Roboparty/RP1

RP1: An OpenSource Full-Stack Platform for Agentic Humanoids

What it solves

Building a humanoid robot typically involves many separate pieces: mechanical parts, electronics, low-level firmware, control software, and AI training. These are often closed or scattered. This project provides everything in one open-source stack so you can understand, reproduce, and extend a real humanoid robot.

How it works

The platform is organized into several repositories. Hardware repositories contain CAD files, schematics, PCB layouts, BOMs, and assembly instructions. Software repositories include Linux firmware, communication bridges, robot model files (URDF/MJCF), and deployment SDKs with modules for IMU, motor, inference, and camera accessible from C++ and Python. Learning repositories provide training workspaces (e.g., Isaac Lab) for locomotion policies, motion imitation, and parkour, using reinforcement learning methods like AMP and BeyondMimic. The architecture follows a pipeline: perception/input -> motion control (retargeting, policies) -> real-time control (500 Hz action decoding) -> whole-body execution with joint modules and a dexterous hand. An E-stop and high-frequency feedback loop are included.

Who it’s for

Researchers, developers, and hobbyists who want to build, understand, or program a humanoid robot. It's also for robotics students and AI engineers who want to experiment with learning-based control on a real platform.

Highlights

  • Fully open source: mechanical, electrical, firmware, control, and learning code all available.
  • Modular repositories for hardware, software, and AI training.
  • Includes joint modules (ROMOMO), a dexterous hand (RP Hand), and deployment SDKs.
  • Supports multiple configurations (24, 28, 30, 42 DoF).
  • Provides learning demos for natural locomotion (AMP), expressive motion imitation (BeyondMimic), and parkour (coming soon).
  • Community channels (Discord, QQ, Telegram, etc.) for collaboration.

Related

  • Project
    althack/vscode_ros2_workspace

    A template that provides a consistent, containerized ROS2 development environment for VSCode, ensuring your local workspace and CI pipeline use the same configuration.

  • Project
    nicklashansen/tdmpc2

    A scalable, robust model-based reinforcement learning algorithm for continuous control that can handle multiple tasks across different domains and embodiments using a single agent.

  • Project
    swc-17/SparseDrive

    SparseDrive is an end-to-end autonomous driving framework that uses sparse scene representation to unify perception and planning, significantly improving inference speed and reducing collision rates.

  • Project
    RussTedrake/underactuated

    A software library providing algorithms for the control of underactuated robots, supporting dynamic movements like walking, running, swimming, and flying.