Dual-eyes visual-based sea docking for sea bottom battery recharging

Abstract

In this paper, a robust stereo vision-based sea docking for autonomous underwater vehicles (AUVs) for sea bottom battery recharging is presented. Docking station is unidirectional one which requires high homing accuracy and robustness against difference disturbances. Real-time pose tracking was developed using stereo vision from two cameras and known 3D marker. Real time relative pose estimation was implemented as software implementation utilizing 3D model-based matching method and real-time multi-step GA. Remotely Operated Vehicle (ROV) with maximum depth 50 m is used as a test bed. Successful docking operation using proposed system demonstrated for underwater battery recharging was conducted in the sea, near Wakayama City, Japan. Experimental results confirmed the robustness of the system proving homing accuracy on the order of centimeter under real-world conditions.