AUV docking experiments based on vision positioning using two cameras

Abstract

An AUV(Autonomous Underwater Vehicle) docking test got finished successfully in the general deep water pool lab of Harbin Engineering University. Two major problems got solved. One is reliable underwater navigation method providing accurate relative position information, which is basically needed while AUV approaching. Four green LED lights are fixed around the station and two cameras are installed in AUV’s head part. A reliable vision positioning method using two cameras is adopted. Core algorithm switches between combining monocular and binocular positioning method according to the number of image in both cameras.. The other problem is that control system should be designed suited to the docking procedure, that is to say, outputs of controller ensures the declining of deviation between AUV position and center line of dock station. A two-layer control system is adopted considering the underactuated feature of our test bed. The guidance controller is designed based on Lyapunov stability theory. It outputs reference heading and trimming angle according to position deviations in horizontal and vertical plane respectively. Control parameters are adaptively adjusted by fuzzy algorithm. The execution controller eliminates errors of attitude angle employing an adaptively S-surface non-linear function. Pool test demonstrates that the navigating and control system works well. The system guarantees successful docking rate greater than 80%.