OpenArm Open-Sources 7-DOF Humanoid Arm: Embodied AI's 'Linux Moment'

OpenArm: Fully Open-Source 7-DOF Humanoid Robotic Arm

The Tokyo robotics team Enactic_ai has released OpenArm (OpenSource Artificial Robotics Machines), a complete open-source platform for a 7-degree-of-freedom humanoid robotic arm. The project publishes everything needed to build and control the arm: 3D-printable CAD manufacturing drawings, a bill of materials (BOM), low-level motor firmware, and dynamics control code via a ROS2 driver. The main repository is hosted at https://github.com/enactic/openarm, containing openarm_hardware for hardware designs and openarm_ros2 for the ROS2 driver.

Key Technical Features

1. Intrinsic Safety Through High Backdrivability

Unlike traditional industrial arms with high gear ratios and rigid structures that pose pinch hazards and can damage gearboxes on collision, OpenArm is designed for contact-rich environments. It achieves high backdrivability, allowing the arm to be compliant and safe during physical interaction. Under real-time gravity compensation, the arm feels light and can be manually guided by hand for trajectory demonstration; when encountering physical obstacles it automatically yields and dissipates force.

2. Zero-Latency 1:1 Master-Slave Force-Feedback Teleoperation (KER Architecture)

The project also open-sources a lightweight leader arm (Leader Arm). An operator manipulates the leader arm, and the follower arm follows with millisecond-level latency. Crucially, the real physical resistance encountered during grasping is instantly fed back to the operator's hand, enabling intuitive teleoperation with haptic feedback.

3. True Sim-to-Real Closed Loop

OpenArm provides deep integration with MuJoCo and NVIDIA Isaac Sim / Isaac Lab. A zero-install web-based simulation using WebAssembly is available at https://enactic.github.io/openarm_mujoco/, allowing immediate browser-based testing with sliders or keyboard. Policies trained in simulation via reinforcement learning can be deployed directly to the physical motors without modification. The platform also supports fine-tuning and deployment of NVIDIA's Isaac-GR00T foundation model.

Accessibility and Cost Reduction

Historically, a dual-arm system capable of advanced force-control experiments cost tens of thousands of dollars. OpenArm reduces this barrier by providing full open-source designs that can be 3D-printed and assembled by individual developers or university labs. Certified vendors also offer pre-made kits or complete units. The article draws a parallel to the RepRap project that democratized 3D printing, arguing that OpenArm lets embodied AI researchers focus on algorithm development rather than being blocked by expensive, closed hardware.

Resources

Official website: https://openarm.dev/ GitHub repository: https://github.com/enactic/openarm Web-based MuJoCo simulation (WASM):

https://enactic.github.io/openarm_mujoco/