A Comparison of Nonlinear Filters for Underwater Geomagnetic Navigation

The present article focuses on the use of Bayesian Filters for Geophysical Navigation (GN) of autonomous underwater vehicles (AUVs) using geomagnetic data. Due to the nonlinear non-Gaussian nature of the problem, three classical Bayesian nonlinear filtering algorithms are considered: the Extended Kalman Filter (EKF), the Unscented Kalman Filter (UKF), and the Particle Filter (PF). Extensive Monte Carlo (MC) simulations using real data acquired with an autonomous marine vehicle of the Medusa class were done to assess the robustness and efficacy of the proposed solutions. For a fair evaluation of the estimation performance, the following scenarios were considered: different initial covariances and kidnapped vehicle. Simulation results showcase the robustness of the PF solution for the geomagnetic navigation problem. However, EKF and UKF are computationally less expensive and may be viable solutions in certain scenarios.