Application of the Sound Localization Algorithm to the Global Localization of a Robot

Abstract

Sound localization is the process of determining the position of the sound sourceexpressed by the azimuth and the elevation angles. Robot localization is the process ofdetermining a robot pose from sensor data. Based upon recent researches done onbinaural sound localization algorithms in the Neural Systems Laboratory at the Institutefor Systems Research, this research sought to apply localization of sound to automaticallylocalize the robot in a small laboratory environment. By placing three sound sources atpredetermined positions and letting the robot find the angles between the three soundsources by rotating in place to orient itself to each source in succession, the robot wasable determine its current pose. The research successfully realized this model of robotlocalization. Simulation of the localization model was created to calculate range possibleerrors at different points in the map.