openarm

A fully open-source humanoid arm for physical AI research and deployment in contact-rich environments.

PathOn Robotics Open Source Repo on building and controlling real robots

OpenARM is an open-source remake version 7DOF humanoid arm hardware designed for physical AI research and deployment in many environments. With high backdrivability and compliance with cheap Chinese joint motor standards, it excels at safe human-robot.

A tested control, planning, perception, bimanual, and learning platform for the Enactic OpenArm v2 (7-DOF x2) in MuJoCo: IK / Cartesian / compliant control, RRT-Connect planning, dynamic catching and throwing, articulated / cloth manipulation, language-commanded skills, a learned ACT vision policy, RL insertion, and OpenArm-Bench.

Pick & place manipulation with OpenArm robot trained via PPO in Isaac Lab

End-effector pose tools and experiments for OpenArm.

OpenArm-related C++ control and integration workspace.

Real-time teleop_xr-to-CAN bridge for a 14-DOF OpenArm robot

LeRobot teleoperator plugin for the OpenArm KER leader arm

Physics-based OpenArm 2.0 bimanual parcel orientation, handoff, and roller conveyor simulation in MuJoCo.

Minimal macOS handoff for two OpenArm Mini leaders controlling bimanual OpenArm V2 in MuJoCo with verified gripper haptics

Isaac Lab extension for the OpenArm manipulator, providing uni-/bimanual tasks for IL/RL.

OpenArm2 reinforcement learning, simulated vision telemetry and VR teleoperation with NVIDIA Isaac Lab

Per-unit angle calibration for the OpenArm KER encoder module — firmware, station procedure and host tool. Wire-compatible with an unmodified M5.

Bimanual OpenArm in Isaac Sim: position-based picks, right-to-left handover, multimodal demo recording — toward teleop, pi0.5 policy integration and evaluation (actively developed)

Mock Open Arm V2 pipeline using locally generated camera stream/joint values. Designed to generate various episodes to hold data