General fixes in code, removed delta action, fixed grasp penalty, added logic to put gripper reward in info

lerobot/common/envs/configs.py

@@ -171,7 +171,6 @@ class VideoRecordConfig:
 class WrapperConfig:
     """Configuration for environment wrappers."""
 
-    delta_action: float | None = None
     joint_masking_action_space: list[bool] | None = None
 
 
@@ -191,7 +190,6 @@ class EnvWrapperConfig:
     """Configuration for environment wrappers."""
 
     display_cameras: bool = False
-    delta_action: float = 0.1
     use_relative_joint_positions: bool = True
     add_joint_velocity_to_observation: bool = False
     add_ee_pose_to_observation: bool = False
@@ -203,11 +201,13 @@ class EnvWrapperConfig:
     joint_masking_action_space: Optional[Any] = None
     ee_action_space_params: Optional[EEActionSpaceConfig] = None
     use_gripper: bool = False
-    gripper_quantization_threshold: float = 0.8
-    gripper_penalty: float = 0.0
+    gripper_quantization_threshold: float | None = 0.8
+    gripper_penalty: float = 0.0 
+    gripper_penalty_in_reward: bool = False
     open_gripper_on_reset: bool = False
 
 
+
 @EnvConfig.register_subclass(name="gym_manipulator")
 @dataclass
 class HILSerlRobotEnvConfig(EnvConfig):

lerobot/common/policies/sac/modeling_sac.py

@@ -428,6 +428,7 @@ class SACPolicy(
         actions_discrete = torch.round(actions_discrete)
         actions_discrete = actions_discrete.long()
 
+        gripper_penalties: Tensor | None = None
         if complementary_info is not None:
             gripper_penalties: Tensor | None = complementary_info.get("gripper_penalty")
 

lerobot/common/robot_devices/control_utils.py

@@ -221,7 +221,6 @@ def record_episode(
         events=events,
         policy=policy,
         fps=fps,
-        # record_delta_actions=record_delta_actions,
         teleoperate=policy is None,
         single_task=single_task,
     )
@@ -267,8 +266,6 @@ def control_loop(
 
         if teleoperate:
             observation, action = robot.teleop_step(record_data=True)
-            # if record_delta_actions:
-            #     action["action"] = action["action"] - current_joint_positions
         else:
             observation = robot.capture_observation()