A hierarchical heterogeneous UAV swarms architecture for casualty detection and task allocation in post-disaster rescue

In post-disaster search and rescue missions supported by unmanned aerial vehicle swarms, efficient task allocation is essential for rapidly locating and confirming survivors under strict time, energy, and communication constraints. However, wide-area scanning can improve spatial coverage while increasing the risk of false alarms, whereas repeated close-range verification can improve detection reliability at the cost of additional flight time and communication resources. To address this trade-off, this study proposes a hierarchical heterogeneous unmanned aerial vehicle swarm architecture in which high-altitude unmanned aerial vehicles provide regional communication relay and coordination, while low-altitude unmanned aerial vehicles perform cell-level search, scanning, and target verification tasks. Based on this architecture, a utility-driven hierarchical cooperative greedy algorithm is developed to dynamically allocate tasks by jointly considering detection uncertainty, energy consumption, communication feasibility, and opportunity costs. In addition, a regional load-aware coordination mechanism is introduced to mitigate task imbalance caused by uneven survivor distributions across disaster areas. Simulation results show that the proposed method improves survivor confirmation performance while reducing false positives and missed detections. The results further demonstrate that the proposed framework achieves an effective balance among rescue efficiency, energy consumption, and computational complexity, indicating its potential for large scale post-disaster search and rescue applications.