Enactic has released OpenArm, an open-source seven-degree-of-freedom humanoid arm for physical AI research in contact-rich settings, according to the project’s GitHub repository. The company says the arm is highly backdrivable and compliant, built for safe human-robot interaction while carrying a usable payload.
A complete bimanual system, two arms, costs $6,500 and ships assembled or as a DIY kit through listed manufacturers. Enactic pitches it at four jobs: collecting data in the real world, running simulations, imitation learning and teleoperation.
The release bundles nine repositories under Apache 2.0 and CERN open-hardware licenses: CAD files, ROS2 integration, CAN motor control, and MuJoCo and Isaac Lab simulation support. Enactic also ships “OpenArm Cell,” a standardized rig with fixed lighting and camera placement meant to let labs reproduce each other’s results, a persistent gap in robotics research where hardware variance makes published results hard to compare.
That reproducibility framing, not the arm itself, is the more consequential part of this release. A cheap, fully open reference platform gives smaller labs a shared baseline for manipulation research that has until now demanded custom rigs. Enactic says it is recruiting contributors and research partners.
Based on Enactic’s OpenArm project page on GitHub.
