Summary

The model combines language reasoning, visual-spatial understanding and whole-body action generation. The company reports scaling experiments across human data, Unitree G1 data and its LightBot platform, alongside public weights, code and a text-to-action preview model.

Why it matters

Using large-scale human action data as a transferable pretraining signal could reduce dependence on expensive robot-specific demonstrations, although the reported evidence remains largely vendor-provided.

Sources