UAV Surveillance Optimization for Networks in Disaster-Stricken Areas

Unmanned aerial vehicle (UAV) cameras have recently been an emerging powerful tool to grasp damage situations in disaster-stricken areas. Whereas previous methods have mainly focused on covering widely spread target fields, we consider UAV trajectory optimization for the network surveillance in disaster-stricken areas. Under the constraint of UAV flight distance, we aim at maximizing a utility function reflecting the priority of monitoring determined by domain-specific knowledge, such as the risk of damage due to disasters. Although this problem is NP-hard, we propose a method for obtaining approximate solutions on the basis of two alternative heuristic algorithms. To investigate the characteristic of our method, we conducted numerical simulations in an artificial network, showing that our algorithms empirically give good approximations of the optimal solution. In addition, to evaluate the effectiveness and feasibility of our method, we carried out numerical experiments using a real road network constructed with open geographic information system (GIS) data, demonstrating that our method generates plausible UAV trajectories in appropriate parameter settings.