Berkeley's Do as I Do turns everyday RGB videos into dexterous robot hand data
JitendraMalikCV · x · 2026-09-20
UC Berkeley's Jitendra Malik, with Pieter Abbeel and Mahi Shafiullah, presents "Do as I Do": an algorithm that reconstructs hand-object interactions from in-the-wild monocular RGB human videos and retargets them into executable actions for multi-fingered dexterous robot hands, outperforming prior SOTA and yielding a data-collection playbook. Paper and code are released; the reply thread notes such pipelines once took teams two years to build.
Related event: Astra hand-object reconstruction sparks clarification from researchers(3 posts)→
More from Embodied
- QUANTA X1 Pro bimanual robot starts real warehouse work at Lululemon's Wuhan DC — chris_j_paxton · 2026-09-21
- jev-use: open-source Mac voice control reads the Accessibility tree, no screenshots, claimed 100x faster than LLMs — _AustinCalvert_ · 2026-09-21
- Physical AI data layer Mecka partners with Whop to pay people for robot training footage — eptwts · 2026-09-21
- Eight cutting heads blind-harvest celery in one of the hardest specialty crop automation feats — anselm · 2026-09-21
- Jitendra Malik: robots ignoring 3D structure are wasting a valuable signal — JitendraMalikCV · 2026-09-21
- Eric Topol: wearable HRV and readiness scores lack evidence of health benefits — EricTopol · 2026-09-21