3DThinkVLA: Endowing Vision-Language-Action Models with Latent 3D Priors via 3D-Thinking-Guided Co-training

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