UC Berkeley and Stanford Develop AI Framework for Advanced Humanoid Robot Motion

Researchers from UC Berkeley and Stanford University have introduced BeyondMimic, a reinforcement learning-based motion tracking framework that enables humanoid robots to learn complex human motions and adapt them to real hardware. This breakthrough allows robots to perform hundreds of actions, including aerial cartwheels and spinning kicks, under a unified training setup.