1 paper
Yuan Zhang, Shiqi Zhang, Yedong Shen +11
Vision-Language-Action (VLA) models achieve strong benchmark performance but still struggle in real-world deployment with unseen objects, background shifts, and different robot emb…