AI RESEARCH
HomeGuard: VLM-based Embodied Safeguard for Identifying Contextual Risk in Household Task
arXiv CS.CV
•
ArXi:2603.14367v1 Announce Type: new Vision-Language Models (VLMs) empower embodied agents to execute complex instructions, yet they remain vulnerable to contextual safety risks where benign commands become hazardous due to subtle environmental states. Existing safeguards often prove inadequate. Rule-based methods lack scalability in object-dense scenes, whereas model-based approaches relying on prompt engineering suffer from unfocused perception, resulting in missed risks or hallucinations.