무인자율잠수정 항법 오차 개선을 위한 광학 영상 및 음향 영상의 신뢰성 평가 방법

Abstract

We propose a method to evaluate optical and acoustic images to compensate for error of backtracking navigation of AUV. With the proposed method, the vehicle can record useful images on its path, and conduct backtracking navigation efficiently and robustly even in turbid water. At each point during the exploration, AUV chooses better image between optical and acoustic image by feature detection. The chosen image is saved in the path array with navigation data. If user commands backtracking to the robot, it generates the optimal path, considering the distribution of feature scores on its trajectory. Using feature matching of optical images and trajectory backtracking method with imaging sonar, the vehicle can compensate for error of dead-reckoning during backtracking. To verify this method, we acquired the field data using the hovering-type AUV Cyclops and tested the feasibility and processing time of image selection and optimal path generation method.