Roboparty/roboto_origin

What it solves

It provides a fully open-source blueprint for building a DIY humanoid robot capable of walking and running. The project lowers the barrier to entry for humanoid robotics by providing all necessary hardware designs, electronics, and software for training and deployment.

How it works

ROBOTO_ORIGIN is a comprehensive ecosystem divided into several specialized modules:

• Hardware & Design: Includes CAD drawings, PCB designs, and a Bill of Materials (BOM) for assembly via common procurement channels.
• Simulation & Training: Uses IsaacLab for reinforcement learning (RL) training and supports Sim2Sim transfer to MuJoCo to develop motion control.
• Deployment: A ROS2-based framework handles the middleware, IMU/motor integration, and real-world execution.
• Firmware: Provides embedded software for USB2CAN communication and system management on platforms like OrangePi/RDK.
• Applications: Includes experimental modules for navigation, localization, and XR teleoperation.

Who it’s for

Robotics enthusiasts, developers, and researchers who want to build and program their own humanoid robot using open-source tools and accessible hardware.

Highlights

• Complete DIY Pipeline: Covers everything from mechanical structure and electronics to training and deployment code.
• RL-Driven Motion: Leverages IsaacLab for training the robot to walk and run.
• Modular Architecture: Organized into distinct repositories for hardware, firmware, deployment, and training for easier contribution.
• XR Integration: Supports experimental teleoperation via XR.