Skild AI's S1 Robot Model Learns New Tasks from a Single Demo
deepakpathak · x · 2026-08-27
Skild AI has introduced S1, a general-purpose robotics foundation model demonstrating strong in-context learning capabilities. S1 can execute unseen tasks on real hardware (e.g., pancake flipping, kit assembly, pour-over coffee) after watching just one human demo video (10 min task) without any fine-tuning. Experiments show that with 100K hours of pre-training, S1 achieves an average per-step success rate of 66% on unseen tasks, vastly outperforming language-prompted VLA models (9%). A single video prompt is reportedly worth roughly 380 post-training demonstrations, enabling zero-shot generalization and significantly lowering the barrier for robots to acquire new skills.
More from Embodied
- Viral meme nails AI's irony: it can write symphonies but can't do the dishes — briannekimmel · 2026-10-16
- WEDGE turns navigation agents' own failures into targeted training worlds — mohitban47 · 2026-10-16
- Knödel Watch: animated vocabulary cards on ESP32-S3, open-sourced 4 days after unboxing — alexcovo_eth · 2026-10-16
- SlipSense fuses pressure and vibration for 23ms zero-shot robotic slip detection — micoolcho · 2026-10-16
- Stanford's Fei-Fei Li team open-sources OpenWAM, a unified world-action model framework for robots — jiqizhixin · 2026-10-16
- Robot folding leaderboard: GPT-6 Astra leads at 1349 Elo, Muse Wonder trails at 1034 — qinzytech · 2026-10-16