cs.CV, cs.MM, cs.RO, eess.IV

E-VLA: Event-Augmented Vision-Language-Action Model for Dark and Blurred Scenes

arXiv:2604.04834v1 Announce Type: new
Abstract: Robotic Vision-Language-Action (VLA) models generalize well for open-ended manipulation, but their perception is fragile under sensing-stage degradations such as extreme low light, motion blur, and black…