Stanford HomeBody drops the VLA: GPT Astra directly orchestrates Unitree G1 humanoid skills

CyberRobooo · x · 2026-09-28

Stanford's HomeBody tests a different path to humanoid intelligence: a Unitree G1 explores an unfamiliar kitchen, builds persistent spatial memory and a Real2Sim digital twin, then uses GPT Astra to plan and compose skills like Navigate, Pick, Place and Open Drawer — no environment-specific training data or extra policy learning.

The open question: as frontier models get better at spatial reasoning and planning, how much of the robot stack still needs custom building?

Related event: Stanford Open-Sources HomeBody: VLM-Driven Humanoid Robots Tackle Unfamiliar Homes(4 posts)→

Original post →

More from Embodied

Embodied channel →