Embodied Intelligence Observer

Motus2: A Self-Evolving General World Model for Dexterous Manipulation

Research

Source: arXiv cs.ROPublish time unverified

arXiv:2608.30237v1 Announce Type: new Abstract: General embodied agents should perceive, predict, act, evaluate, and improve within a unified system. World models have shown great promise in building such agents, yet existing models typically append an action output head to a world simulator, without coupling them into a closed decision-and-learning loop for policy improvement. We present Motus2, a self-evolving general world model for dexterous manipulation.

Motus2: A Self-Evolving General World Model for Dexterous Manipulation | Embodied Intelligence Observer