac.modeling_sac import SACPolicy
 from lerobot.policies.sac.reward_model.modeling_classifier import Classifier
 from lerobot.rl.buffer import ReplayBuffer
 from lerobot.rl.gym_manipulator import make_robot_env
-from lerobot.robots.so_follower import SO100FollowerConfig
-from lerobot.teleoperators.so_leader import SO100LeaderConfig
+from lerobot.robots.so100_follower import SO100FollowerConfig
+from lerobot.teleoperators.so100_leader import SO100LeaderConfig
 from lerobot.teleoperators.utils import TeleopEvents
 
 LOG_EVERY = 10

examples/tutorial/smolvla/using_smolvla_example.py

@@ -5,7 +5,8 @@ from lerobot.datasets.utils import hw_to_dataset_features
 from lerobot.policies.factory import make_pre_post_processors
 from lerobot.policies.smolvla.modeling_smolvla import SmolVLAPolicy
 from lerobot.policies.utils import build_inference_frame, make_robot_action
-from lerobot.robots.so_follower import SO100Follower, SO100FollowerConfig
+from lerobot.robots.so100_follower.config_so100_follower import SO100FollowerConfig
+from lerobot.robots.so100_follower.so100_follower import SO100Follower
 
 MAX_EPISODES = 5
 MAX_STEPS_PER_EPISODE = 20

examples/unitree_g1/gr00t_locomotion.py

@@ -13,9 +13,16 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+"""
+Example: GR00T Locomotion with Pre-loaded Policies
+
+This example demonstrates the NEW pattern for loading GR00T policies externally
+and passing them to the robot class.
+"""
 
 import argparse
 import logging
+import threading
 import time
 from collections import deque
 
@@ -24,26 +31,24 @@ import onnxruntime as ort
 from huggingface_hub import hf_hub_download
 
 from lerobot.robots.unitree_g1.config_unitree_g1 import UnitreeG1Config
-from lerobot.robots.unitree_g1.g1_utils import G1_29_JointIndex
 from lerobot.robots.unitree_g1.unitree_g1 import UnitreeG1
 
-logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 
-
 GROOT_DEFAULT_ANGLES = np.zeros(29, dtype=np.float32)
-GROOT_DEFAULT_ANGLES[[0, 6]] = -0.1  # Hip pitch
-GROOT_DEFAULT_ANGLES[[3, 9]] = 0.3  # Knee
-GROOT_DEFAULT_ANGLES[[4, 10]] = -0.2  # Ankle pitch
+GROOT_DEFAULT_ANGLES[[0, 6]] = -0.1  # hip pitch
+GROOT_DEFAULT_ANGLES[[3, 9]] = 0.3  # knee
+GROOT_DEFAULT_ANGLES[[4, 10]] = -0.2  # ankle pitch
 
 MISSING_JOINTS = []
-G1_MODEL = "g1_23"  # Or "g1_29"
+G1_MODEL = "g1_23"  # or "g1_29"
 if G1_MODEL == "g1_23":
-    MISSING_JOINTS = [12, 14, 20, 21, 27, 28]  # Waist yaw/pitch, wrist pitch/yaw
+    MISSING_JOINTS = [12, 14, 20, 21, 27, 28]  # waist yaw/pitch, wrist pitch/yaw
+
+LOCOMOTION_ACTION_SCALE = 0.25
+
+LOCOMOTION_CONTROL_DT = 0.02
 
-# Control parameters
-ACTION_SCALE = 0.25
-CONTROL_DT = 0.02  # 50Hz
 ANG_VEL_SCALE: float = 0.25
 DOF_POS_SCALE: float = 1.0
 DOF_VEL_SCALE: float = 0.05
@@ -56,12 +61,12 @@ DEFAULT_GROOT_REPO_ID = "nepyope/GR00T-WholeBodyControl_g1"
 def load_groot_policies(
     repo_id: str = DEFAULT_GROOT_REPO_ID,
 ) -> tuple[ort.InferenceSession, ort.InferenceSession]:
-    """Load GR00T dual-policy system (Balance + Walk) from the hub.
+    """Load GR00T dual-policy system (Balance + Walk) from Hugging Face Hub.
 
     Args:
         repo_id: Hugging Face Hub repository ID containing the ONNX policies.
     """
-    logger.info(f"Loading GR00T dual-policy system from the hub ({repo_id})...")
+    logger.info(f"Loading GR00T dual-policy system from Hugging Face Hub ({repo_id})...")
 
     # Download ONNX policies from Hugging Face Hub
     balance_path = hf_hub_download(
@@ -83,7 +88,15 @@ def load_groot_policies(
 
 
 class GrootLocomotionController:
-    """GR00T lower-body locomotion controller for the Unitree G1."""
+    """
+    Handles GR00T-style locomotion control for the Unitree G1 robot.
+
+    This controller manages:
+    - Dual-policy system (Balance + Walk)
+    - 29-joint observation processing
+    - 15D action output (legs + waist)
+    - Policy inference and motor command generation
+    """
 
     def __init__(self, policy_balance, policy_walk, robot, config):
         self.policy_balance = policy_balance
@@ -91,9 +104,9 @@ class GrootLocomotionController:
         self.robot = robot
         self.config = config
 
-        self.cmd = np.array([0.0, 0.0, 0.0], dtype=np.float32)  # vx, vy, theta_dot
+        self.locomotion_cmd = np.array([0.0, 0.0, 0.0], dtype=np.float32)  # vx, vy, theta_dot
 
-        # Robot state
+        # GR00T-specific state
         self.groot_qj_all = np.zeros(29, dtype=np.float32)
         self.groot_dqj_all = np.zeros(29, dtype=np.float32)
         self.groot_action = np.zeros(15, dtype=np.float32)
@@ -103,39 +116,47 @@ class GrootLocomotionController:
         self.groot_height_cmd = 0.74  # Default base height
         self.groot_orientation_cmd = np.array([0.0, 0.0, 0.0], dtype=np.float32)
 
-        # Input to GR00T is 6 frames (6*86D=516)
+        # input to gr00t is 6 frames (6*86D=516)
         for _ in range(6):
             self.groot_obs_history.append(np.zeros(86, dtype=np.float32))
 
+        # Thread management
+        self.locomotion_running = False
+        self.locomotion_thread = None
+
         logger.info("GrootLocomotionController initialized")
 
-    def run_step(self):
-        # Get current observation
-        obs = self.robot.get_observation()
+    def groot_locomotion_run(self):
+        # get current observation
+        robot_state = self.robot.get_observation()
 
-        if not obs:
+        if robot_state is None:
             return
 
-        # Get command from remote controller
-        if obs["remote.buttons"][0]:  # R1 - raise waist
-            self.groot_height_cmd += 0.001
-            self.groot_height_cmd = np.clip(self.groot_height_cmd, 0.50, 1.00)
-        if obs["remote.buttons"][4]:  # R2 - lower waist
-            self.groot_height_cmd -= 0.001
-            self.groot_height_cmd = np.clip(self.groot_height_cmd, 0.50, 1.00)
+        # get command from remote controller
+        if robot_state.wireless_remote is not None:
+            self.robot.remote_controller.set(robot_state.wireless_remote)
+            if self.robot.remote_controller.button[0]:  # R1 - raise waist
+                self.groot_height_cmd += 0.001
+                self.groot_height_cmd = np.clip(self.groot_height_cmd, 0.50, 1.00)
+            if self.robot.remote_controller.button[4]:  # R2 - lower waist
+                self.groot_height_cmd -= 0.001
+                self.groot_height_cmd = np.clip(self.groot_height_cmd, 0.50, 1.00)
+        else:
+            self.robot.remote_controller.lx = 0.0
+            self.robot.remote_controller.ly = 0.0
+            self.robot.remote_controller.rx = 0.0
+            self.robot.remote_controller.ry = 0.0
 
-        self.cmd[0] = obs["remote.ly"]  # Forward/backward
-        self.cmd[1] = obs["remote.lx"] * -1  # Left/right
-        self.cmd[2] = obs["remote.rx"] * -1  # Rotation rate
+        self.locomotion_cmd[0] = self.robot.remote_controller.ly  # forward/backward
+        self.locomotion_cmd[1] = self.robot.remote_controller.lx * -1  # left/right
+        self.locomotion_cmd[2] = self.robot.remote_controller.rx * -1  # rotation rate
 
