Autonomous docking for work class ROVs

This thesis describes research work in the domain of underwater robotics. It is aimed towards improving performance and achieving partial or full autonomy in the rapidly increasing underwater resident robotics field, with the emphasis on the development of a suite of technologies for autonomous Remotely operated vehicle (ROV) docking. Fundamentally resident ROVs operating from shore demand high bandwidth, zero latency communication link which is often unavailable, thus high levels of automation are needed to compensate. This is especially important for time-critical tasks such as ROV docking. In addition, the docking station provides a download/upload link, charging point, and overall mechanical protection for the resident vehicles. Therefore, the docking of the ROV at the end of a mission is one of the crucial ROV tasks and often dictates weather window extents. Research in the literature mainly focuses on the docking to a static docking station, and although docking to a docking station deployed to the seabed is part of the thesis, the emphasis is on the docking to a Tether Management System (TMS) garage suspended from the floating platform/ship. This is used as a ROV docking scenario close to that of docking to a dock on a floating production platform such as floating oil production platform, floating wind platform or floating offshore fish farm seacage. Typically it is difficult to compensate for all motion between the ROV and the suspendedTMS/dock due to underpowered thrust,large inertia,and high drag forces, presenting a significant challenge. If the docking of a conventional work-class ROV to a suspended TMS proves successful, it presents a significant contribution and accelerates the path towards collaborative and integrated ROV and Autonomous surface vehicle (ASV) systems. This advancement will drive further improvements in the autonomous transition of the existing intervention subsea vehicles across large areas, primarily associated with the offshore wind sector. The work presented in the thesis consists of three major parts. The first is the visual pose estimation system developed to provide accurate relative position measurements between the ROV and the docking station/TMS. The developed system is based on active light beacons asymmetrically arranged to form a unique marker, and a machine vision camera. The system is built around conventional, industry-standard subsea LED lights and camera and it has been successfully tested both in a lake and in the ocean. The additional algorithm has been developed to reduce the pose estimation errors due to the low camera sensitivity to angle measurements from longer distances. The system has been developed for standard work-class ROV systems found throughout the sector, deployed from suspended cage type TMS. The second is autonomous docking of an industry-standard work-class ROV to cage type TMS using a visual-based pose estimation approach. This included both, autonomous docking to static TMS deployed to the seabed, and docking to TMS suspended from the ship. Evaluation of the system has been demonstrated through completion of offshore trials in the North Atlantic Ocean during January 2019. The third is a suspended TMS heave motion prediction method for ROV docking, based on the Adaptive Neuro-Fuzzy Inference System (ANFIS). With large ROV inertia and drag forces acting against it, the ROV is not agile enough to match a cage type TMS heaving motion. Therefore the ROV docking manoeuvre has to start before the ROV and the TMS align. This also includes matching the ROV to the docking depth that covers top or the bottom half of the TMS heave range, where TMS vertical speed is low. The method includes on-site neural network training based on previous TMS depth measurements and the TMS depth prediction. In addition, this method could be used standalone as a ROV pilot aiding tool.