RL robot running form hailed as potential 'Move 37' of biomechanics

rohanpaul_ai · x · 2026-08-30

A post showcases a video of a robot mimicking human running form, comparing it to AlphaGo's 'Move 37' and suggesting it could be a milestone in biomechanics. Comments explain that reinforcement learning helped the robot maintain a posture superior to human form: humans swing arms to counter torso rotation from leg strides, while robots using motors and gears generate high torque; the forward lean shifts the center of mass, reducing effort to stay upright and increasing forward motion efficiency.

Related event: Reinforcement Learning Enables Humanoid Robots to Run Like Humans(2 posts)→

Original post →

More from Embodied

Embodied channel →