-        # Get joint positions and velocities from flat dict
-        for motor in G1_29_JointIndex:
-            name = motor.name
-            idx = motor.value
-            self.groot_qj_all[idx] = obs[f"{name}.q"]
-            self.groot_dqj_all[idx] = obs[f"{name}.dq"]
+        for i in range(29):
+            self.groot_qj_all[i] = robot_state.motor_state[i].q
+            self.groot_dqj_all[i] = robot_state.motor_state[i].dq
 
-        # Adapt observation for g1_23dof
+        # adapt observation for g1_23dof
         for idx in MISSING_JOINTS:
             self.groot_qj_all[idx] = 0.0
             self.groot_dqj_all[idx] = 0.0
@@ -144,18 +165,18 @@ class GrootLocomotionController:
         qj_obs = self.groot_qj_all.copy()
         dqj_obs = self.groot_dqj_all.copy()
 
-        # Express IMU data in gravity frame of reference
-        quat = [obs["imu.quat.w"], obs["imu.quat.x"], obs["imu.quat.y"], obs["imu.quat.z"]]
-        ang_vel = np.array([obs["imu.gyro.x"], obs["imu.gyro.y"], obs["imu.gyro.z"]], dtype=np.float32)
+        # express imu data in gravity frame of reference
+        quat = robot_state.imu_state.quaternion
+        ang_vel = np.array(robot_state.imu_state.gyroscope, dtype=np.float32)
         gravity_orientation = self.robot.get_gravity_orientation(quat)
 
-        # Scale joint positions and velocities before policy inference
+        # scale joint positions and velocities before policy inference
         qj_obs = (qj_obs - GROOT_DEFAULT_ANGLES) * DOF_POS_SCALE
         dqj_obs = dqj_obs * DOF_VEL_SCALE
         ang_vel_scaled = ang_vel * ANG_VEL_SCALE
 
-        # Build single frame observation
-        self.groot_obs_single[:3] = self.cmd * np.array(CMD_SCALE)
+        # build single frame observation
+        self.groot_obs_single[:3] = self.locomotion_cmd * np.array(CMD_SCALE)
         self.groot_obs_single[3] = self.groot_height_cmd
         self.groot_obs_single[4:7] = self.groot_orientation_cmd
         self.groot_obs_single[7:10] = ang_vel_scaled
@@ -173,76 +194,113 @@ class GrootLocomotionController:
             end_idx = start_idx + 86
             self.groot_obs_stacked[start_idx:end_idx] = obs_frame
 
-        cmd_magnitude = np.linalg.norm(self.cmd)
+        # Run policy inference (ONNX) with 516D stacked observation
+
+        cmd_magnitude = np.linalg.norm(self.locomotion_cmd)
+
         selected_policy = (
             self.policy_balance if cmd_magnitude < 0.05 else self.policy_walk
-        )  # Balance/standing policy for small commands, walking policy for movement commands
+        )  # balance/standing policy for small commands, walking policy for movement commands
 
-        # Run policy inference
+        # run policy inference
         ort_inputs = {selected_policy.get_inputs()[0].name: np.expand_dims(self.groot_obs_stacked, axis=0)}
         ort_outs = selected_policy.run(None, ort_inputs)
         self.groot_action = ort_outs[0].squeeze()
 
-        # Transform action back to target joint positions
-        target_dof_pos_15 = GROOT_DEFAULT_ANGLES[:15] + self.groot_action * ACTION_SCALE
+        # transform action back to target joint positions
+        target_dof_pos_15 = GROOT_DEFAULT_ANGLES[:15] + self.groot_action * LOCOMOTION_ACTION_SCALE
 
-        # Build action dict (only first 15 joints for GR00T)
-        action_dict = {}
+        # command motors
         for i in range(15):
-            motor_name = G1_29_JointIndex(i).name
-            action_dict[f"{motor_name}.q"] = float(target_dof_pos_15[i])
+            motor_idx = i
+            self.robot.msg.motor_cmd[motor_idx].q = target_dof_pos_15[i]
+            self.robot.msg.motor_cmd[motor_idx].qd = 0
+            self.robot.msg.motor_cmd[motor_idx].kp = self.robot.kp[motor_idx]
+            self.robot.msg.motor_cmd[motor_idx].kd = self.robot.kd[motor_idx]
+            self.robot.msg.motor_cmd[motor_idx].tau = 0
 
-        # Zero out missing joints for g1_23dof
+        # adapt action for g1_23dof
         for joint_idx in MISSING_JOINTS:
-            motor_name = G1_29_JointIndex(joint_idx).name
-            action_dict[f"{motor_name}.q"] = 0.0
+            self.robot.msg.motor_cmd[joint_idx].q = 0.0
+            self.robot.msg.motor_cmd[joint_idx].qd = 0
+            self.robot.msg.motor_cmd[joint_idx].kp = self.robot.kp[joint_idx]
+            self.robot.msg.motor_cmd[joint_idx].kd = self.robot.kd[joint_idx]
+            self.robot.msg.motor_cmd[joint_idx].tau = 0
 
-        # Send action to robot
-        self.robot.send_action(action_dict)
+        # send action to robot
+        self.robot.send_action(self.robot.msg)
 
-
-def run(repo_id: str = DEFAULT_GROOT_REPO_ID) -> None:
-    """Main function to run the GR00T locomotion controller.
-
-    Args:
-        repo_id: Hugging Face Hub repository ID for GR00T policies.
-    """
-    # Load policies
-    policy_balance, policy_walk = load_groot_policies(repo_id=repo_id)
-
-    # Initialize robot
-    config = UnitreeG1Config()
-    robot = UnitreeG1(config)
-
-    robot.connect()
-
-    # Initialize gr00T locomotion controller
-    groot_controller = GrootLocomotionController(
-        policy_balance=policy_balance,
-        policy_walk=policy_walk,
-        robot=robot,
-        config=config,
-    )
-
-    try:
-        robot.reset(CONTROL_DT, GROOT_DEFAULT_ANGLES)
-
-        logger.info("Use joystick: LY=fwd/back, LX=left/right, RX=rotate, R1=raise waist, R2=lower waist")
-        logger.info("Press Ctrl+C to stop")
-
-        # Run step
-        while not robot._shutdown_event.is_set():
+    def _locomotion_thread_loop(self):
+        """Background thread that runs the locomotion policy at specified rate."""
+        logger.info("Locomotion thread started")
+        while self.locomotion_running:
             start_time = time.time()
-            groot_controller.run_step()
+            try:
+                self.groot_locomotion_run()
+            except Exception as e:
+                logger.error(f"Error in locomotion loop: {e}")
+
+            # Sleep to maintain control rate
             elapsed = time.time() - start_time
-            sleep_time = max(0, CONTROL_DT - elapsed)
+            sleep_time = max(0, LOCOMOTION_CONTROL_DT - elapsed)
             time.sleep(sleep_time)
-    except KeyboardInterrupt:
-        logger.info("Stopping locomotion...")
-    finally:
-        if robot.is_connected:
-            robot.disconnect()
-        logger.info("Done!")
+        logger.info("Locomotion thread stopped")
+
+    def start_locomotion_thread(self):
+        if self.locomotion_running:
+            logger.warning("Locomotion thread already running")
+            return
+
+        logger.info("Starting locomotion control thread...")
+        self.locomotion_running = True
+        self.locomotion_thread = threading.Thread(target=self._locomotion_thread_loop, daemon=True)
+        self.locomotion_thread.start()
+
+        logger.info("Locomotion control thread started!")
+
+    def stop_locomotion_thread(self):
+        if not self.locomotion_running:
+            return
+
+        logger.info("Stopping locomotion control thread...")
+        self.locomotion_running = False
+        if self.locomotion_thread:
+            self.locomotion_thread.join(timeout=2.0)
+        logger.info("Locomotion control thread stopped")
+
+    def reset_robot(self):
+        """Move robot legs to default standing position over 2 seconds (arms are not moved)."""
+        total_time = 3.0
+        num_step = int(total_time / self.robot.control_dt)
+
+        # Only control legs, not arms (first 12 joints)
+        default_pos = GROOT_DEFAULT_ANGLES  # First 12 values are leg angles
+        dof_size = len(default_pos)
+
+        # Get current lowstate
+        robot_state = self.robot.get_observation()
+
+        # Record the current leg positions
+        init_dof_pos = np.zeros(dof_size, dtype=np.float32)
+        for i in range(dof_size):
+            init_dof_pos[i] = robot_state.motor_state[i].q
+
+        # Move legs to default pos
+        for i in range(num_step):
+            alpha = i / num_step
+            for motor_idx in range(dof_size):
+                target_pos = default_pos[motor_idx]
+                self.robot.msg.motor_cmd[motor_idx].q = (
+                    init_dof_pos[motor_idx] * (1 - alpha) + target_pos * alpha
+                )
+                self.robot.msg.motor_cmd[motor_idx].qd = 0
+                self.robot.msg.motor_cmd[motor_idx].kp = self.robot.kp[motor_idx]
+                self.robot.msg.motor_cmd[motor_idx].kd = self.robot.kd[motor_idx]
+                self.robot.msg.motor_cmd[motor_idx].tau = 0
+            self.robot.msg.crc = self.robot.crc.Crc(self.robot.msg)
+            self.robot.lowcmd_publisher.Write(self.robot.msg)
+            time.sleep(self.robot.control_dt)
+        logger.info("Reached default position (legs only)")
 
 
 if __name__ == "__main__":
@@ -255,4 +313,35 @@ if __name__ == "__main__":
     )
     args = parser.parse_args()
 
