A three-dimensional locating method for underwater robot based on sensors fusion

Abstract

Among various vision sensors used in localizing targets in robotics, the locating precision of monocular vision is low, and binocular vision requires difficult stereo matching and a heavy computation. Hence a novel application of sensors fusion for underwater robot locating is presented in this paper, which is combined with a linear laser emitter and an underwater camera to obtain 3D coordinates of targets. Due to the distortions especially the edge of the images contributing significantly to the accurate locating, camera radial aberration is considered when calibrating camera parameters to weaken the image distortion. Considering refraction affects locating precision in water, a method of modifying underwater refraction based on sensors fusion is presented to reduce locating error of depth. Experiment results verify the method based on sensors fusion is better than the secondary focus method. Also the camera calibration method and refraction influence modifying method proposed in this paper are proved to be able to improve locating precision obviously.