On the Natural Robustness of Vision-Language Models Against Visual Perception Attacks in Autonomous Driving

Open in new window