-    run(repo_id=args.repo_id)
+    # load policies
+    policy_balance, policy_walk = load_groot_policies(repo_id=args.repo_id)
+
+    # initialize robot
+    config = UnitreeG1Config()
+    robot = UnitreeG1(config)
+
+    # initialize gr00t locomotion controller
+    groot_controller = GrootLocomotionController(
+        policy_balance=policy_balance,
+        policy_walk=policy_walk,
+        robot=robot,
+        config=config,
+    )
+
+    # reset legs and start locomotion thread
+    try:
+        groot_controller.reset_robot()
+        groot_controller.start_locomotion_thread()
+
+        # log status
+        logger.info("Robot initialized with GR00T locomotion policies")
+        logger.info("Locomotion controller running in background thread")
+        logger.info("Press Ctrl+C to stop")
+
+        # keep robot alive
+        while True:
+            time.sleep(1.0)
+    except KeyboardInterrupt:
+        print("\nStopping locomotion...")
+        groot_controller.stop_locomotion_thread()
+        print("Done!")

examples/unitree_g1/holosoma_locomotion.py

@@ -1,264 +0,0 @@
-#!/usr/bin/env python
-
-# Copyright 2025 The HuggingFace Inc. team. All rights reserved.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-import argparse
-import json
-import logging
-import time
-
-import numpy as np
-import onnx
-import onnxruntime as ort
-from huggingface_hub import hf_hub_download
-
-from lerobot.robots.unitree_g1.config_unitree_g1 import UnitreeG1Config
-from lerobot.robots.unitree_g1.g1_utils import G1_29_JointIndex
-from lerobot.robots.unitree_g1.unitree_g1 import UnitreeG1
-
-logging.basicConfig(level=logging.INFO)
-logger = logging.getLogger(__name__)
-
-DEFAULT_ANGLES = np.zeros(29, dtype=np.float32)
-DEFAULT_ANGLES[[0, 6]] = -0.312  # Hip pitch
-DEFAULT_ANGLES[[3, 9]] = 0.669  # Knee
-DEFAULT_ANGLES[[4, 10]] = -0.363  # Ankle pitch
-DEFAULT_ANGLES[[15, 22]] = 0.2  # Shoulder pitch
-DEFAULT_ANGLES[16] = 0.2  # Left shoulder roll
-DEFAULT_ANGLES[23] = -0.2  # Right shoulder roll
-DEFAULT_ANGLES[[18, 25]] = 0.6  # Elbow
-
-MISSING_JOINTS = []
-G1_MODEL = "g1_23"  # Or "g1_29"
-if G1_MODEL == "g1_23":
-    MISSING_JOINTS = [12, 14, 20, 21, 27, 28]  # Waist yaw/pitch, wrist pitch/yaw
-
-# Control parameters
-ACTION_SCALE = 0.25
-CONTROL_DT = 0.02  # 50Hz
-ANG_VEL_SCALE = 0.25
-DOF_POS_SCALE = 1.0
-DOF_VEL_SCALE = 0.05
-GAIT_PERIOD = 1.0
-
-
-DEFAULT_HOLOSOMA_REPO_ID = "nepyope/holosoma_locomotion"
-
-# Policy filename mapping
-POLICY_FILES = {
-    "fastsac": "fastsac_g1_29dof.onnx",
-    "ppo": "ppo_g1_29dof.onnx",
-}
-
-
-def load_policy(
-    repo_id: str = DEFAULT_HOLOSOMA_REPO_ID,
-    policy_type: str = "fastsac",
-) -> tuple[ort.InferenceSession, np.ndarray, np.ndarray]:
-    """Load Holosoma locomotion policy and extract KP/KD from metadata.
-
-    Args:
-        repo_id: Hugging Face Hub repo ID
-        policy_type: Either "fastsac" (default) or "ppo"
-
-    Returns:
-        (policy, kp, kd) tuple
-    """
-    if policy_type not in POLICY_FILES:
-        raise ValueError(f"Unknown policy type: {policy_type}. Choose from: {list(POLICY_FILES.keys())}")
-
-    filename = POLICY_FILES[policy_type]
-    logger.info(f"Loading {policy_type.upper()} policy from: {repo_id}/{filename}")
-    policy_path = hf_hub_download(repo_id=repo_id, filename=filename)
-
-    policy = ort.InferenceSession(policy_path)
-    logger.info(f"Policy loaded: {policy.get_inputs()[0].shape} → {policy.get_outputs()[0].shape}")
-
-    # Extract KP/KD from ONNX metadata
-    model = onnx.load(policy_path)
-    metadata = {prop.key: prop.value for prop in model.metadata_props}
-
-    if "kp" not in metadata or "kd" not in metadata:
-        raise ValueError("ONNX model must contain 'kp' and 'kd' in metadata")
-
-    kp = np.array(json.loads(metadata["kp"]), dtype=np.float32)
-    kd = np.array(json.loads(metadata["kd"]), dtype=np.float32)
-    logger.info(f"Loaded KP/KD from ONNX ({len(kp)} joints)")
-
-    return policy, kp, kd
-
-
-class HolosomaLocomotionController:
-    """Holosoma whole-body locomotion controller for Unitree G1."""
-
-    def __init__(self, policy, robot, kp: np.ndarray, kd: np.ndarray):
-        self.policy = policy
-        self.robot = robot
-
-        # Override robot's PD gains with policy gains
-        self.robot.kp = kp
-        self.robot.kd = kd
-
-        self.cmd = np.zeros(3, dtype=np.float32)
-
-        # Robot state
-        self.qj = np.zeros(29, dtype=np.float32)
-        self.dqj = np.zeros(29, dtype=np.float32)
-        self.obs = np.zeros(100, dtype=np.float32)
-        self.last_action = np.zeros(29, dtype=np.float32)
-
-        # Gait phase
-        self.phase = np.array([[0.0, np.pi]], dtype=np.float32)
-        self.phase_dt = 2 * np.pi / ((1.0 / CONTROL_DT) * GAIT_PERIOD)
-        self.is_standing = True
-
-    def run_step(self):
-        # Get current observation
-        obs = self.robot.get_observation()
-
-        if not obs:
-            return
-
-        # Get command from remote controller
-        ly = obs["remote.ly"] if abs(obs["remote.ly"]) > 0.1 else 0.0
-        lx = obs["remote.lx"] if abs(obs["remote.lx"]) > 0.1 else 0.0
-        rx = obs["remote.rx"] if abs(obs["remote.rx"]) > 0.1 else 0.0
-        self.cmd[:] = [ly, -lx, -rx]
-
-        # Get joint positions and velocities
-        for motor in G1_29_JointIndex:
-            name = motor.name
-            idx = motor.value
-            self.qj[idx] = obs[f"{name}.q"]
-            self.dqj[idx] = obs[f"{name}.dq"]
-
-        # Adapt observation for g1_23dof
-        for idx in MISSING_JOINTS:
-            self.qj[idx] = 0.0
-            self.dqj[idx] = 0.0
-
-        # Express IMU data in gravity frame of reference
-        quat = [obs["imu.quat.w"], obs["imu.quat.x"], obs["imu.quat.y"], obs["imu.quat.z"]]
-        ang_vel = np.array([obs["imu.gyro.x"], obs["imu.gyro.y"], obs["imu.gyro.z"]], dtype=np.float32)
-        gravity = self.robot.get_gravity_orientation(quat)
-
-        # Scale joint positions and velocities before policy inference
-        qj_obs = (self.qj - DEFAULT_ANGLES) * DOF_POS_SCALE
-        dqj_obs = self.dqj * DOF_VEL_SCALE
-        ang_vel_s = ang_vel * ANG_VEL_SCALE
-
-        # Update gait phase
-        if np.linalg.norm(self.cmd[:2]) < 0.01 and abs(self.cmd[2]) < 0.01:
-            self.phase[0, :] = np.pi
-            self.is_standing = True
-        elif self.is_standing:
-            self.phase = np.array([[0.0, np.pi]], dtype=np.float32)
-            self.is_standing = False
-        else:
-            self.phase = np.fmod(self.phase + self.phase_dt + np.pi, 2 * np.pi) - np.pi
-
-        sin_ph = np.sin(self.phase[0])
-        cos_ph = np.cos(self.phase[0])
-
-        # Build observations
-        self.obs[0:29] = self.last_action
-        self.obs[29:32] = ang_vel_s
-        self.obs[32] = self.cmd[2]
-        self.obs[33:35] = self.cmd[:2]
-        self.obs[35:37] = cos_ph
-        self.obs[37:66] = qj_obs
-        self.obs[66:95] = dqj_obs
-        self.obs[95:98] = gravity
-        self.obs[98:100] = sin_ph
-
-        # Run policy inference
-        ort_in = {self.policy.get_inputs()[0].name: self.obs.reshape(1, -1).astype(np.float32)}
-        raw_action = self.policy.run(None, ort_in)[0].squeeze()
-        action = np.clip(raw_action, -100.0, 100.0)
-        self.last_action = action.copy()
-
-        # Transform action back to target joint positions
-        target = DEFAULT_ANGLES + action * ACTION_SCALE
-
-        # Build action dict
-        action_dict = {}
-        for motor in G1_29_JointIndex:
-            action_dict[f"{motor.name}.q"] = float(target[motor.value])
-
-        # Zero out missing joints for g1_23dof
-        for joint_idx in MISSING_JOINTS:
-            motor_name = G1_29_JointIndex(joint_idx).name
-            action_dict[f"{motor_name}.q"] = 0.0
-
-        # Send action to robot
-        self.robot.send_action(action_dict)
-
-
-def run(repo_id: str = DEFAULT_HOLOSOMA_REPO_ID, policy_type: str = "fastsac") -> None:
-    """Main function to run the Holosoma locomotion controller.
-
-    Args:
-        repo_id: Hugging Face Hub repository ID for Holosoma policies.
-        policy_type: Policy type to use ('fastsac' or 'ppo').
-    """
-    # Load policy and gains
-    policy, kp, kd = load_policy(repo_id=repo_id, policy_type=policy_type)
-
-    # Initialize robot
-    config = UnitreeG1Config()
-    robot = UnitreeG1(config)
-    robot.connect()
-
-    holosoma_controller = HolosomaLocomotionController(policy, robot, kp, kd)
-
-    try:
-        robot.reset(CONTROL_DT, DEFAULT_ANGLES)
-
-        logger.info("Use joystick: LY=fwd/back, LX=left/right, RX=rotate")
-        logger.info("Press Ctrl+C to stop")
-
-        # Run step
-        while not robot._shutdown_event.is_set():
-            start_time = time.time()
-            holosoma_controller.run_step()
-            elapsed = time.time() - start_time
-            sleep_time = max(0, CONTROL_DT - elapsed)
-            time.sleep(sleep_time)
-    except KeyboardInterrupt:
-        logger.info("Stopping locomotion...")
-    finally:
-        if robot.is_connected:
-            robot.disconnect()
-        logger.info("Done!")
-
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser(description="Holosoma Locomotion Controller for Unitree G1")
-    parser.add_argument(
-        "--repo-id",
-        type=str,
-        default=DEFAULT_HOLOSOMA_REPO_ID,
-        help=f"Hugging Face Hub repo ID for Holosoma policies (default: {DEFAULT_HOLOSOMA_REPO_ID})",
-    )
-    parser.add_argument(
-        "--policy",
-        type=str,
-        choices=["fastsac", "ppo"],
-        default="fastsac",
-        help="Policy type to use: 'fastsac' (default) or 'ppo'",
-    )
-    args = parser.parse_args()
-
-    run(repo_id=args.repo_id, policy_type=args.policy)

