A progression of MuJoCo environments and deep reinforcement-learning tools working toward humanoid tennis. Moves from simple ball control through racket-and-wall tasks to an experimental two-humanoid curriculum.

프로젝트 사이트 방문