Implementation and calibration of a Bayesian binaural system for 3D localisation

Abstract

In this text we present a Bayesian system of auditory localisation in distance, azimuth and elevation using binaural cues only; we focus mainly on implementation details and the calibration procedure, and present experimental results. This binaural system is also integrated in a spatial representation framework for multimodal perception of 3D structure and motion - the Bayesian volumetric map (BVM). This solution will enable the implementation of an active perception system with great potential in applications as diverse as social robots or even robotic navigation.