Robot learns to walk on 1K+ moving cylinders via 10M RL steps
hbouammar · x · 2026-09-21
A reinforcement learning demo where a quadruped robot learns to walk on a shifting ground made of 1,000+ moving cylinders in MuJoCo, forcing the policy to adapt in real time.
Training took roughly 10 million steps with the terrain maze constantly changing; the robot survived several near-falls before converging to a stable policy. The video shows real policy rollouts, not animation.
More from Embodied
- Scoble bets $1,000 Musk will never add lidar to FSD — Scobleizer · 2026-09-21
- Humanoid Bodies Are Impressive, Brains Are the Bottleneck — Spirit AI Bets on 2027 — TansuYegen · 2026-09-21
- The catapult analogy: why AI models are jagged and robots face a deployment gap — lateinteraction · 2026-09-21
- QUANTA X1 Pro bimanual robot starts real warehouse work at Lululemon's Wuhan DC — chris_j_paxton · 2026-09-21
- jev-use: open-source Mac voice control reads the Accessibility tree, no screenshots, claimed 100x faster than LLMs — _AustinCalvert_ · 2026-09-21
- Physical AI data layer Mecka partners with Whop to pay people for robot training footage — eptwts · 2026-09-21