Skild AI Post-Trains Humanoid S1 via Self-Play: 140 Sim Years of Soccer, No Human Demos
Scobleizer · x · 2026-09-24
Skild AI brings AlphaGo-style self-play to humanoids: its robotics foundation model S1, given only the objective of scoring goals, trained over 140 simulated years in NVIDIA Isaac Sim and went from falling over to dribbling, shielding, tackling, shooting, and getting back up—with no human demonstrations. Behaviors emerged from competing against past versions of itself, forming an automatic curriculum.
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