[ECCV 2026] VLA-JEPA:利用潜隐世界模型提升视觉-语言-动作模型性能
[ECCV 2026] VLA-JEPA: Enhancing Vision-Language-Action Model with Latent World Model