roboterax/humanoid-gym
A reinforcement learning framework for training humanoid robot locomotion skills with zero-shot sim-to-real transfer.
Velocity · 7d
+2.4
★ / day
Trend
→steady
star history
Humanoid-Gym is an RL training framework built on Nvidia Isaac Gym for developing locomotion skills in humanoid robots, emphasizing zero-shot transfer from simulation to real-world deployment. It provides a sim-to-sim pipeline between Isaac Gym and Mujoco for policy verification, and has been validated on RobotEra’s XBot-S and XBot-L humanoid robots in real-world environments.