GEOPHYSICAL NAVIGATION OF AUTONOMOUS UNDERWATER VEHICLES

The paper addresses the problem of Geophysical Navigation (GN) of Autonomous Underwater Vehicles (AUVs). The objective is to use bathymetric and geomagnetic information to estimate the position and velocity of an AUV in the presence of unknown ocean currents. The navigation estimator derived mechanizes a sequential Bayesian estimation strategy consisting of a novel particle filter algorithm that we name Smooth Kernel Particle Filter. The estimator relies essentially on the AUV kinematics. It integrates measurements derived from the terrain with dead-reckoning based on Doppler and attitude sensors. Central to the strategy derived is the concept of maps of invariant gradients of the geomagnetic field as an efficient form of representing geomagnetic terrain information. The results obtained through computer simulations show the effectiveness of the GN approach proposed.