On multi-UAV scheduling for human operator target identification

We address the problem of a single human operator in charge of monitoring multiple UAVs in a reconnaissance mission. The human operator must classify targets as they appear on video feeds from the various aircraft as friends or foes. We introduce the idea of the human monitoring a single screen and a decision aid that automatically selects the video feed to be displayed on that screen. In this setting and given that the location of the targets is unknown and decision must be made in real time, the challenge here addressed is the development of a scheduling strategy which queues the video feeds in such a way that no target is missed and all targets are looked at for a specified amount of time. We present a Linear Programming (LP) formulation to this problem for the simple case of two UAVs and two equidistant targets. We develop an online algorithm that uses this LP as a building block to be applied to the more general case of multiple non- equidistant targets. We further show that in order to guarantee the feasibility of the scheduling strategy and ensure that no targets are missed it is sufficient to enforce a lower bound on the distance between targets.