Robust Speaker Localization System for a Humanoid Robot in Noise and Reverberation Environments

摘要

Human-robot interaction (HRI) is an important study tornallow more natural and effective communication betweenrnhumans and robots. Recently, research on HRI has beenrnreceiving an increasing amount of attention. In this paper,rnwe report on a speaker localization system for a humanoidrnrobot, as part of an auditory-based system of HRI. Ourrnsystem uses a six-channel microphone array on a humanoidrnrobot's head. To achieve good results in environments ofrnnoise and reverberation, we utilize a pre-emphasis filter; arnvoice activity detection (VAD) algorithm; and generalizedrncross correlation with phase transform weighting (GCCPHAT)rnmethod. Experimental results show that our speakerrnlocalization system for a humanoid robot can successfullyrnlocalize a speaker's position over the entire azimuth andrnaltitude in noise and reverberation environments.