pyproject.toml

@@ -25,9 +25,9 @@ discord = "https://discord.gg/s3KuuzsPFb"
 
 [project]
 name = "lerobot"
-version = "0.4.4"
+version = "0.4.3"
 description = "🤗 LeRobot: State-of-the-art Machine Learning for Real-World Robotics in Pytorch"
-dynamic = ["readme"]
+readme = "README.md"
 license = { text = "Apache-2.0" }
 requires-python = ">=3.10"
 authors = [
@@ -74,7 +74,7 @@ dependencies = [
     "packaging>=24.2,<26.0",
     "pynput>=1.7.7,<1.9.0",
     "pyserial>=3.5,<4.0",
-    "wandb>=0.24.0,<0.25.0",
+    "wandb>=0.20.0,<0.22.0", # TODO: Bumb dependency (compatible with protobuf)
 
     "torch>=2.2.1,<2.8.0", # TODO: Bumb dependency
     "torchcodec>=0.2.1,<0.6.0; sys_platform != 'win32' and (sys_platform != 'linux' or (platform_machine != 'aarch64' and platform_machine != 'arm64' and platform_machine != 'armv7l')) and (sys_platform != 'darwin' or platform_machine != 'x86_64')", # TODO: Bumb dependency
@@ -96,8 +96,8 @@ dependencies = [
 # Common
 pygame-dep = ["pygame>=2.5.1,<2.7.0"]
 placo-dep = ["placo>=0.9.6,<0.10.0"]
-transformers-dep = ["transformers>=4.57.1,<5.0.0"]
-grpcio-dep = ["grpcio==1.73.1", "protobuf>=6.31.1,<6.32.0"]
+transformers-dep = ["transformers>=4.53.0,<5.0.0"]
+grpcio-dep = ["grpcio==1.73.1", "protobuf==6.31.0"] # TODO: Bumb dependency (compatible with wandb)
 
 # Motors
 feetech = ["feetech-servo-sdk>=1.0.0,<2.0.0"]
@@ -109,9 +109,9 @@ hopejr = ["lerobot[feetech]", "lerobot[pygame-dep]"]
 lekiwi = ["lerobot[feetech]", "pyzmq>=26.2.1,<28.0.0"]
 unitree_g1 = [
     "pyzmq>=26.2.1,<28.0.0",
-    "onnxruntime>=1.16.0,<2.0.0"
+    "onnxruntime>=1.16.0"
 ]
-reachy2 = ["reachy2_sdk>=1.0.15,<1.1.0"]
+reachy2 = ["reachy2_sdk>=1.0.14,<1.1.0"]
 kinematics = ["lerobot[placo-dep]"]
 intelrealsense = [
     "pyrealsense2>=2.55.1.6486,<2.57.0 ; sys_platform != 'darwin'",
@@ -120,14 +120,7 @@ intelrealsense = [
 phone = ["hebi-py>=2.8.0,<2.12.0", "teleop>=0.1.0,<0.2.0", "fastapi<1.0"]
 
 # Policies
-wallx = [
-    "transformers==4.49.0",
-    "peft==0.17.1",
-    "scipy==1.15.3",
-    "torchdiffeq==0.2.5",
-    "qwen_vl_utils==0.0.11"
-]
-pi = ["transformers @ git+https://github.com/huggingface/transformers.git@fix/lerobot_openpi", "scipy>=1.10.1,<1.15"]
+pi = ["transformers @ git+https://github.com/huggingface/transformers.git@fix/lerobot_openpi"]
 smolvla = ["lerobot[transformers-dep]", "num2words>=0.5.14,<0.6.0", "accelerate>=1.7.0,<2.0.0", "safetensors>=0.4.3,<1.0.0"]
 groot = [
     "lerobot[transformers-dep]",
@@ -140,16 +133,13 @@ groot = [
     "ninja>=1.11.1,<2.0.0",
     "flash-attn>=2.5.9,<3.0.0 ; sys_platform != 'darwin'"
 ]
-sarm = ["lerobot[transformers-dep]", "faker>=33.0.0,<35.0.0", "matplotlib>=3.10.3,<4.0.0", "qwen-vl-utils>=0.0.14,<0.1.0"]
-xvla = ["lerobot[transformers-dep]"]
 hilserl = ["lerobot[transformers-dep]", "gym-hil>=0.1.13,<0.2.0", "lerobot[grpcio-dep]", "lerobot[placo-dep]"]
 
