1 paper · 1 filter
Hyunse Lee, Jiwoo Jeong, Haneul Lee +3
Vision-language-model-based embodied agents can complete instructed tasks but often violate safety constraints in the process, a problem recently framed as interactive safety. Trai…