Docking experiment of underwater vehicle by dual-eye visual servoing (Control performance verification with autonomous underwater vehicle)

Abstract

A variety of robots has been studied and developed for undersea exploration. One of the applications for undersea exploration is mining of undersea resources such as methane hydrate and rare metal by Autonomous Underwater Vehicle (AUV). To extend active duration time of the robot in the sea, a system through which the AUVs recharge autonomously batteries is a key technology. Our reseach group has confirmed in a pool environment the dual-eye visual servoing system made a pole attached to the vehicle dock into a hole. This experiment simulates situation where the vehicle approaches recharging station under deep sea. But previous studies were conducted with ROV (Remotely Operated Vehicle) that is controlled by remote computer instead of human operator. In those experiments, power cables affected performance of control accuracy. Therefore the performance of visual feedback with AUV named Tuna-Sand2 that is compeletely indepent from cables or wires and has a general structure of modernized AUV system, has been verified to confirm the practicability of dual-eye dockig system.

Keywords

• visual servoing
• genetic algorithm
• auv