 # Features
 async = ["lerobot[grpcio-dep]", "matplotlib>=3.10.3,<4.0.0"]
-peft = ["lerobot[transformers-dep]", "peft>=0.18.0,<1.0.0"]
 
 # Development
-dev = ["pre-commit>=3.7.0,<5.0.0", "debugpy>=1.8.1,<1.9.0", "lerobot[grpcio-dep]", "grpcio-tools==1.73.1", "mypy>=1.19.1"]
+dev = ["pre-commit>=3.7.0,<5.0.0", "debugpy>=1.8.1,<1.9.0", "lerobot[grpcio-dep]", "grpcio-tools==1.73.1"]
 test = ["pytest>=8.1.0,<9.0.0", "pytest-timeout>=2.4.0,<3.0.0", "pytest-cov>=5.0.0,<8.0.0", "mock-serial>=0.0.1,<0.1.0 ; sys_platform != 'win32'"]
 video_benchmark = ["scikit-image>=0.23.2,<0.26.0", "pandas>=2.2.2,<2.4.0"]
 
@@ -168,11 +158,9 @@ all = [
     "lerobot[reachy2]",
     "lerobot[kinematics]",
     "lerobot[intelrealsense]",
-    # "lerobot[wallx]",
-    # "lerobot[pi]", TODO(Pepijn): Update pi to transformers v5
+    "lerobot[pi]",
     "lerobot[smolvla]",
     # "lerobot[groot]", TODO(Steven): Gr00t requires specific installation instructions for flash-attn
-    "lerobot[xvla]",
     "lerobot[hilserl]",
     "lerobot[async]",
     "lerobot[dev]",
@@ -183,8 +171,6 @@ all = [
     "lerobot[phone]",
     "lerobot[libero]",
     "lerobot[metaworld]",
-    "lerobot[sarm]",
-    "lerobot[peft]",
 ]
 
 [project.scripts]
@@ -197,7 +183,6 @@ lerobot-setup-motors="lerobot.scripts.lerobot_setup_motors:main"
 lerobot-teleoperate="lerobot.scripts.lerobot_teleoperate:main"
 lerobot-eval="lerobot.scripts.lerobot_eval:main"
 lerobot-train="lerobot.scripts.lerobot_train:main"
-lerobot-train-tokenizer="lerobot.scripts.lerobot_train_tokenizer:main"
 lerobot-dataset-viz="lerobot.scripts.lerobot_dataset_viz:main"
 lerobot-info="lerobot.scripts.lerobot_info:main"
 lerobot-find-joint-limits="lerobot.scripts.lerobot_find_joint_limits:main"
@@ -240,7 +225,6 @@ ignore = [
 
 [tool.ruff.lint.per-file-ignores]
 "__init__.py" = ["F401", "F403"]
-"src/lerobot/policies/wall_x/**" = ["N801", "N812", "SIM102", "SIM108", "SIM210", "SIM211", "B006", "B007", "SIM118"] # Supprese these as they are coming from original Qwen2_5_vl code TODO(pepijn): refactor original
 
 [tool.ruff.lint.isort]
 combine-as-imports = true
@@ -277,7 +261,6 @@ default.extend-ignore-identifiers-re = [
     "ein",
     "thw",
     "inpt",
-    "ROBOTIS",
 ]
 
 # TODO: Uncomment when ready to use
@@ -332,9 +315,9 @@ disallow_untyped_defs = true
 disallow_incomplete_defs = true
 check_untyped_defs = true
 
-[[tool.mypy.overrides]]
-module = "lerobot.optim.*"
-ignore_errors = false
+# [[tool.mypy.overrides]]
+# module = "lerobot.optim.*"
+# ignore_errors = false
 
 [[tool.mypy.overrides]]
 module = "lerobot.model.*"
@@ -377,92 +360,10 @@ ignore_errors = false
 # module = "lerobot.async_inference.*"
 # ignore_errors = false
 
-[[tool.mypy.overrides]]
-module = "lerobot.transport.*"
-ignore_errors = false
+# [[tool.mypy.overrides]]
+# module = "lerobot.transport.*"
+# ignore_errors = false
 
 # [[tool.mypy.overrides]]
 # module = "lerobot.scripts.*"
 # ignore_errors = false
-
-[tool.uv]
-# wallx requires transformers==4.49.0 which conflicts with other extras that need >=4.53.0
-conflicts = [
-    [
-        { extra = "wallx" },
-        { extra = "transformers-dep" },
-    ],
-    [
-        { extra = "wallx" },
-        { extra = "pi" },
-    ],
-    [
-        { extra = "wallx" },
-        { extra = "smolvla" },
-    ],
-    [
-        { extra = "wallx" },
-        { extra = "groot" },
-    ],
-    [
-        { extra = "wallx" },
-        { extra = "xvla" },
-    ],
-    [
-        { extra = "wallx" },
-        { extra = "sarm" },
-    ],
-    [
-        { extra = "wallx" },
-        { extra = "hilserl" },
-    ],
-    [
-        { extra = "wallx" },
-        { extra = "libero" },
-    ],
-    [
-        { extra = "wallx" },
-        { extra = "peft" },
-    ],
-    [
-        { extra = "wallx" },
-        { extra = "all" },
-    ],
-    # pi uses custom branch which conflicts with transformers-dep
-    [
-        { extra = "pi" },
-        { extra = "transformers-dep" },
-    ],
-    [
-        { extra = "pi" },
-        { extra = "smolvla" },
-    ],
-    [
-        { extra = "pi" },
-        { extra = "groot" },
-    ],
-    [
-        { extra = "pi" },
-        { extra = "xvla" },
-    ],
-    [
-        { extra = "pi" },
-        { extra = "sarm" },
-    ],
-    [
-        { extra = "pi" },
-        { extra = "hilserl" },
-    ],
-    [
-        { extra = "pi" },
-        { extra = "libero" },
-    ],
-    [
-        { extra = "pi" },
-        { extra = "peft" },
-    ],
-    [
-        { extra = "pi" },
-        { extra = "all" },
-    ],
-]

setup.py

@@ -1,72 +0,0 @@
-# Copyright 2025 The HuggingFace Inc. team. All rights reserved.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-
-from setuptools import setup
-
-
-def get_version_from_toml() -> str:
-    """Return the project's version string parsed from `pyproject.toml`.
-
-    The function scans `pyproject.toml` line-by-line looking for a line
-    that starts with ``version`` (for example: ``version = "1.2.3"``)
-    and returns the value without surrounding quotes. If no such line is
-    found a :class:`ValueError` is raised.
-
-    Returns:
-        The version string from `pyproject.toml` (e.g. ``"1.2.3"`` ->
-        ``1.2.3``).
-    """
-
-    version = None
-    with open("pyproject.toml", encoding="utf-8") as f:
-        for line in f:
-            if line.strip().startswith("version"):
-                version = line.split("=")[1].strip().strip('"')
-                break
-    if version is None:
-        raise ValueError("Version not found in pyproject.toml")
-    return version
-
-
-def read_long_description() -> str:
-    """Read and return the project's long description for setup.
-
-    This function reads `README.md` and replaces image links that point
-    to the local `./media/` directory with absolute raw GitHub URLs that
-    reference the release tag corresponding to the version parsed from
-    `pyproject.toml` (for example, ``v1.2.3``). The modified README
-    content is returned as a string suitable for passing to
-    ``setuptools.setup(long_description=...)``.
-
-    Returns:
-        The README content with rewritten media links.
-    """
-
-    with open("README.md", encoding="utf-8") as f:
-        content = f.read()
-
-    version = get_version_from_toml()
-    git_tag = f"v{version}"
-
-    base_raw_url = f"https://raw.githubusercontent.com/huggingface/lerobot/{git_tag}/"
-    content = content.replace('src="./media/', f'src="{base_raw_url}media/')
-
-    return content
-
-
-setup(
-    long_description=read_long_description(),
-    long_description_content_type="text/markdown",
-)

src/lerobot/async_inference/configs.py

@@ -126,12 +126,6 @@ class RobotClientConfig:
 
