#43 · Primary category: Robotics & Embodied AI Simulation

motion_imitation

Code accompanying the paper "Learning Agile Robotic Locomotion Skills by Imitating Animals"

Project last updated:03/24/23

GitHub Stars

1.5K

Forks

314

Contributors

8

License

Apache-2.0

Why we included this project

Legged-locomotion researchers and engineers who want to see imitation learning in action will get more from this than from most demo repos. It ships a Gym environment for a simulated quadruped plus reference motion clips recorded from real animals, so you can train a policy to copy gaits like a dog's pace and then check it against pretrained models. There's also a whole-body model predictive control example that runs in PyBullet and on a real Unitree A1, which helps if you're planning the jump from simulation to hardware. Just be ready for research code: you'll be driving the training scripts and command-line flags yourself rather than clicking through a polished interface. If you're curious how motion-capture data can steer agile robot behavior, this is a solid reference to dig into.

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