Unmanned surface vehicles (USVs) have gained increased attention in environmental monitoring, navigation assistance, search-and-rescue, and other fields over the past three decades. USVs provide an effective platform for mobile sensing applications and offer flexibility in specific capabilities. This work presents a network of compact USVs that are capable of deploying underwater sensors using an automated winch. The small, maneuverable nature of each USV is ideal for inland bodies of water, and the relatively large payload capacity allows for surveys that last multiple hours. Motivated by the application of acoustic telemetry-based fish movement tracking, this work focuses on using a network of USVs to localize an underwater acoustic tag by exploiting the time-difference-of-arrival (TDOA) of the emitted signal. A distributed TDOA-based particle filter (PF) algorithm is proposed for localizing a moving target modeled by a discrete-time correlated random walk (DCRW). Furthermore, an online model learning method is explored, where target position estimates are used to update the unknown probability distributions of the target’s movement model. Through numerical simulations, the distributed PF is shown to result in effective estimation of the target position when a node is connected to a network that collectively has an adequate number of TDOA measurements. Additionally, the efficacy of online model learning in handling model uncertainties is demonstrated in simulation studies.TDOA-based localization algorithms are further validated in field experiments using a network of four USVs carrying acoustic telemetry equipment. In particular, TDOA and GPS data are collected and used to assess the target estimation performance for the distributed TDOA-based PF and a distributed TDOA-based extended Kalman filter (EKF) under different settings for the network topology.
Read
