Active Sound Localization in a Symmetric Environment

Abstract

Localization for humanoid robots becomes difficult when events that disrupt robot positioning information occur. This holds especially true in symmetric environments because landmark data may not be sufficient to determine orientation. We propose a system of localizing humanoid robots in a known, symmetric environment using a Rao-Blackwellized particle filter and a sound localization system. This system was used in the RoboCup Standard Platform League, and has been found to reduce the amount of own-goals scored as compared with the previously used localization system without sound.