Efficient sensor coverage for acoustic localization

In this paper we consider a network of mobile agents carrying a simple sensor, able to measure the time an impulsive signal propagating isotropically from an unknown point source is detected, with a probability depending on the distance between the signal source and the sensor. Given an a priori probability distribution for the location of the source, we address the problem of controlling the motion of the agents, in such a way that the quality of the localization of the signal source is maximized, assuming an ideal estimation process. The performance criterion is the expected value of the determinant of the Fisher Information Matrix, as computed from the location of the sensors that detect the signal. We develop a gradient control law, ensuring convergence of the agents to a critical point with respect to the stated performance. An application to acoustic detection and localization of sniper fire is discussed and simulation results are presented.