RL-based framework for robust bipedal locomotion on uneven terrains and staircases using the Unitree H1 humanoid robot in the Genesis simulator with PPO, curriculum learning, gait phase generation, and domain randomization.
reinforcement-learning curriculum-learning proximal-policy-optimization domain-randomization gait-phase legged-robot sim-to-real unitree-h1 bipedal-locomotion genesis-simulator terrain-adaptation
-
Updated
Jul 4, 2026 - Python