FlowPRO: Reward-Free Reinforced Fine-Tuning of Flow-Matching VLAs via Proximalized Preference Optimization

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