Terrain-relative navigation for autonomous underwater vehicles

Abstract

Navigation is a key technology for autonomous underwater vehicles (AUVs), and currently, it limits potential and existing vehicle capabilities and applications. This thesis presents a terrain-relative navigation system for AUVs that does not require the deployment of acoustic beacons or other navigational aids, but instead depends on a supplied digital bathymetric map and the ability of the vehicle to image the seafloor. At each time step, a bathymetric profile is measured and compared to a local region of the supplied map using a mean absolute difference criterion. The region size is determined by the current navigation uncertainty. For large regions, a coarse-to-fine algorithm with a modified beam search is used to intelligently search for good matches while reducing the computational requirements. A validation gate is defined around the position estimate using the navigation uncertainty, which is explicitly represented through a covariance matrix. A probabilistic data association filter with amplitude information (PDAFAI), grounded in the Kalman Filter framework, probabilistically weights each good match that lies within the validation gate. Weights are a function of both the match quality and the size of the innovation. Navigation updates are then a function of the predicted position, the gate size, all matches within the gate, and the uncertainties on both the prediction and the matches. The system was tested in simulation on several terrain types using a deep-ocean bathymetry map of the western flank of the Mid-Atlantic Ridge between the Kane and Atlantis Transforms. Results show more accurate navigation in the areas with greater bathymetric variability and less accurate navigation in flatter areas with more gentle terrain contours. In most places, the uncertainties assigned to the navigation positions reflect the ability of the system to follow the true track. In no case did the navigation diverge from the true track beyond the point of recovery.