A £1.4 million UK project is currently trialling the use of new artificial intelligence (AI) control systems for autonomous underwater robots, dedicated to offshore wind farm repairs and maintenance.
The project, dubbed Underwater Intervention for Offshore Renewable Energies (UNITE), is an Engineering and Physical Science Research Council (EPSRC) Prosperity Partnership programme, led by Heriot-Watt University, Edinburgh and hosted in collaboration with Imperial College London, geo-data company Fugro and underwater software developer Frontier Robotics. UNITE is also supported by the National Robotarium, sited at the Heriot-Watt campus, which will serve as a test bed for the project’s R&D.
The participants believe that using autonomous and remote-controlled underwater bots to replace the deployment of larger maintenance vessels could slash carbon emissions by up to 97% in the long term, while shielding personnel from the risks associated with entering hazardous offshore environments. UNITE states: “The UK has more than 2,600 offshore wind turbines, with plans to quadruple capacity by 2030. On average, each turbine requires up to three maintenance check-ups per year, a frequency that increases as turbines age.”
Speaking in late October, David Morrison, National Robotarium project manager, commented: “Our trials are showing promising results in enabling underwater robots to maintain stable contact with offshore structures in challenging conditions. The technology could transform offshore wind maintenance, potentially reducing fuel consumption of maintenance missions from 7,000litres per day to just 200litres.” Yvan Petillot, professor of robotics at Heriot Watt University and National Robotarium academic co-lead, added: “We are especially interested in subsea inspection and manipulation in dynamic environments where existing solutions cannot be used.” Common tasks would include using the bots to take precise measurements and conduct visual inspections, as well as to clean structures and repair defects.
One challenge UNITE seeks to resolve is the so-called ‘chicken head problem’, which refers to the difficulty of maintaining stability for robotic arms or tools while working on underwater structures, including wind turbines, in the presence of strong currents and waves. Morrison said: “Just as a chicken can keep its head stable while its body moves, robots need to maintain precise positioning in turbulent seas. This stability is crucial for tasks like measuring the corrosion rate of underwater structures, where consistent contact is essential.” To resolve this problem, the project partners will develop advanced control systems and use machine-learning algorithms to allow the bots to “adapt in real time to changing conditions”, UNITE says.
The project also aims to adopt advanced 3D semantic mapping capabilities; these would permit the robots to create detailed maps of their underwater environment. Such maps would boost the bots’ ability to navigate complex structures and identify critical components in need of attention. UNITE will also push for the enhanced coordination of ROVs and electric ROVs (eROVs), deployed from uncrewed surface vessels (USVs), to undertake missions in tandem.
Frontier Robotics, which is based at the National Robotarium, will provide perception, mapping and autonomous technologies for the project. This will also involve integrating advanced stereo camera systems with edge computing to support the AI systems being tested. The partners’ goal is for the resulting solution to “deliver data insights in just three hours, compared to the current industry standard of three weeks – a potential 1,500-fold improvement in speed”, UNITE says.
According to Jonatan Scharff Willners, Frontier Robotics CEO: “With the exponential growth of offshore infrastructure, we need to look towards deploying more AI, robots and autonomy to enable the industry to take advantage of new technologies to work even more efficiently and to scale with the global demand.” UNITE adds that it would welcome fresh input from prospective new partners.
