This paper addresses an automation method of underwater tasks using AUV (Autonomous Underwater Vehicles).An AUV has a small agent vehicle to connect to an underwater docking station which is connected with operators. If the agent docks, the operator can fully monitor and control the AUV as an ROV (Remotely Operated Vehicles).Under human supervision, it can carry out complicated tasks such as underwater manipulations. If the agent undocks and is recovered, the AUV navigates as an untethered vehicle. In order to evaluate the proposed docking system, an experimental agent and a docking device were developed. For efficient and reliable docking, a visual landmark and a smart cable-based docking method were developed. A preliminary docking test was carried out at an indoor tank to evaluate the proposed docking system.


Risky and complicated underwater tasks, such as underwater structure's maintenance, have been manually performed by divers and ROV (Remotely Operated Vehicles). These tasks are dangerous for humans and ROV require heavy deck supports, such as tether maintenance using a specific mother ship. Automation of the tasks using AUV (Autonomous Underwater Vehicles) can be an alternative. They have excellent mobility and hardly need the deck support. They have been successfully automated for simple tasks such as wide area data collection, and challenging for complicated task's automation. However, their intelligence and reliability have not yet reached a sufficient level to undertake the tasks. Human intervention in or monitoring of the tasks can be an interim solution. During the task, if a communication link between the AUV and an operator is secured, they can carry out the tasks under human supervision. They will gradually be able to improve their performance. In the end, they will have automated most risky tasks. For this link, there are 2 feasible underwater methods. One is an acoustic link, and the other is a tethered link.

This content is only available via PDF.
You do not currently have access to this content.