Waveguide invariant navigation of an autonomous underwater vehicle

Reliable navigation of an autonomous underwater vehicle (AUV) remains a challenge due to the unavailability of a global positioning system underwater. However, the self-localization uncertainty of an AUV with an acoustic receiver can potentially be reduced by leveraging passive acoustic ranging techniques. In particular, in shallow water, the range between an acoustic source and a passive receiver can be estimated using the waveguide invariant. In this work, we develop a sequential Bayes navigation filter that fuses passive recordings of an acoustic source with the measurements of an inertial sensor. The acoustic source can either be cooperative or a source of opportunity, e.g., a container ship. Using particle-based processing, our filter can either estimate the waveguide invariant or the AUV position by making use of a detailed nonlinear statistical model of the received signal. If the position of the AUV is accurately known, e.g., in the beginning of the AUV’s mission, the waveguide invariant can be estimated. Alternatively, assuming that accurate waveguide invariant information is available, range information can be obtained and used to improve the estimate of the AUV’s position. The navigation capability provided by the proposed sequential Bayes filter is demonstrated using simulated and real data.