Visual-servo-based autonomous docking system for underwater vehicle using dual-eyes camera 3D-pose tracking

A visual-servo type remotely operated vehicle (ROV) system with binocular wide-angle lens was developed to survey submarine resources, decontaminate radiation from mud in dam lake and so on. This paper explores the experiments on regulator performance and underwater docking of the robot system utilizing Genetic Algorithm (GA) for real-time recognition of the robot's relative pose (position and posture) through 3D marker. The visual servoing performances have been verified as follows; (1) The stability performances of the proposed regulator system have been evaluated by exerting abrupt distrubane force while the ROV is controlled by visual servoing, (2) The proposed system can track time-variant desired target position in z-axis (front-back direction of the robot), (3) The underwater docking can be completed by switching visual servoing and docking modes based on the error threshold, and by giving time-varying desired target pose to the controller as a desired pose.