Reinforcement Learning

Single-leg stance looks simpler than acrobatics, yet it requires a robot to track not only where its center of mass is but where it is moving. We examine DDC’s 89-of-90 simulation result and the important limits of its qualitative Unitree G1 demonstration.