AI RESEARCH

Risk-Bounded Multi-Agent Visual Navigation via Iterative Risk Allocation

arXiv CS.AI

ArXi:2509.08157v3 Announce Type: replace-cross Safe navigation is essential for autonomous systems operating in hazardous environments, especially when multiple agents must coordinate using only high-dimensional visual observations. While recent approaches successfully combine Goal-Conditioned RL (GCRL) for graph construction with Conflict-Based Search (CBS) for planning, they typically rely on deleting edges with high risk before running CBS to enforce safety.