Humanoid robots learn dynamic parkour skills using motion matching and RL

Researchers have developed a framework called Perceptive Humanoid Parkour (PHP) that enables humanoid robots to perform dynamic parkour maneuvers. The system uses motion matching to compose human skills into complex trajectories and then trains reinforcement learning policies for these motions. This allows robots to autonomously navigate and interact with obstacles using only depth sensing and basic velocity commands. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Demonstrates advanced humanoid robot capabilities in dynamic environments, potentially influencing future robotics research and applications.

RANK_REASON This is a research paper detailing a new framework for humanoid robot locomotion. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Perceptive Humanoid Parkour
• PHP
• Unitree G1
• arXiv
• Lujie Yang

• paper
• other