General embodied intelligence
is built on the compression
and generalization of human behavior.

Pretrained on human actions captured in videos, Light-O1 learns a transferable human action prior for scalable whole-body robot intelligence.

Whole-Body In-Context Learning helps humanoids adapt to changes in physical capability, recover from falls, and keep moving despite hardware damage.

Our first general-purpose navigation brain, demonstrates zero-shot generalization across diverse robot embodiments, tasks, and scenes.