Vision-language-action and world-action models have demonstrated impressive capabilities in robotics, yet generalization to unseen tasks remains challenging.…
机构:复旦
来源:arXiv 2609.37810 | AI4Papers 论文推荐平台