Terrain Aided Navigation for Remus Autonomous Underwater Vehicle.

摘要

This research investigates the ability to create an undersea bathymetry map and navigate relative to the map. This is known as terrain aided navigation (TAN). In our particular case, the goal was for an autonomous underwater vehicle (AUV) to reduce positional uncertainty through the use of downward-looking swath sonar and employing TAN techniques. This is considered important for undersea operations where positioning systems such as GPS are either not available or difficult to put in place. There are several challenges associated with TAN that are presented: The image processing necessary to extract altitude data from the sonar image, the initial building of the bathymetry map, incorporating a system and measurement model that takes into consideration AUV motion and sensor uncertainty and near-optimal, real-time estimation algorithms. The thesis presents a methodology coupled with analysis on datasets collected from joint Naval Postgraduate School/National Aeronautical Space Administration experimentation conducted at the Aquarius undersea habitat near Key Largo, Florida.