Universal Pose Pretraining for Generalizable Vision-Language-Action Policies

作者：Haitao Lin, Hanyang Yu, Jingshun Huang, He Zhang, Yonggen Ling, Ping Tan, Xiangyang Xue, Yanwei Fu · 单位：Tencent Robotics X, The Hong Kong University of Science and Technology, Fudan University · 会议/期刊：arXiv · 日期：2026-02-23 · 来源：Low-Level Learning-Based Action Modelling / Input Modelling / 3D Vision Language Action Models