#34 · Primary category: Robotics & Embodied AI Simulation

PBHC

Official Implementation of "KungfuBot: Physics-Based Humanoid Whole-Body Control for Learning Highly-Dynamic Skills"

Project last updated:07/01/26

GitHub Stars

1.1K

Forks

141

Contributors

5

License

Other

Why we included this project

Whole-body control is one of the harder problems in humanoid robotics, and this repo is the official code behind the KungfuBot papers, which train a simulated humanoid to pull off highly dynamic moves like kungfu-style motions and horse-stance poses using physics-based reinforcement learning. The code ships pretrained policies for some of the motions, so you can run evaluation without retraining everything from scratch, and it is laid out as a policy training pipeline with a humanoidverse environment layer on top. If you are working on legged locomotion or comparing whole-body control baselines, this is a solid codebase to study and adapt.

Articles for this project

No articles for this project yet.

To suggest a topic or contribute an article, contact us.

Related projects in this category