feat(envs): add RoboTwin 2.0 benchmark integration

- RoboTwinEnvConfig with 4-camera setup (head/front/left_wrist/right_wrist)
- Docker image with SAPIEN, mplib, CuRobo, pytorch3d (Python 3.12)
- CI workflow: 1-episode smoke eval with pepijn223/smolvla_robotwin
- RoboTwinProcessorStep for state float32 casting
- Camera rename_map: head_camera/front_camera/left_wrist -> camera1/2/3

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Pepijn
2026-04-14 11:49:15 +02:00
parent cff4bcf4a0
commit 5558ea2207
7 changed files with 1364 additions and 0 deletions
+91
View File
@@ -310,3 +310,94 @@ jobs:
name: metaworld-metrics
path: /tmp/metaworld-artifacts/metrics.json
if-no-files-found: warn
# ── ROBOTWIN 2.0 ──────────────────────────────────────────────────────────
# Isolated image: full RoboTwin 2.0 stack — SAPIEN, mplib, CuRobo,
# pytorch3d, + simulation assets (~4 GB).
# Build takes ~20 min on first run; subsequent runs hit the layer cache.
# Requires an NVIDIA GPU runner with CUDA 12.1 drivers.
robotwin-integration-test:
name: RoboTwin 2.0 — build image + 1-episode eval
runs-on:
group: aws-g6-4xlarge-plus
env:
HF_USER_TOKEN: ${{ secrets.LEROBOT_HF_USER }}
steps:
- uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
with:
persist-credentials: false
lfs: true
- name: Set up Docker Buildx
uses: docker/setup-buildx-action@v3 # zizmor: ignore[unpinned-uses]
with:
cache-binary: false
# Build the full-install image: SAPIEN, mplib, CuRobo, pytorch3d +
# simulation assets (~4 GB). Layer cache lives in the runner's local
# Docker daemon — reused across re-runs on the same machine.
- name: Build RoboTwin 2.0 benchmark image
uses: docker/build-push-action@v6 # zizmor: ignore[unpinned-uses]
with:
context: .
file: docker/Dockerfile.benchmark.robotwin
push: false
load: true
tags: lerobot-benchmark-robotwin:ci
cache-from: type=local,src=/tmp/.buildx-cache-robotwin
cache-to: type=local,dest=/tmp/.buildx-cache-robotwin,mode=max
- name: Run RoboTwin 2.0 smoke eval (1 episode)
run: |
# Named container (no --rm) so we can docker cp artifacts out.
docker run --name robotwin-eval --gpus all \
--shm-size=4g \
-e HF_HOME=/tmp/hf \
-e HF_USER_TOKEN="${HF_USER_TOKEN}" \
lerobot-benchmark-robotwin:ci \
bash -c "
hf auth login --token \"\$HF_USER_TOKEN\" --add-to-git-credential 2>/dev/null || true
cd /opt/robotwin && lerobot-eval \
--policy.path=pepijn223/smolvla_robotwin \
--env.type=robotwin \
--env.task=beat_block_hammer \
--eval.batch_size=1 \
--eval.n_episodes=1 \
--eval.use_async_envs=false \
--policy.device=cuda \
'--rename_map={\"observation.images.head_camera\": \"observation.images.camera1\", \"observation.images.front_camera\": \"observation.images.camera2\", \"observation.images.left_wrist\": \"observation.images.camera3\"}' \
--output_dir=/tmp/eval-artifacts
"
- name: Copy RoboTwin artifacts from container
if: always()
run: |
mkdir -p /tmp/robotwin-artifacts
docker cp robotwin-eval:/tmp/eval-artifacts/. /tmp/robotwin-artifacts/ 2>/dev/null || true
docker rm -f robotwin-eval || true
- name: Parse RoboTwin eval metrics
if: always()
run: |
python3 scripts/ci/parse_eval_metrics.py \
--artifacts-dir /tmp/robotwin-artifacts \
--env robotwin \
--task beat_block_hammer \
--policy pepijn223/smolvla_robotwin
- name: Upload RoboTwin rollout video
if: always()
uses: actions/upload-artifact@v4
with:
name: robotwin-rollout-video
path: /tmp/robotwin-artifacts/videos/
if-no-files-found: warn
- name: Upload RoboTwin eval metrics
if: always()
uses: actions/upload-artifact@v4
with:
name: robotwin-metrics
path: /tmp/robotwin-artifacts/metrics.json
if-no-files-found: warn