Uncrewed submarines, underwater gliders, and surface vessels are the future of environmental monitoring and naval surveillance in the ocean. These autonomous robots will cooperate with and, in many cases, replace humans (reducing danger to life) to give us a wide-reaching and persistent view of our underwater environment. However, there is “an elephant in the room”: artificial intelligence (AI) consumes power (potentially a lot of power!) and marine robots have a limited supply, e.g., from batteries or renewable sources such as solar. Therefore, there is a critical engineering need for low-power AI algorithms and hardware for this vision to become a practical reality.
My PhD will explore the trade-off between AI capability and reliability versus the realistic power limitations imposed on robots in the challenging real-world ocean environment. Furthermore, it will develop, test, and demonstrate low-power AI solutions on a real marine robot. The project is funded by the Seiche Water Technology Group and will be centred around their Autonaut wave-propelled, solar powered uncrewed surface vessel (USV) with its underwater acoustic and above-water electro-optic sensors. Relevant use cases include monitoring of marine mammals, oceanographic science, and naval surveillance for underwater intrusion into protected areas.
My research interests are AI and the application of power intensive models to low powered marine robotics. I’m specifically interested in optimising and reducing the computational complexity of an AI model’s forward pass/inference, as well as the development of new power-efficient-by-design model architectures.
Master in Science in Computer Science (First Class Honours), University of Exeter (2019-2023)
Dr Alan Hunter
Dr Andy Barnes
