Abstract

Abstract. This paper addresses the problem of actively controlling robotic teams with range-only sensors to a) discover and b) localize an unknown number of devices. We develop separate information based objectives to achieve both goals, and examine ways of combining them into a uni-fied approach. Despite the computational complexity of calculating these policies for multiple robots over long time horizons, a series of approxi-mations enable all calculations to be performed in polynomial time. We demonstrate the tangible benefits of our approaches through a series of simulations in complex indoor environments. 1