Visual-based deep sea docking simulation of underwater vehicle using dual-eyes cameras with lighting adaptation

This paper proposes visual-based underwater vehicle docking/homing system under the environment especially simulated for deep sea trial. Instead of measuring absolute position of vehicle using other non-contact sensors, estimation of the robot's relative position and posture (pose) using dual-eyes cameras and 3D target object is proposed. For relative pose estimation, 3D model-based recognition approach is applied because of its real-time effective performance. According to effectiveness, simplicity and repeatable evaluation ability for real-time performance, Genetic Algorithm (GA) is utilized here as a modified form of " Multi-step GA" to evaluate the gene candidates that represent relative poses until getting the best gene with the most trustful pose in dynamic images input by videorate. P controller is used to control the vehicle for the desired pose using real-time images from dual-eyes cameras. Since the underwater environment is complex, this work also addresses to the experiment for deep sea docking operation under changeable lighting environment derived from pose fluctuation of underwater vehicle with two LED-lighting direction altered accordingly that offers huge challenges for visual servoing. As the main contribution for this paper, therefore, we have developed visual servoing using adaptive system for unknown lighting environment. Remotely Operated Vehicle (ROV) is used as a test bed and the experiments are conducted in simulated indoor pool. Experimental results show visual servoing performance under varying lighting conditions and docking performance using proposed system.