Skild AI Trains Soccer Robot via 140 Years of Simulated Self-Play, No Human Demos
deepakpathak · x · 2026-09-24
Skild AI's "Messinator" learned to dribble past defenders, shield the ball and tackle purely through self-play — 140+ years of simulated soccer with only goals rewarded — before transferring to a real humanoid. Ilya Lehmann notes self-play is the same trick behind AlphaGo, but far harder with bodies: a humanoid must set every joint angle 50 times a second, making the move space effectively infinite. Still, he argues this is likely the path to superhuman robots.
More from Embodied
- Guava distills GPT-5.4 into a 4B robot agent with just 2,000 simulated trajectories — jbhuang0604 · 2026-11-09
- Dex-One2Many learns dexterous manipulation from one human video, 95%+ success across 4 hands — jbhuang0604 · 2026-11-09
- Fly-by-Code: coding agents pilot drones via active visual and physical feedback loops — jbhuang0604 · 2026-11-09
- μ0 world model learns robot physics from video only, beats π0 by 5 points on RoboCasa365 — jbhuang0604 · 2026-11-09
- NVIDIA unveils GR00T 2 open-world robot action model, hitting 95% success with 5 demos — CyberRobooo · 2026-11-09
- CoRL keynote to explore open-source data and hardware for robotics — micoolcho · 2026-11-09