     # Device configuration
     policy_device: str = field(default="cpu", metadata={"help": "Device for policy inference"})
-    client_device: str = field(
-        default="cpu",
-        metadata={
-            "help": "Device to move actions to after receiving from server (e.g., for downstream planners)"
-        },
-    )
 
     # Control behavior configuration
     chunk_size_threshold: float = field(default=0.5, metadata={"help": "Threshold for chunk size control"})
@@ -167,9 +161,6 @@ class RobotClientConfig:
         if not self.policy_device:
             raise ValueError("policy_device cannot be empty")
 
-        if not self.client_device:
-            raise ValueError("client_device cannot be empty")
-
         if self.chunk_size_threshold < 0 or self.chunk_size_threshold > 1:
             raise ValueError(f"chunk_size_threshold must be between 0 and 1, got {self.chunk_size_threshold}")
 
@@ -193,7 +184,6 @@ class RobotClientConfig:
             "policy_type": self.policy_type,
             "pretrained_name_or_path": self.pretrained_name_or_path,
             "policy_device": self.policy_device,
-            "client_device": self.client_device,
             "chunk_size_threshold": self.chunk_size_threshold,
             "fps": self.fps,
             "actions_per_chunk": self.actions_per_chunk,

src/lerobot/async_inference/constants.py

@@ -23,7 +23,7 @@ DEFAULT_INFERENCE_LATENCY = 1 / DEFAULT_FPS
 DEFAULT_OBS_QUEUE_TIMEOUT = 2
 
 # All action chunking policies
-SUPPORTED_POLICIES = ["act", "smolvla", "diffusion", "tdmpc", "vqbet", "pi0", "pi05", "groot"]
+SUPPORTED_POLICIES = ["act", "smolvla", "diffusion", "tdmpc", "vqbet", "pi0", "pi05"]
 
 # TODO: Add all other robots
-SUPPORTED_ROBOTS = ["so100_follower", "so101_follower", "bi_so_follower", "omx_follower"]
+SUPPORTED_ROBOTS = ["so100_follower", "so101_follower", "bi_so100_follower"]

src/lerobot/async_inference/helpers.py

@@ -18,7 +18,6 @@ import os
 import time
 from dataclasses import dataclass, field
 from pathlib import Path
-from typing import Any
 
 import torch
 
@@ -40,8 +39,8 @@ from lerobot.utils.utils import init_logging
 
 Action = torch.Tensor
 
-# observation as received from the robot (can be numpy arrays, floats, etc.)
-RawObservation = dict[str, Any]
+# observation as received from the robot
+RawObservation = dict[str, torch.Tensor]
 
 # observation as those recorded in LeRobot dataset (keys are different)
 LeRobotObservation = dict[str, torch.Tensor]

src/lerobot/async_inference/policy_server.py

@@ -381,8 +381,6 @@ class PolicyServer(services_pb2_grpc.AsyncInferenceServicer):
         action_tensor = torch.stack(processed_actions, dim=1).squeeze(0)
         self.logger.debug(f"Postprocessed action shape: {action_tensor.shape}")
 
-        action_tensor = action_tensor.detach().cpu()
-
         """5. Convert to TimedAction list"""
         action_chunk = self._time_action_chunk(
             observation_t.get_timestamp(), list(action_tensor), observation_t.get_timestep()

src/lerobot/async_inference/robot_client.py

@@ -25,7 +25,6 @@ python src/lerobot/async_inference/robot_client.py \
     --policy_type=act \
     --pretrained_name_or_path=user/model \
     --policy_device=mps \
-    --client_device=cpu \
     --actions_per_chunk=50 \
     --chunk_size_threshold=0.5 \
     --aggregate_fn_name=weighted_average \
@@ -52,11 +51,11 @@ from lerobot.cameras.realsense.configuration_realsense import RealSenseCameraCon
 from lerobot.robots import (  # noqa: F401
     Robot,
     RobotConfig,
-    bi_so_follower,
+    bi_so100_follower,
     koch_follower,
     make_robot_from_config,
-    omx_follower,
-    so_follower,
+    so100_follower,
+    so101_follower,
 )
 from lerobot.transport import (
     services_pb2,  # type: ignore
@@ -286,21 +285,6 @@ class RobotClient:
                 timed_actions = pickle.loads(actions_chunk.data)  # nosec
                 deserialize_time = time.perf_counter() - deserialize_start
 
-                # Log device type of received actions
-                if len(timed_actions) > 0:
-                    received_device = timed_actions[0].get_action().device.type
-                    self.logger.debug(f"Received actions on device: {received_device}")
-
-                # Move actions to client_device (e.g., for downstream planners that need GPU)
-                client_device = self.config.client_device
-                if client_device != "cpu":
-                    for timed_action in timed_actions:
-                        if timed_action.get_action().device.type != client_device:
-                            timed_action.action = timed_action.get_action().to(client_device)
-                    self.logger.debug(f"Converted actions to device: {client_device}")
-                else:
-                    self.logger.debug(f"Actions kept on device: {client_device}")
-
                 self.action_chunk_size = max(self.action_chunk_size, len(timed_actions))
 
                 # Calculate network latency if we have matching observations

src/lerobot/cameras/reachy2_camera/configuration_reachy2_camera.py

@@ -35,19 +35,18 @@ class Reachy2CameraConfig(CameraConfig):
         name="teleop",
         image_type="left",
         ip_address="192.168.0.200",  # IP address of the robot
-        port=50065,  # Port of the camera server
+        fps=15,
         width=640,
         height=480,
-        fps=30,  # Not configurable for Reachy 2 cameras
         color_mode=ColorMode.RGB,
-    )  # Left teleop camera, 640x480 @ 30FPS
+    )  # Left teleop camera, 640x480 @ 15FPS
     ```
 
     Attributes:
         name: Name of the camera device. Can be "teleop" or "depth".
         image_type: Type of image stream. For "teleop" camera, can be "left" or "right".
                     For "depth" camera, can be "rgb" or "depth". (depth is not supported yet)
-        fps: Requested frames per second for the color stream. Not configurable for Reachy 2 cameras.
+        fps: Requested frames per second for the color stream.
         width: Requested frame width in pixels for the color stream.
         height: Requested frame height in pixels for the color stream.
         color_mode: Color mode for image output (RGB or BGR). Defaults to RGB.
@@ -63,6 +62,7 @@ class Reachy2CameraConfig(CameraConfig):
     color_mode: ColorMode = ColorMode.RGB
     ip_address: str | None = "localhost"
     port: int = 50065
+    # use_depth: bool = False
 
     def __post_init__(self) -> None:
         if self.name not in ["teleop", "depth"]:

src/lerobot/cameras/reachy2_camera/reachy2_camera.py

@@ -16,13 +16,12 @@
 Provides the Reachy2Camera class for capturing frames from Reachy 2 cameras using Reachy 2's CameraManager.
 """
 
-from __future__ import annotations
-
 import logging
 import os
 import platform
 import time
-from typing import TYPE_CHECKING, Any
+from threading import Event, Lock, Thread
+from typing import Any
 
 from numpy.typing import NDArray  # type: ignore  # TODO: add type stubs for numpy.typing
 
@@ -31,19 +30,10 @@ if platform.system() == "Windows" and "OPENCV_VIDEOIO_MSMF_ENABLE_HW_TRANSFORMS"
     os.environ["OPENCV_VIDEOIO_MSMF_ENABLE_HW_TRANSFORMS"] = "0"
 import cv2  # type: ignore  # TODO: add type stubs for OpenCV
 import numpy as np  # type: ignore  # TODO: add type stubs for numpy
-
-from lerobot.utils.import_utils import _reachy2_sdk_available
-
-if TYPE_CHECKING or _reachy2_sdk_available:
-    from reachy2_sdk.media.camera import CameraView
-    from reachy2_sdk.media.camera_manager import CameraManager
-else:
-    CameraManager = None
-
-    class CameraView:
-        LEFT = 0
-        RIGHT = 1
-
+from reachy2_sdk.media.camera import CameraView  # type: ignore  # TODO: add type stubs for reachy2_sdk
+from reachy2_sdk.media.camera_manager import (  # type: ignore  # TODO: add type stubs for reachy2_sdk
+    CameraManager,
+)
 
 from lerobot.utils.errors import DeviceNotConnectedError
 
@@ -79,10 +69,17 @@ class Reachy2Camera(Camera):
 
         self.config = config
 
+        self.fps = config.fps
         self.color_mode = config.color_mode
 
         self.cam_manager: CameraManager | None = None
 
+        self.thread: Thread | None = None
+        self.stop_event: Event | None = None
+        self.frame_lock: Lock = Lock()
+        self.latest_frame: NDArray[Any] | None = None
+        self.new_frame_event: Event = Event()
+
     def __str__(self) -> str:
         return f"{self.__class__.__name__}({self.config.name}, {self.config.image_type})"
 
@@ -103,23 +100,44 @@ class Reachy2Camera(Camera):
     def connect(self, warmup: bool = True) -> None:
         """
         Connects to the Reachy2 CameraManager as specified in the configuration.
-
-        Raises:
-            DeviceNotConnectedError: If the camera is not connected.
         """
         self.cam_manager = CameraManager(host=self.config.ip_address, port=self.config.port)
-        if self.cam_manager is None:
-            raise DeviceNotConnectedError(f"Could not connect to {self}.")
         self.cam_manager.initialize_cameras()
 
         logger.info(f"{self} connected.")
 
