UAV search optimization for recording emerging targets with camouflaging capabilities

Abstract

This paper extends the emerging target information gathering domain by introducing a camouflaging component, investigating the benefit of using multiple UAVs, and studying the impact of allowing re-visits. Previous work in this area addresses the UAV Orienteering Problem for target detection where targets emerge according to non-homogeneous space–time Poisson processes. Our extension considers that emerged targets will camouflage themselves to become undetectable after a period of time and nodes can be revisited. Routes for single or multiple UAVs are generated using the Team Orienteering Problem with Time Windows and evaluated using simulation. In addition, a framework is developed for comparing routes, and the value added by increasing the solving time is investigated. Our computational testing reveals increasing the number of time windows used increases the expected route value. A factorial analysis is conducted which indicates the network topology, the number of time windows used, and the coefficient of variation for camouflage time generally have significant effects on the expected number of targets detected regardless of the number of UAVs used. In addition, increasing the amount of time spent solving the problem does not always increase the number of expected target detected.