AI & ML New Capability

RoboClaw introduces 'Entangled Action Pairs' to allow robots to autonomously collect data by learning to reset their own environment.

arXiv · March 13, 2026 · 2603.11558

Ruiying Li, Yunlang Zhou, YuYao Zhu, Kylin Chen, Jingyuan Wang, Sukai Wang, Kongtao Hu, Minhui Yu, Bowen Jiang, Zhan Su, Jiayao Ma, Xin He, Yongjian Shen, Yangyang, Guanghui Ren, Maoqing Yao, Wenhao Wang, Yao Mu

Why it matters

The massive bottleneck in robotic learning is human intervention for resets. By coupling forward manipulation with inverse recovery actions, RoboClaw enables continuous on-policy data acquisition, reducing human effort by 53% and improving long-horizon task success.

From the abstract

Vision-Language-Action (VLA) systems have shown strong potential for language-driven robotic manipulation. However, scaling them to long-horizon tasks remains challenging. Existing pipelines typically separate data collection, policy learning, and deployment, resulting in heavy reliance on manual environment resets and brittle multi-policy execution. We present RoboClaw, an agentic robotics framework that unifies data collection, policy learning, and task execution under a single VLM-driven cont