Sound Localization Based on Excitation Source Information for Intelligent Home Service Robots

摘要

This paper is concerned with Sound Localization (SL) using Excitation Source Information (ESI) and effective angle estimation for intelligent home service robots that are equipped with multi-channel sound board and three low-cost condenser microphones. The main goal is to localize a caller by estimating time-delay with features obtained from the excitation source based on Linear Prediction (LP) residual and Hilbert envelop, when the speaker calls robot's name in all directions. For performance analysis, we collected SL-DB (sound localization database) with the variation of distance and angle under test-bed environments like home. Here the localization success rate (LSR) and average localization error (ALE) from field of view (FOV) range of robot camera are used as localization performance criterion. The experimental results reveal that the presented method shows a good performance in comparison with the well-known Time Delay of Arrival (TDOA) and Generalized Cross Correlation-Phase Transform (GCC-PHAT) method.