Vision-Language-Action models provide a strong foundation for general-purpose robot control, yet a vast majority of policies do not preserve and leverage epi…
机构:NTU
来源:arXiv 2609.28256 | AI4Papers 论文推荐平台