University of Tokyo Uses RL for 12 Giant Swings by Inverted Bipedal Robot
Researchers at the University of Tokyo have trained an inverted bipedal robot to complete 12 consecutive giant swings—full 360-degree rotations—on a real high-bar setup using reinforcement learning. The controller was developed to manage the demanding contact dynamics and continuous rotating motion rather than only steady balancing.
✓ Verified
2026-11-22
Source:立德共创 Leaderobot ↗