Vision-language models struggle to detect robot collisions safely

Original: Probing Collision Grounding in Vision-Language Models for Safe Human-Robot Collaboration

Source: arxiv.org ↗