Towards a multi-robot architecture for bearing-only tracking of underwater acoustic sources using vector sensors

This paper presents a multi-vehicle distributed system for the real-time estimation of the position of an underwater acoustic source, based on Direction of Arrival (DOA) observations obtained through onboard Acoustic Vector Sensors (AVSs) mounted on Autonomous Underwater Vehicles (AUVs). In particular, the paper describes the onboard processing of the AVS data and a Bearing Only Tracking (BOT) algorithm employing a pseudo-linear formulation suitable for the underwater domain. Such algorithm efficiently handles communication latency and asynchronous packet arrivals by constructing local regressors. The results of a scaled experiment at sea in controlled and simplified conditions are reported, to verify the feasibility of the approach and to experimentally determine some of the system characteristics, together with the performance variability as influenced by selected processing parameters and by the state of the vehicles. Finally, a simulative study is reported in which, starting from the characteristics experimentally determined, system performance is evaluated vis-a-vis relevant parameters (as relative AUVs distance, sound source distance and motion). The combination of experimental and simulative results validates the practical viability of the proposed distributed multi-AUV framework for real-time underwater target tracking under communication-constrained conditions, while highlighting the critical need to carefully tune design trade-offs to optimize system-level performance in real-world marine applications.