Recursive Harness Distillation lifts robot manipulation success from 37.3% to 64.0%
SeoulNatlUniv · hf · 2026-09-29
Researchers propose Recursive Harness Distillation: a strong agent distills intervention experience into a playbook, recursively refined via a lightweight agent's execution feedback — no parameter updates needed. Real-world manipulation success rises from 37.3% to 64.0%; on SimplerEnv Bridge, the light agent with playbook hits 66.7% vs 41.7% for GR00T-only, and the strong agent reaches 79.2%.
More from Embodied
- Wayve's autonomous driving goes live in London with Uber, lit up across the city — alexgkendall · 2026-09-29
- Tesla FSD Supervised approved in Croatia, its 8th European market — mitchdeg · 2026-09-29
- Idle time in demos teaches robots to do nothing — dev open-sources a LeRobot dataset editor — lhoestq · 2026-09-29
- Microagi hires Georgia Tech professor Animesh Garg plus 7 researchers via stealth acquisition — lukas_m_ziegler · 2026-09-29
- Tsinghua humanoid robot plays badminton with one policy from just 30 min of human motion data — ChongZzZhang · 2026-09-29
- Bose's $99 USB-C wired ANC earbuds ditch the battery brick by drawing power from your phone — aakashgupta · 2026-09-29