SO101 ROS 2 Workspace

Summary

The SO101 ROS 2 workspace connects Lerobot’s leader and follower controllers to the ROS 2 ecosystem. It packages the bridge nodes, bring-up launch files, teleoperation utilities, and documentation for building datasets that support imitation learning research.

Overview

This documentation walks through the full workflow:

  • Preparing the Lerobot and Python environments required by the bridge.

  • Building the ROS 2 workspace and enabling camera pipelines.

  • Configuring the bridge parameters and launching the system end-to-end.

  • Capturing demonstrations and preparing them for imitation learning pipelines.

Packages

  • ros2_externals/system_data_recording – Nodes and launch files for capturing datasets for imitation learning using bag files.

  • so101_bringup – End-to-end bring-up launch files for the SO101 robot.

  • so101_controller – ROS 2 control launches and configurations.

  • so101_description – URDF, xacro, meshes, usd files for the robot + robot_state_publisher launch file.

  • so101_hardware_interface – Hardware interface for the SO101 robot.

  • so101_ros2 – Meta package for the SO101 ROS 2 workspace.

  • so101_ros2_bridge – Bridge nodes connecting Lerobot APIs to ROS 2 topics and services.

  • so101_teleop – Teleoperation component and launch file.

Acknowledgments

This project builds directly on the excellent Lerobot framework from Hugging Face.

Guides