Optimal tracking of multiple targets using UAVs

Target tracking problems have been studied fairly extensively by researchers in the last few years. However, the problem of continuous tracking of all mobile targets using the fewest number of mobile trackers, even when the trajectories of all the targets are known in advance, has received very little attention. In this paper we study this problem, where the goal is to find the fewest number of trackers needed to track all the targets for the entire period of observation. Specifically, given a set of n targets moving in n different (known) trajectories in a two (or three) dimensional space, our objective is to find the fewest number of velocity-bounded UAVs (mobile sensors, trackers) and their trajectories, so that all the targets are tracked during the entire period of observation. We also study two other versions of the problem where not only the number of trackers but also the time during which the trackers are active is also taken into account. We formulate these problems as network flow problems and propose algorithms for their solution. We evaluate the performance of our algorithms through simulation and study the impact of parameters such as the speed and sensing range of the trackers.