협소 공간 내 인명 탐색 로봇의 생존자 음원 위치 추정과 운용에 관한 연구

Abstract

This paper describes the study on Sound Source Localization of survivors for the robot searching victims in a narrow space. Casualties due to collapse accidents continue to occur. To respond them, research cases using robot technology are expanding. Sound Source Localization identifies the direction of sound, and if it applied to a robot that performs a life search mission in a narrow space, the mission performance will be improved. In order to monitor the result of Sound Source Localization and the data of multi sensor module including microphones, monitoring GUI was implemented with a focus on data visualization so that operators can intuitively grasp the situation of the work site. We proposed Generalized Cross-Correlation Phase Transform algorithm using TDOA method performs Sound Source Localization on 180 degrees in front using two microphones arranged at an interval of 0.03m. Through frequency domain analysis, we designed filter to make algorithm react only to the sound with a human voice frequency. Furthermore, calibration algorithm was integrated to solve the problem that the update cycle of result value becomes very short when passing through the filter, making it difficult to check the value. Experimental results obtained through experiments using constructed testbed verify the performance of the proposed microphone array and integrated Sound Source Localization algorithm.