Return-to-site of an AUV using terrain-relative navigation: Field trials

This paper presents the results from recent field trials in Monterey Bay, CA which demonstrate the use of Terrain-Relative Navigation (TRN) to perform return-to-site missions using a 53cm diameter Dorado-class Autonomous Underwater Vehicle (AUV). For these tests, a series of targets were identified on a bathymetry map that were recognizable features (e.g. large boulders and outcroppings), and a trajectory was constructed for the AUV to fly over those targets (i.e. the latitudes and longitudes as defined on the map). Multiple missions were flown using a real-time TRN algorithm to identify and correct for georeferencing errors in the bathymetry map. The results were repeatable. They demonstrated that there was a 20m georeferencing error in the map. Without TRN, the targets were missed. With TRN, the AUV flew directly over the targets. Performance was independently validated using acoustic imagery from an on-board mapping multibeam sonar. These field results are presented and described. The TRN algorithm is also discussed.