Hierarchical Coordination and HOCBF-Based Safety Filtering for Connectivity-Aware Multi-UAV Navigation

This paper presents a hierarchical coordination and safety-filtering framework for connectivity-aware cooperative multi-UAVnavigation in obstacle-dense corridor environments. The architecture separates low-rate centralized team coordination from high-rate decentralized execution. The centralized layer updates a belief map, samples boundary-aligned exploration targets, assigns agents using the Hungarian algorithm, updates leader–follower roles, and constructs a directed minimum spanning tree (D-MST) as a sparse range connectivity backbone. Each agent replans an A* path on its local map and filters the nominal command through a decentralized high-order control barrier function quadratic program (HOCBF-QP) with hard obstacle- and inter-agent collision avoidance constraints and softened, link-activated connectivity constraints. Simulation studies on cluttered corridor maps demonstrate improved task completion, regulated team separation, and recovery from local trap-like regions while maintaining high empirical safety and range connectivity satisfaction. The results show that team-level coordination mitigates congestion-induced stalling, while the prioritized HOCBF-QP maintains high empirical safety and limits connectivity violations.