     @staticmethod
-    def find_cameras() -> list[dict[str, Any]]:
+    def find_cameras(ip_address: str = "localhost", port: int = 50065) -> list[dict[str, Any]]:
         """
-        Detection not implemented for Reachy2 cameras.
+        Detects available Reachy 2 cameras.
+
+        Returns:
+            List[Dict[str, Any]]: A list of dictionaries,
+            where each dictionary contains 'name', 'stereo',
+            and the default profile properties (width, height, fps).
         """
-        raise NotImplementedError("Camera detection is not implemented for Reachy2 cameras.")
+        initialized_cameras = []
+        camera_manager = CameraManager(host=ip_address, port=port)
+
+        for camera in [camera_manager.teleop, camera_manager.depth]:
+            if camera is None:
+                continue
+
+            height, width, _, _, _, _, _ = camera.get_parameters()
+
+            camera_info = {
+                "name": camera._cam_info.name,
+                "stereo": camera._cam_info.stereo,
+                "default_profile": {
+                    "width": width,
+                    "height": height,
+                    "fps": 30,
+                },
+            }
+            initialized_cameras.append(camera_info)
+
+        camera_manager.disconnect()
+        return initialized_cameras
 
     def read(self, color_mode: ColorMode | None = None) -> NDArray[Any]:
         """
@@ -137,49 +155,95 @@ class Reachy2Camera(Camera):
                        (height, width, channels), using the specified or default
                        color mode and applying any configured rotation.
         """
-        start_time = time.perf_counter()
-
         if not self.is_connected:
             raise DeviceNotConnectedError(f"{self} is not connected.")
 
-        if self.cam_manager is None:
-            raise DeviceNotConnectedError(f"{self} is not connected.")
+        start_time = time.perf_counter()
 
         frame: NDArray[Any] = np.empty((0, 0, 3), dtype=np.uint8)
 
-        if self.config.name == "teleop" and hasattr(self.cam_manager, "teleop"):
-            if self.config.image_type == "left":
-                frame = self.cam_manager.teleop.get_frame(
-                    CameraView.LEFT, size=(self.config.width, self.config.height)
-                )[0]
-            elif self.config.image_type == "right":
-                frame = self.cam_manager.teleop.get_frame(
-                    CameraView.RIGHT, size=(self.config.width, self.config.height)
-                )[0]
-        elif self.config.name == "depth" and hasattr(self.cam_manager, "depth"):
-            if self.config.image_type == "depth":
-                frame = self.cam_manager.depth.get_depth_frame()[0]
-            elif self.config.image_type == "rgb":
-                frame = self.cam_manager.depth.get_frame(size=(self.config.width, self.config.height))[0]
+        if self.cam_manager is None:
+            raise DeviceNotConnectedError(f"{self} is not connected.")
         else:
-            raise ValueError(f"Invalid camera name '{self.config.name}'. Expected 'teleop' or 'depth'.")
+            if self.config.name == "teleop" and hasattr(self.cam_manager, "teleop"):
+                if self.config.image_type == "left":
+                    frame = self.cam_manager.teleop.get_frame(CameraView.LEFT, size=(640, 480))[0]
+                elif self.config.image_type == "right":
+                    frame = self.cam_manager.teleop.get_frame(CameraView.RIGHT, size=(640, 480))[0]
+            elif self.config.name == "depth" and hasattr(self.cam_manager, "depth"):
+                if self.config.image_type == "depth":
+                    frame = self.cam_manager.depth.get_depth_frame()[0]
+                elif self.config.image_type == "rgb":
+                    frame = self.cam_manager.depth.get_frame(size=(640, 480))[0]
 
-        if frame is None:
-            return np.empty((0, 0, 3), dtype=np.uint8)
+            if frame is None:
+                return np.empty((0, 0, 3), dtype=np.uint8)
 
-        if self.config.color_mode == "rgb":
-            frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
+            if self.config.color_mode == "rgb":
+                frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
 
         read_duration_ms = (time.perf_counter() - start_time) * 1e3
         logger.debug(f"{self} read took: {read_duration_ms:.1f}ms")
 
         return frame
 
+    def _read_loop(self) -> None:
+        """
+        Internal loop run by the background thread for asynchronous reading.
+
+        On each iteration:
+        1. Reads a color frame
+        2. Stores result in latest_frame (thread-safe)
+        3. Sets new_frame_event to notify listeners
+
+        Stops on DeviceNotConnectedError, logs other errors and continues.
+        """
+        if self.stop_event is None:
+            raise RuntimeError(f"{self}: stop_event is not initialized before starting read loop.")
+
+        while not self.stop_event.is_set():
+            try:
+                color_image = self.read()
+
+                with self.frame_lock:
+                    self.latest_frame = color_image
+                self.new_frame_event.set()
+
+            except DeviceNotConnectedError:
+                break
+            except Exception as e:
+                logger.warning(f"Error reading frame in background thread for {self}: {e}")
+
+    def _start_read_thread(self) -> None:
+        """Starts or restarts the background read thread if it's not running."""
+        if self.thread is not None and self.thread.is_alive():
+            self.thread.join(timeout=0.1)
+        if self.stop_event is not None:
+            self.stop_event.set()
+
+        self.stop_event = Event()
+        self.thread = Thread(target=self._read_loop, args=(), name=f"{self}_read_loop")
+        self.thread.daemon = True
+        self.thread.start()
+
+    def _stop_read_thread(self) -> None:
+        """Signals the background read thread to stop and waits for it to join."""
+        if self.stop_event is not None:
+            self.stop_event.set()
+
+        if self.thread is not None and self.thread.is_alive():
+            self.thread.join(timeout=2.0)
+
+        self.thread = None
+        self.stop_event = None
+
     def async_read(self, timeout_ms: float = 200) -> NDArray[Any]:
         """
-        Reads the latest available frame.
+        Reads the latest available frame asynchronously.
 
-        This method retrieves the most recent frame available in Reachy 2's low-level software.
+        This method retrieves the most recent frame captured by the background
+        read thread. It does not block waiting for the camera hardware directly,
+        but may wait up to timeout_ms for the background thread to provide a frame.
 
         Args:
             timeout_ms (float): Maximum time in milliseconds to wait for a frame
@@ -197,10 +261,22 @@ class Reachy2Camera(Camera):
         if not self.is_connected:
             raise DeviceNotConnectedError(f"{self} is not connected.")
 
-        frame = self.read()
+        if self.thread is None or not self.thread.is_alive():
+            self._start_read_thread()
+
+        if not self.new_frame_event.wait(timeout=timeout_ms / 1000.0):
+            thread_alive = self.thread is not None and self.thread.is_alive()
+            raise TimeoutError(
+                f"Timed out waiting for frame from camera {self} after {timeout_ms} ms. "
+                f"Read thread alive: {thread_alive}."
+            )
+
+        with self.frame_lock:
+            frame = self.latest_frame
+            self.new_frame_event.clear()
 
         if frame is None:
-            raise RuntimeError(f"Internal error: No frame available for {self}.")
+            raise RuntimeError(f"Internal error: Event set but no frame available for {self}.")
 
         return frame
 
@@ -211,9 +287,12 @@ class Reachy2Camera(Camera):
         Raises:
             DeviceNotConnectedError: If the camera is already disconnected.
         """
-        if not self.is_connected:
+        if not self.is_connected and self.thread is None:
             raise DeviceNotConnectedError(f"{self} not connected.")
 
