Robust methods of magnetic navigation of marine robotic vehicles

Abstract This paper describes the development of robust estimation methods of magnetic navigation for small, affordable underwater robotic vehicles. Relying on prior work by the authors, the paper proposes a particle filter navigation algorithm which is integrated with other position estimation methods to increase the robustness of the estimation process. The new techniques embedded in the navigation particle filter include an adaptive-likelihood particle filter and a magnetic contour matching algorithm implemented in the Fourier domain to achieve superior computation speeds. The performance of the navigation methods proposed is assessed through a series of tests which exploit real data acquired in the water with a marine robotic vehicle. The results obtained illustrate the high potential of the approach for magnetic navigation of autonomous underwater vehicles.