When would Vision-Proprioception Policies Fail in Robotic Manipulation?

作者：Jingxian Lu, Wenke Xia, Yuxuan Wu, Zhiwu Lu, Di Hu · 单位：Gaoling School of Artificial Intelligence, Renmin University of China, Beijing Key Laboratory of Research · 会议/期刊：arXiv · 日期：2026-02-12 · 来源：Low-Level Learning-Based Action Modelling / Input Modelling / Vision Action Models / Image-Action Models