Man-made versus biological in-air sonar systems

Abstract

In this chapter we will argue that biologically inspired sonar systems, i.e. man-made systems that implement functional principles of their biological counterparts, are capable of significantly improving the performance of current in-air sonar systems. Instead of collecting large numbers of sonar range readings from multiple observation points and combining them into a reliable environment map we advocate the use of intelligent sonar sensors capable of extracting significantly more information from a single measurement. As an example of this bio-inspired approach we present a binaural sonar sensor capable of localizing reflectors in 3 D-space using broadband spectral cues introduced by the emitter and receiver directional filters. Acoustic simulations indicate that duplicating the outer ears and combining them with an emitter that acts by directing emitted energy in the frontal direction should be sufficient to approximate the significant features of the directional properties of a real bat’s sonar system. Localisation is performed by a template matching scheme whereby the spectrum of the received echo signal is compared with a set of stored spectral templates, one for every direction. This bio-inspired 3 D localisation scheme was implemented on a robotic bat head and validated in a series of experiments. The results from these experiments show that both the monaural and the binaural spectral cues introduced by the emitter/receiver directional filters carry sufficient information to discriminate between different reflector locations in realistic noise conditions. The experiments further show that to track a moving spherical targetwith our robotic system spectral information from both receivers is required.