Autonomous Underwater Vehicle-Based Concurrent Detection and Classification of Buried Targets Using Higher Order Spectral Analysis

Abstract

This paper presents a processing concept for autonomous underwater vehicle (AUV)-based concurrent detection and classification (CDAC) of mine-like objects. In the detection phase, the AUV seeks objects of interest using a simple energy detector combined with a peak tracking mechanism. Upon detection, the processing mechanism changes to a higher order spectral (HOS) classification process. The system is demonstrated through theory, simulation and at-sea experiments to have promise in reducing the false alarm rate of mine detections. The HOS classification mechanism is also shown to have some benefit over classical spectral estimation in all cases. Components of the system concept were also demonstrated live onboard the AUV during the Generic Oceanographic Array Technology Sonar (GOATS 2002) experiment off the coast of Italy, while others are demonstrated using a comprehensive AUV sonar simulation framework