100,000 hours of human hands doing real work. LeRobot format, fully annotated and open.
EgoSuite-Open100K from @LightwheelAI X @huggingface: 15,000+ tasks across 15,000+ real scenes, hand + body pose and wrist cams.
Recorded across 7 environment categories and 128 scene types, from kitchens and bedrooms to warehouse floors and assembly lines.
It's open for academic research and commercial training, and streams straight from the Hub. Always great to see a dataset at this scale released openly on the Hub.
More info: https://t.co/1XNb4sWLL5
FlashSAC at @holidayrobots just got a Best Paper Award at #RSS2026 🎉
One thing that stood out at #RSS2026 was how widely RSL-RL & PPO are used for sim-to-real RL. So we’re open-sourcing rsl_rl_flashsac, a FlashSAC implementation built on top of RSL-RL.
If you're already on RSL-RL, you can try FlashSAC without changing your existing stack. We're also working on Newton support and sim2sim2real deployment on the Unitree G1. More to come!
Repo: https://t.co/xCgWEtaAuK
Egocentric data has a brutal failure mode: you can record hours of footage before realizing the robot-relevant part was never captured properly.
EGO1GS moves that check into the headset itself. Its EgoHand model knows whether the hands in frame actually belong to the wearer, hitting 0.9925 AUC, 99.6% recall and 97.8% precision.