Sound Source Localization

Abstract

Localization tasks on robotic platforms are typically based on camera, radar or lidar systems. In the recent time, microphones are implemented as well for speech input. Naturally, this information could also be used for the localization of the sound source. This thesis shall show the feasibility of such a system and that this system could be implemented on hardware with limited processing capabilities. For the localization of sound sources the well known MUSIC algorithm is used and adapted with several improvements being implemented to speed up the processing while assuring high accuracy. The hardware system used in this thesis is composed of two microphone arrays, each with four microphones, will demonstrate the function. Experimental evaluations in an anechoic chamber show the feasibility of the approach.