Motus2 world model hits 84% success on dexterous tasks with self-evolving closed loop
burny_tech · x · 2026-09-08
Motus2 is a self-evolving general world model for dexterous manipulation that closes the loop between prediction, evaluation, and policy improvement.
- A single shared-weight model serves as policy, simulator, and value model: it proposes action chunks, imagines their visual consequences, and scores them, improving via model-based RL
- Failed and suboptimal interactions feed dynamics and value learning
- Data scaling: 130K hours of egocentric pretraining, moving to stereo egocentric data plus robot-domain adaptation
- Adds tactile feedback and global-autoregressive/hybrid-memory context extensions, deployed on a biomimetic robot platform
It reaches 84% average success across five dexterous tasks, with MBRL and test-time planning further boosting performance.
More from Embodied
- 360CityArena at ECCV 2026: best AI agent scores 17.1% vs humans' 77.3% on city navigation — CSProfKGD · 2026-09-08
- Open-source C-based LLM engine runs Gemma voice conversations on Jetson Orin, faster than llama.cpp — cortexist · 2026-09-08
- Robotics veteran Jitendra Malik challenges LLM robotics hype with a dexterity benchmark — JitendraMalikCV · 2026-09-08
- Self-driving demo runs on nothing but logic gates, moves like an insect — burny_tech · 2026-09-08
- Xpeng announces first automated production line where robots mass-produce robots — Anxious-Yoghurt-9207 · 2026-09-08
- Robotics researcher: post-training and recursive improvement are the next bottleneck — animesh_garg · 2026-09-08