ydy0615/lerobot-clone
1
0
Fork 0
mirror of https://github.com/huggingface/lerobot.git synced 2026-06-04 12:51:27 +00:00
Code Issues Packages Projects Releases Wiki Activity

refactor(processors): enhance transform_features method across multiple processors (#1849)

Browse Source 
* refactor(processors): enhance transform_features method across multiple processors

- Updated the transform_features method in various processors to utilize a copy of the features dictionary, ensuring immutability of the original features.
- Added handling for new feature keys and removed obsolete ones in the MapTensorToDeltaActionDict, JointVelocityProcessor, and others.
- Improved readability and maintainability by following consistent patterns in feature transformation.

* refactor(processors): standardize action and observation keys in delta_action_processor and joint_observations_processor

- Updated action and observation keys to use constants for improved readability and maintainability.
- Refactored the transform_features method in multiple processors to ensure consistent handling of feature keys.
- Enhanced error handling by raising exceptions for missing required components in action and observation processing.
- Removed obsolete code and improved overall structure for better clarity.

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* refactor(processors): remove unused import in joint_observations_processor

* refactor(processors): simplify transform_features method in delta_action_processor

* refactor(processors): streamline transform_features method in ImageCropResizeProcessor

* refactor(processors): improve error handling and streamline transform_features method in phone_processor

- Raised a ValueError for missing position and rotation in action to enhance error handling.

* refactor(processors): enhance error handling in JointVelocityProcessor

- Added a ValueError raise for missing current joint positions in the observation method to improve error handling and ensure the integrity of the transform_features method.

* refactor(processors): simplify transform_features method in robot kinematic processors

* refactor(processors): standardize action keys in phone_processor

* fix(processor): RKP feature obs -> act

---------

Signed-off-by: Adil Zouitine <adilzouitinegm@gmail.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Steven Palma <steven.palma@huggingface.co>
This commit is contained in:
Adil Zouitine
2025-09-03 16:54:41 +02:00
committed by GitHub
parent 2fcc358e98
commit 4ebe482a7e
6 changed files with 111 additions and 93 deletions
Download Patch File Download Diff File Expand all files Collapse all files

51
src/lerobot/teleoperators/phone/phone_processor.py
View File

@@ -17,6 +17,7 @@
from dataclasses import dataclass, field
from lerobot.configs.types import FeatureType, PolicyFeature
from lerobot.constants import ACTION
from lerobot.processor import ActionProcessor, ProcessorStepRegistry
from lerobot.teleoperators.phone.config_phone import PhoneOS
@@ -48,13 +49,13 @@ class MapPhoneActionToRobotAction(ActionProcessor):
def action(self, act: dict) -> dict:
# Pop them from the action
enabled = bool(act.pop("action.phone.enabled", 0))
pos = act.pop("action.phone.pos", None)
rot = act.pop("action.phone.rot", None)
inputs = act.pop("action.phone.raw_inputs", {})
enabled = bool(act.pop(f"{ACTION}.phone.enabled", 0))
pos = act.pop(f"{ACTION}.phone.pos", None)
rot = act.pop(f"{ACTION}.phone.rot", None)
inputs = act.pop(f"{ACTION}.phone.raw_inputs", {})
if pos is None or rot is None:
return act
raise ValueError("pos and rot must be present in action")
rotvec = rot.as_rotvec() # Absolute orientation as rotvec
@@ -69,28 +70,28 @@ class MapPhoneActionToRobotAction(ActionProcessor):
) # Positive if a is pressed, negative if b is pressed, 0 if both or neither are pressed
# For some actions we need to invert the axis
act["action.enabled"] = enabled
act["action.target_x"] = -pos[1] if enabled else 0.0
act["action.target_y"] = pos[0] if enabled else 0.0
act["action.target_z"] = pos[2] if enabled else 0.0
act["action.target_wx"] = rotvec[1] if enabled else 0.0
act["action.target_wy"] = rotvec[0] if enabled else 0.0
act["action.target_wz"] = -rotvec[2] if enabled else 0.0
act["action.gripper"] = gripper # Still send gripper action when disabled
act[f"{ACTION}.enabled"] = enabled
act[f"{ACTION}.target_x"] = -pos[1] if enabled else 0.0
act[f"{ACTION}.target_y"] = pos[0] if enabled else 0.0
act[f"{ACTION}.target_z"] = pos[2] if enabled else 0.0
act[f"{ACTION}.target_wx"] = rotvec[1] if enabled else 0.0
act[f"{ACTION}.target_wy"] = rotvec[0] if enabled else 0.0
act[f"{ACTION}.target_wz"] = -rotvec[2] if enabled else 0.0
act[f"{ACTION}.gripper"] = gripper # Still send gripper action when disabled
return act
def transform_features(self, features: dict[str, PolicyFeature]) -> dict[str, PolicyFeature]:
features.pop("action.phone.enabled", None)
features.pop("action.phone.pos", None)
features.pop("action.phone.rot", None)
features.pop("action.phone.raw_inputs", None)
features.pop(f"{ACTION}.phone.enabled", None)
features.pop(f"{ACTION}.phone.pos", None)
features.pop(f"{ACTION}.phone.rot", None)
features.pop(f"{ACTION}.phone.raw_inputs", None)
features["action.enabled"] = (PolicyFeature(type=FeatureType.ACTION, shape=(1,)),)
features["action.target_x"] = (PolicyFeature(type=FeatureType.ACTION, shape=(1,)),)
features["action.target_y"] = (PolicyFeature(type=FeatureType.ACTION, shape=(1,)),)
features["action.target_z"] = (PolicyFeature(type=FeatureType.ACTION, shape=(1,)),)
features["action.target_wx"] = (PolicyFeature(type=FeatureType.ACTION, shape=(1,)),)
features["action.target_wy"] = (PolicyFeature(type=FeatureType.ACTION, shape=(1,)),)
features["action.target_wz"] = (PolicyFeature(type=FeatureType.ACTION, shape=(1,)),)
features["action.gripper"] = (PolicyFeature(type=FeatureType.ACTION, shape=(1,)),)
features[f"{ACTION}.enabled"] = PolicyFeature(type=FeatureType.ACTION, shape=(1,))
features[f"{ACTION}.target_x"] = PolicyFeature(type=FeatureType.ACTION, shape=(1,))
features[f"{ACTION}.target_y"] = PolicyFeature(type=FeatureType.ACTION, shape=(1,))
features[f"{ACTION}.target_z"] = PolicyFeature(type=FeatureType.ACTION, shape=(1,))
features[f"{ACTION}.target_wx"] = PolicyFeature(type=FeatureType.ACTION, shape=(1,))
features[f"{ACTION}.target_wy"] = PolicyFeature(type=FeatureType.ACTION, shape=(1,))
features[f"{ACTION}.target_wz"] = PolicyFeature(type=FeatureType.ACTION, shape=(1,))
features[f"{ACTION}.gripper"] = PolicyFeature(type=FeatureType.ACTION, shape=(1,))
return features