AtomVLA: Scalable Post-Training for Robotic Manipulation via Predictive Latent World Models

返回详情 VLA / Vision-Language-Action 每日论文卡