Skild AI Taps NVIDIA Physical AI to Teach Robots New Tasks From a Single Video
Skild AI's S1 learns new robot tasks from a single video — a step toward adaptive manufacturing that doesn't require reprogramming.
Why it matters
A foundation model for physical robotics that generalizes across unseen long-horizon tasks from minimal demonstration data represents a meaningful capability advance in embodied AI — with direct implications for how practitioners think about robot deployment flexibility and retraining costs.
The key facts
10 to knowSkild AI S1: robot foundation model learns long-horizon tasks from single video
Targets manufacturing, warehousing, production-line adaptability
Built on NVIDIA Physical AI
Launched week of Sept 10, 2026
Key capability: generalizes to previously unseen tasks without reprogramming
Skild AI launches S1 robot foundation model
Model learns previously unseen, long-horizon tasks from single video demonstration
Targets manufacturing, warehousing, and production environments with variable layouts and changing tasks
Built on NVIDIA Physical AI framework
Launched week of September 10, 2026
Go to the source
NVIDIA Blogblogs.nvidia.com
Publisher excerpt: Manufacturing floors, warehouses and production lines rarely stay fixed — tasks change, layouts shift and new products arrive, and most robots can’t keep up without significant reprogramming. Skild AI’s new S1 robot foundation model helps address this, designed to learn previously unseen,…