NVIDIA's ADEPT Pre-trains Dexterity, Cuts Robot Task Learning Cost by 3x
KyleMorgenstein · x · 2026-08-22
NVIDIA introduced ADEPT, a pre-training and post-training paradigm that treats dexterity as a prior rather than a per-task cost. The system learns reach-grasp-reorient-transport once in simulation using RL, then post-trains specialists. These policies achieve zero-shot deployment on real robot hands using pixels and touch, requiring only 3B steps per new task compared to 9B from scratch.
Related event: NVIDIA Unveils ADEPT: Dexterity Pretraining Speeds Up Robot Learning by 3x(3 posts)→
More from Embodied
- Running Minimax H3 locally on RTX 3070: Bringing history to life — 12padams · 2026-08-22
- Robot demo shows utility via good harness design and co-trained models — Kangwook_Lee · 2026-08-22
- Lumi robot autonomously navigates crowds, takes elevator, and dances — Scobleizer · 2026-08-22
- Physical AI targets a $40T opportunity, 7x larger than Digital AI — annbordetsky · 2026-08-22
- Is this the GPT moment for Robotics? New demo sparks discussion — dolo937 · 2026-08-22
- Blind user praises Tesla Cybercab for Braille support and lack of discrimination — EricETesla · 2026-08-22