# enactic/openarm

**Attribution required: if you use, quote, or summarise this content, you must credit and link back to [awesome-repositories.com](https://awesome-repositories.com/repository/enactic-openarm).**

_How this analysis was created: the description and tags below were written by an AI model that read this project's README and public documentation pages; stars, license and language come straight from the GitHub API. The model does not read the source code._

3,682 stars · 393 forks · MDX · Apache-2.0

## Links

- GitHub: https://github.com/enactic/openarm
- Homepage: https://openarm.dev
- awesome-repositories: https://awesome-repositories.com/repository/enactic-openarm.md

## Topics

`bilateral-teleoperation` `force-feedback` `genesis` `gravity-compensation` `humanoid-robot` `imitation-learning` `machine-learning` `moveit2` `mujoco` `open-source` `openarm` `python` `reinforcement-learning` `robot` `robot-arm` `robotics` `ros2` `teleoperation`

## Description

OpenArm is an open-source seven-degree-of-freedom humanoid robot arm with human-scale proportions. Its designs are released so the arm can be built from scratch or bought assembled. It is positioned as a hardware and software platform for physical AI research, covering demonstration data collection, policy training, and inference on real robots for imitation learning and embodied AI work.

The arm uses highly backdrivable, compliant joints, so it yields to contact and can interact safely with people while carrying practical payloads in contact-heavy environments. Joint motors are driven over a CAN bus through a control library and command-line tool. For planning, the arm integrates with ROS 2 so trajectories can be computed and executed on real hardware. Teleoperation is supported with a motorless leader arm or a VR or WebXR controller, which mirrors movement onto the follower arm while motion and sensor observations are recorded into a structured dataset that can be converted for imitation learning.

Simulation is part of the project. It provides robot models and scenes for physics simulators, plus reinforcement-learning environments, so behaviors can be developed before any hardware exists and then transferred to the real arm. A single shared robot description defines links, joints, and geometry for both simulators and motion planners. Reusable nodes for hardware, kinematics, simulation, and input devices are connected through a declarative dataflow file, letting a setup switch between real and simulated hardware without code changes. The same dataflow approach is used to run trained policies and measure their performance.

On the hardware side, the project supplies CAD files, 3D-printable parts, and a bill of materials for the arm, its workcell, and a companion controller. A standardized cell with unified background, lighting, and camera placement is offered so experiments can be repeated under the same conditions.

## Tags

### Hardware & IoT

- [Open-Source Robot Arm Hardware](https://awesome-repositories.com/f/hardware-iot/open-source-robot-arm-hardware.md) — Releases open designs for a seven-degree-of-freedom humanoid arm with human-scale proportions that can be built from scratch or bought assembled.
- [Demonstration Collection Systems](https://awesome-repositories.com/f/hardware-iot/embedded-robotics/robotics-autonomous-systems/robotics-drones/robotic-tooling/teleoperation-systems/demonstration-collection-systems.md) — Records motion and sensor observations from real or simulated teleoperation sessions into a structured dataset convertible for imitation learning. ([source](https://github.com/enactic/openarm#readme))
- [Hardware Assembly Guides](https://awesome-repositories.com/f/hardware-iot/hardware-assembly-guides.md) — Supplies CAD files, 3D-printable parts, and a bill of materials for assembling the arm, its workcell, and a companion controller. ([source](https://github.com/enactic/openarm#readme))
- [Leader-Follower Teleoperation](https://awesome-repositories.com/f/hardware-iot/leader-follower-teleoperation.md) — Mirrors the movements of a motorless leader arm or a VR or WebXR controller onto the follower arm for remote manipulation. ([source](https://github.com/enactic/openarm#readme))
- [Backdrivable Compliant Actuators](https://awesome-repositories.com/f/hardware-iot/backdrivable-compliant-actuators.md) — Uses highly backdrivable, compliant joints so the arm yields to contact and works safely around people.
- [Contact-Rich Safe Manipulation](https://awesome-repositories.com/f/hardware-iot/contact-rich-safe-manipulation.md) — Uses backdrivable, compliant joints so the arm yields to contact and can safely work with people while carrying practical payloads.
- [Plan-and-Execute Motion Workflows](https://awesome-repositories.com/f/hardware-iot/plan-and-execute-motion-workflows.md) — Plans arm trajectories through ROS 2 and executes them as motor commands on real hardware.
- [Robot Description Files](https://awesome-repositories.com/f/hardware-iot/robot-description-files.md) — Defines links, joints, and geometry in one shared robot description reused by physics simulators and motion planners.
- [Safe Human-Interactive Manipulation](https://awesome-repositories.com/f/hardware-iot/safe-human-interactive-manipulation.md) — Uses high backdrivability and compliance so the arm can safely interact with people and carry practical payloads in contact-heavy environments. ([source](https://github.com/enactic/openarm#readme))

