音声処理を用いた移動ロボット操作インタフェース

抄録

This paper describes a speech-based interface for a human support robot. We model a transition of the robot state based on human voice commands and camera images using a finite state machine (FSM). For recognizing human voice，we use HARK and Google Speech API. We perform a morphogical analysis of voice recognition results using Mecab. In addition，we develop a human detection and localization system using the sound source localization and sound source separation by HARK and a human detection by a 2D laser scanner. We carried out two kinds of experiments. One is to evaluate the sound source localization results using a microphone array (TAMAGO-03) and other is a real robot control experiment.