+        if self.thread is not None:
+            self._stop_read_thread()
+
         if self.cam_manager is not None:
             self.cam_manager.disconnect()
 

src/lerobot/cameras/utils.py

@@ -43,11 +43,6 @@ def make_cameras_from_configs(camera_configs: dict[str, CameraConfig]) -> dict[s
 
             cameras[key] = Reachy2Camera(cfg)
 
-        elif cfg.type == "zmq":
-            from .zmq.camera_zmq import ZMQCamera
-
-            cameras[key] = ZMQCamera(cfg)
-
         else:
             try:
                 cameras[key] = cast(Camera, make_device_from_device_class(cfg))

src/lerobot/cameras/zmq/__init__.py

@@ -1,20 +0,0 @@
-#!/usr/bin/env python
-
-# Copyright 2026 The HuggingFace Inc. team. All rights reserved.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-from .camera_zmq import ZMQCamera
-from .configuration_zmq import ZMQCameraConfig
-
-__all__ = ["ZMQCamera", "ZMQCameraConfig"]

src/lerobot/cameras/zmq/camera_zmq.py

@@ -1,235 +0,0 @@
-#!/usr/bin/env python
-
-# Copyright 2025 The HuggingFace Inc. team. All rights reserved.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-"""
-ZMQCamera - Captures frames from remote cameras via ZeroMQ using JSON protocol in the
-following format:
-    {
-        "timestamps": {"camera_name": float},
-        "images": {"camera_name": "<base64-jpeg>"}
-    }
-"""
-
-import base64
-import json
-import logging
-import time
-from threading import Event, Lock, Thread
-from typing import Any
-
-import cv2
-import numpy as np
-from numpy.typing import NDArray
-
-from lerobot.utils.errors import DeviceAlreadyConnectedError, DeviceNotConnectedError
-
-from ..camera import Camera
-from ..configs import ColorMode
-from .configuration_zmq import ZMQCameraConfig
-
-logger = logging.getLogger(__name__)
-
-
-class ZMQCamera(Camera):
-    """
-    Example usage:
-        ```python
-        from lerobot.cameras.zmq import ZMQCamera, ZMQCameraConfig
-
-        config = ZMQCameraConfig(server_address="192.168.123.164", port=5555, camera_name="head_camera")
-        camera = ZMQCamera(config)
-        camera.connect()
-        frame = camera.read()
-        camera.disconnect()
-        ```
-    """
-
-    def __init__(self, config: ZMQCameraConfig):
-        super().__init__(config)
-        import zmq
-
-        self.config = config
-        self.server_address = config.server_address
-        self.port = config.port
-        self.camera_name = config.camera_name
-        self.color_mode = config.color_mode
-        self.timeout_ms = config.timeout_ms
-
-        self.context: zmq.Context | None = None
-        self.socket: zmq.Socket | None = None
-        self._connected = False
-
-        self.thread: Thread | None = None
-        self.stop_event: Event | None = None
-        self.frame_lock: Lock = Lock()
-        self.latest_frame: NDArray[Any] | None = None
-        self.new_frame_event: Event = Event()
-
-    def __str__(self) -> str:
-        return f"ZMQCamera({self.camera_name}@{self.server_address}:{self.port})"
-
-    @property
-    def is_connected(self) -> bool:
-        return self._connected and self.context is not None and self.socket is not None
-
-    def connect(self, warmup: bool = True) -> None:
-        """Connect to ZMQ camera server."""
-        if self.is_connected:
-            raise DeviceAlreadyConnectedError(f"{self} is already connected.")
-
-        logger.info(f"Connecting to {self}...")
-
-        try:
-            import zmq
-
-            self.context = zmq.Context()
-            self.socket = self.context.socket(zmq.SUB)
-            self.socket.setsockopt_string(zmq.SUBSCRIBE, "")
-            self.socket.setsockopt(zmq.RCVTIMEO, self.timeout_ms)
-            self.socket.setsockopt(zmq.CONFLATE, True)
-            self.socket.connect(f"tcp://{self.server_address}:{self.port}")
-            self._connected = True
-
-            # Auto-detect resolution
-            if self.width is None or self.height is None:
-                h, w = self.read().shape[:2]
-                self.height = h
-                self.width = w
-                logger.info(f"{self} resolution: {w}x{h}")
-
-            logger.info(f"{self} connected.")
-
-            if warmup:
-                time.sleep(0.1)
-
-        except Exception as e:
-            self._cleanup()
-            raise RuntimeError(f"Failed to connect to {self}: {e}") from e
-
-    def _cleanup(self):
-        """Clean up ZMQ resources."""
-        self._connected = False
-        if self.socket:
-            self.socket.close()
-            self.socket = None
-        if self.context:
-            self.context.term()
-            self.context = None
-
-    @staticmethod
-    def find_cameras() -> list[dict[str, Any]]:
-        """ZMQ cameras require manual configuration (server address/port)."""
-        return []
-
-    def read(self, color_mode: ColorMode | None = None) -> NDArray[Any]:
-        """
-        Read a single frame from the ZMQ camera.
-
-        Returns:
-            np.ndarray: Decoded frame (height, width, 3)
-        """
-        if not self.is_connected or self.socket is None:
-            raise DeviceNotConnectedError(f"{self} is not connected.")
-
-        try:
-            message = self.socket.recv_string()
-        except Exception as e:
-            if type(e).__name__ == "Again":
-                raise TimeoutError(f"{self} timeout after {self.timeout_ms}ms") from e
-            raise
-
-        # Decode JSON message
-        data = json.loads(message)
-
-        if "images" not in data:
-            raise RuntimeError(f"{self} invalid message: missing 'images' key")
-
-        images = data["images"]
-
-        # Get image by camera name or first available
-        if self.camera_name in images:
-            img_b64 = images[self.camera_name]
-        elif images:
-            img_b64 = next(iter(images.values()))
-        else:
-            raise RuntimeError(f"{self} no images in message")
-
-        # Decode base64 JPEG
-        img_bytes = base64.b64decode(img_b64)
-        frame = cv2.imdecode(np.frombuffer(img_bytes, np.uint8), cv2.IMREAD_COLOR)
-
-        if frame is None:
-            raise RuntimeError(f"{self} failed to decode image")
-
-        return frame
-
-    def _read_loop(self) -> None:
-        while self.stop_event and not self.stop_event.is_set():
-            try:
-                frame = self.read()
-                with self.frame_lock:
-                    self.latest_frame = frame
-                self.new_frame_event.set()
-            except DeviceNotConnectedError:
-                break
-            except TimeoutError:
-                pass
-            except Exception as e:
-                logger.warning(f"Read error: {e}")
-
-    def _start_read_thread(self) -> None:
-        if self.thread and self.thread.is_alive():
-            return
-        self.stop_event = Event()
-        self.thread = Thread(target=self._read_loop, daemon=True)
-        self.thread.start()
-
-    def _stop_read_thread(self) -> None:
-        if self.stop_event:
-            self.stop_event.set()
-        if self.thread and self.thread.is_alive():
-            self.thread.join(timeout=2.0)
-        self.thread = None
-        self.stop_event = None
-
-    def async_read(self, timeout_ms: float = 10000) -> NDArray[Any]:
-        """Read latest frame asynchronously (non-blocking)."""
-        if not self.is_connected:
-            raise DeviceNotConnectedError(f"{self} is not connected.")
-
-        if not self.thread or not self.thread.is_alive():
-            self._start_read_thread()
-
-        if not self.new_frame_event.wait(timeout=timeout_ms / 1000.0):
-            raise TimeoutError(f"{self} async_read timeout after {timeout_ms}ms")
-
-        with self.frame_lock:
-            frame = self.latest_frame
-            self.new_frame_event.clear()
-
-        if frame is None:
-            raise RuntimeError(f"{self} no frame available")
-
-        return frame
-
-    def disconnect(self) -> None:
-        """Disconnect from ZMQ camera."""
-        if not self.is_connected and not self.thread:
-            raise DeviceNotConnectedError(f"{self} not connected.")
-
-        self._stop_read_thread()
-        self._cleanup()
-        logger.info(f"{self} disconnected.")