GenRobot.AI releases Gen-HumanWholeBody: 600 hours of whole-body manipulation data with 1,281 human skills
AdinaYakup · x · 2026-10-23
GenRobot.AI has released Gen-HumanWholeBody, a high-precision whole-body manipulation dataset for Physical AI research on Hugging Face.
- Scale: 600 hours of real-world data, 15,943 episodes, 16,400 unique tasks across 26 domains, and 1,281 distinct human skills
- Capture: recorded via lightweight DAS-Ego egocentric devices and reconstructed with the company's proprietary Data Foundation Model, with 3 cm average joint position error
- Format: 6 native RGB views (1600×1300 @ 30 FPS) plus body/hand tracking in mcap format, covering two-handed manipulation, locomotion, and whole-body coordination
The team positions it as training-ready whole-body manipulation data and a template for scalable data pipelines built on lightweight egocentric hardware.
Related event: GenRobot.AI Open-Sources 600-Hour Whole-Body Manipulation Dataset(2 posts)→
More from Embodied
- Inside a Shenzhen robot-arm factory: 2,000 arms/month at ~$15k, AGIBot bought ~20,000 — DominiqueCAPaul · 2026-10-23
- Dexmal's open-source VLA model DM0.5 lands in LeRobot, tops RoboColiseum — AdinaYakup · 2026-10-23
- Guy uses Ray-Ban Meta Smart Glasses, AI and a MacBook to cheat at chess — CodeByPoonam · 2026-10-23
- A classic world model (s, a → s′) acting as a neural MuJoCo simulator — yshan2u · 2026-10-23
- Gen-HumanWholeBody: open 600-hour whole-body manipulation dataset for robots — AdinaYakup · 2026-10-23
- Stanford releases free 520-page robot autonomy course book and lectures — lukas_m_ziegler · 2026-10-23