作者：Yiming Mao, Zixi Yu, Weixin Mao, Yinhao Li, Qirui Hu, Zihan Lan, Minzhao Zhu, Hua Chen, LimX Dynamics · 单位：Beijing University of Posts and Telecommunications, Zhejiang University · 会议/期刊：arXiv · 日期：2026-04-03 · 来源：High-Level Structured Planning / Multimodal Reasoning / Reward Reasoning