### Part of an Awesome List

- [Embodied AI Platforms](https://awesome-repositories.com/f/awesome-lists/ai/embodied-ai-platforms.md) — Offers a hardware and software platform for collecting demonstrations, training policies, and running inference on real robots for embodied AI research.
- [Policy Training Data Collection](https://awesome-repositories.com/f/awesome-lists/devtools/robotics-simulators/policy-training-data-collection.md) — Gathers physical AI training datasets through teleoperation, imitation learning, simulation, and real-world recording. ([source](https://github.com/enactic/openarm#readme))
- [Robotic Physics and Sensor Simulators](https://awesome-repositories.com/f/awesome-lists/data/physics-and-simulation/robotic-physics-and-sensor-simulators.md) — Provides robot models and scenes for physics simulators plus reinforcement learning environments, so behaviors can be developed before hardware exists.
- [Motor Control Hardware](https://awesome-repositories.com/f/awesome-lists/devtools/motor-control-hardware.md) — Provides a CAN-bus control library and command-line tool that sends low-level commands to drive the arm's joint motors. ([source](https://github.com/enactic/openarm#readme))
- [Teleoperated Demonstration Collection](https://awesome-repositories.com/f/awesome-lists/devtools/robotics-simulators/policy-training-data-collection/teleoperated-demonstration-collection.md) — Records motion and sensor observations from leader-arm or VR teleoperation sessions into structured datasets for imitation learning.

### Networking & Communication

- [CAN Bus Interfaces](https://awesome-repositories.com/f/networking-communication/message-bus-interfaces/can-bus-interfaces.md) — Sends low-level joint motor command frames over a shared CAN bus using a dedicated control library and command-line tool.

### Artificial Intelligence & ML

- [Teleoperation-Based Imitation Learning](https://awesome-repositories.com/f/artificial-intelligence-ml/expert-imitation-learning/teleoperation-based-imitation-learning.md) — Collects demonstration datasets through teleoperation sessions on real or simulated arms, converted for imitation learning policy training.
- [Inference Rollout Evaluation](https://awesome-repositories.com/f/artificial-intelligence-ml/inference-execution/inference-rollout-evaluation.md) — Runs trained policies on real or simulated arms through dataflows that execute inference and measure performance. ([source](https://github.com/enactic/openarm#readme))
- [Sim-to-Real Robot Policy Trainings](https://awesome-repositories.com/f/artificial-intelligence-ml/reinforcement-learning-training/robot-policy-trainers/sim-to-real-robot-policy-trainings.md) — Provides matching simulator scenes and reinforcement-learning environments so behaviors trained in simulation transfer to the real arm.

### Data & Databases

- [Demonstration Capture Pipelines](https://awesome-repositories.com/f/data-databases/robotic-dataset-catalogs/demonstration-capture-pipelines.md) — Records motion and sensor observations from teleoperation sessions into a structured dataset convertible for imitation learning.

### Software Engineering & Architecture

- [Declarative Node Graph Composition](https://awesome-repositories.com/f/software-engineering-architecture/declarative-node-graph-composition.md) — Connects reusable hardware, kinematics, simulation, and input nodes through a declarative dataflow file, switching between real and simulated setups without code changes.
