Reinforcement learning (RL) post-training improves robot foundation models (RFMs) beyond their demonstrations, but real-world deployment demands more: learni…
机构:NVIDIA
来源:arXiv 2610.18596 | AI4Papers 论文推荐平台