Multiple-Target Tracking With Binary Proximity Sensors

Abstract

Recent work has shown that, despite the minimal information provided by a binary proximity sensor, a network of these sensors can provide remarkably good target tracking performance. In this article, we examine the performance of such a sensor network for tracking multiple targets. We begin with geometric arguments that address the problem of counting the number of distinct targets, given a snapshot of the sensor readings. We provide necessary and sufficient criteria for an accurate target count in a one-dimensional setting, and provide a greedy algorithm that determines the minimum number of targets that is consistent with the sensor readings. While these combinatorial arguments bring out the difficulty of target counting based on sensor readings at a given time, they leave open the possibility of accurate counting and tracking by exploiting the evolution of the sensor readings over time. To this end, we develop a particle filtering algorithm based on a cost function that penalizes changes in velocity. An extensive set of simulations, as well as experiments with passive infrared sensors, are reported. We conclude that, despite the combinatorial complexity of target counting, probabilistic approaches based on fairly generic models of trajectories yield respectable tracking performance.