Gradient search with autonomous underwater vehicles using scalar measurements

Abstract

We examine how AUVs can be used to locate features of interest from scalar measurements without exhaustive search. It is first shown how the Autonomous Benthic Explorer (ABE) was used to successfully locate the deepest part of a pond. ABE computed the total water depth by combining the computed vehicle depth with the return of a single beam sonar altimeter. While moving in a straight line, ABE was able to estimate the gradient of total water depth along the vehicle track. By executing circular maneuvers, ABE was further able to estimate the 2D horizontal slope of the bathymetry. These two functions were combined to produce a procedure to locate the deepest point in the search region. This was accomplished without the need for a global navigation system. It is discussed how a vehicle with the capabilities of ABE could be used to locate hydrothermal vents using temperature, optical backscatter or chemical tracers as the search parameter. Data from an actual preprogrammed survey conducted by ABE in an area of sea-bed hydrothermal activity is illustrated. This data has been combined with measurements conducted by moorings, towed vehicles and the manned deep submersible Alvin to create a mathematical model of a vent field as might be seen by an AUV maneuvering through the region. Using this model we will show techniques for locating hydrothermal vents based on gradient following and other optimization methods that exploit the current understanding of the structure and physics of hydrothermal plumes.