KingKongRobotics/jumper

🦀 Jumper — an crab robot.

What it solves

It provides an AI-powered toolkit for the Jumper 22-DoF crab robot, simplifying the complex process of designing the robot's physical appearance, training its movements, and creating simulated environments for it to operate in.

How it works

The toolkit allows users to use natural language prompts via an AI coding assistant to generate specific assets. It integrates with several technical frameworks to handle different tasks:

  • Appearance and Scenes: Uses the jumper-design resource to create .skin files for visuals and .map files for environments.
  • Motion Training: Leverages mjlab, rsl_rl, MuJoCo, and MuJoCo Warp to train motion policies (such as walking, jumping, or grasping), which are then packaged as .app files.

Who it’s for

It is designed for developers and robotics enthusiasts who own or are working with the Jumper crab robot and want to quickly iterate on its design and behavior without manually coding every movement or asset.

Highlights

  • Natural Language Design: Create robot skins and environment maps using simple text descriptions.
  • Motion Training: Support for training complex behaviors like tripod gaits, dancing, and grasping.
  • Simulation Integration: Built on industry-standard physics engines like MuJoCo.
  • Exportable Assets: Generates ready-to-use files for appearance (.skin), maps (.map), and motion policies (.app).

Written about in

Related

  • Project
    ace-trump-tech/MindPaw

    An open-source, low-cost desktop quadruped robot that integrates LLM-powered dialogue, gesture recognition, and 3D reconstruction for interactive AI companionship.

  • Project
    Roboparty/roboto_origin

    A fully open-source DIY humanoid robot project providing the hardware designs, RL training workflows, and ROS2 deployment framework needed to build a walking and running robot.

  • Project
    ARISE-Initiative/robosuite

    A MuJoCo-powered simulation framework and benchmark suite for robot learning, providing modular tools to create environments and test AI control algorithms.

  • Project
    open-gigaai/giga-brain-0

    GigaBrain‑0.7 is an open‑source, large‑scale vision‑language‑action model for robots. It is pretrained on 37 k h of heterogeneous embodied data, supports eight robot embodiments, and ships with training scripts, benchmark suites (VLM‑Benchmark, RoboColiseum, RoboTwin 2.0, EBench), and ready‑to‑run inference servers that integrate with ROS.