arXiv Robotics — research abstracts· Zijun Wang, Tianyi Yan, Haokun Lin, Panwen Hu, Bohang He, Siyu Ren, Xiaoyu Guo, Weijia Liufu, Shuoshuo Xue, Fengjiao Chen, Zipeng Ye, Xuezhi Cao, Zutao Jiang, Hanhui Li, Xiaodan Liang·· 6 days agoEditorial score65
ProWorld: Multi-View World Model with Proprioceptive Feedback for Closed-Loop Policy Evaluation
ProWorld: Multi-View World Model with Proprioceptive Feedback for Closed-Loop Policy Evaluation
Summary
ProWorld is a world model that jointly predicts multi-view video and robot state, enabling closed-loop policy evaluation. It incorporates proprioceptive feedback and failed-action outcomes to avoid overestimating policy success. The model achieves high correlation between generated and real success rates, supporting its use in policy evaluation frameworks.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
Source：arXiv Robotics — research abstracts · arxiv.org