Ex-Google PhD Fellow's robot startup bets on full-modal human data over video learning
快鲤鱼 · wechat · 2026-10-14
Poxiao Intelligence, founded by 28-year-old HIT professor Yang Shuo, launched SeePhysHub — a wearable rig capturing first-person vision, hand touch, force and full-body pose — to train native full-modal world models from low-cost human data, claiming an order-of-magnitude reduction in robot task data needs versus vision-only approaches.
Related event: SeePhysHub Adds Full-Body Motion and Touch to Egocentric Robot Data(2 posts)→
More from Embodied
- Ex-Boston Dynamics team launches Hugo, a companion robot running the first VLE model — Scobleizer · 2026-10-17
- Prototype combines Atlas API and 8th Wall to turn one photo into a walkable AR scene — XRarchitect · 2026-10-16
- Tesla's Cybertruck kickflip video: jet controls right the truck midair like a cat — astralmatrix · 2026-10-16
- Sentdex trains a fall-recovery policy for the Asimov humanoid: getting up is hard — Sentdex · 2026-10-16
- LeRobot adds VLM-powered feature that turns plain-language goals into robot skills — Gradio · 2026-10-16
- Optimus Robot Shows Off Its Driving in a Tesla Roadster — debreuil · 2026-10-16