Passive bio-inspired flow sensing for autonomous underwater vehicles: A digital twin framework for object detection and localisation

Abstract

Throughout the natural world, organisms have evolved a wide range of strategies to sense their environment, often surpassing the performance of artificial sensors despite significant technological advancements. Biomimicry presents a compelling pathway for innovation in emerging technologies, as nature may already offer solutions to complex sensing challenges. The fish lateral line system, for example, is a distributed network of neuromast sensors capable of detecting local fluid velocities, accelerations, and pressure gradients. Flow sensing is of particular relevance to this research, as it offers the potential to significantly improve the performance of underwater vehicles operating in complex and dynamic environments by enabling the measurement and interpretation of the surrounding fluid. Drawing inspiration from distributed biological sensing systems, this paper describes the development of a bio-inspired digital twin model, accompanied by a signal processing algorithm designed to extract information about upstream obstacles from the flow surrounding a vehicle. The objective is to utilise flow sensing to support a range of autonomous underwater vehicle behaviours, including environmental interpretation for obstacle detection, unsupervised decision-making, and energy harvesting. Leveraging data from flow simulations in a virtual environment, the digital twin sensors are used to investigate how passive flow sensing can classify and localise an upstream bluff body. The proof-of-concept results presented here demonstrate that a passive sensor array, positioned downstream of a bluff body, can detect the wake, estimate the approximate size of the upstream object, and provide critical information to support collision avoidance.