Phoenix: A Motion-based Self-Reflection Framework for Fine-grained Robotic Action Correction

CVPR2025 / Vision-Language-Action Models