수중 로봇의 자율 주행을 위한 위치 추정 방법

Abstract

This paper describes a particle filter method and Kalman fiilter method for localization of an underwater robot. The method uses external measurement information as well as the internal motion information. The particle filter and Kalman filter have been used for localization of ground mobile robot. The method can take the uncertainty of robot motion and measurement into account. It consists of two iterative steps: prediction step and correction step. The prediction step calculates pose of the robot using the internal motion information either through internal sensing or presumed motion command. The second step corrects the predicted pose using the measurement relative to external environment. The measurement is largely based on acoustic signal processing. Unlike the ground environment, the underwater environment doesn't afford the use of radio signal or visual information. Only the acoustic signal can propagate and be detected for use. The acoustic signal is used for distance measurement using the time of arrival(TOA) or time difference of arrival(TDOA). This paper extends the localization capability of particle filter method to be applied in three dimensional underwater environment. It depicts how the filtering methods can be used for localization of an underwater robot. Simulation shows the localization performance of the method in cases of TOA and TDOA measurements. The performance is compared with that of deadreckoning, trilateration, and least square methods. To compare the accuracy and robustness of localization, some measures are introduced: the repeatability and the length of the estimated trajectory. An experiment in test tank shows the feasibility and availability of the proposed method.