Exploiting autonomous underwater vehicle mobility for enhanced sonar performance

Abstract

Autonomous underwater vehicles (AUVs) provide a mobile sonar platform for local environmental characterization and mine hunting missions in areas that are prohibitively dangerous or otherwise inaccessible to manned vessels. Such missions are typically implemented by providing the vehicle with a series of way-points that uniformly sample the region of interest. While these preplanned sampling paths can be effective for many purposes, a higher degree of efficiency and efficacy can be achieved when the AUV is able to react to its perceived environment. Such response to the sensory stimulus represents a preliminary step toward the AUV achieving tasks in the way that a dolphin or human might. Upon sensing an object or feature of interest, the AUV can further interrogate the object by obtaining multiple views from preferred vantage points. Probabilities of detection and correct classification can be greatly impacted by a wise choice of sonar-adaptive AUV behavior. In this paper, the ability of the AUV to improve sonar performance through sonar-adaptive mission planning is investigated in the context of the mine hunting problem. Trade-offs are discussed for continuous versus scripted mission adaptations, single-minded versus multiple-objective mission control, and task-wise versus mission-wise behavior. [Work supported by ONR and SACLANTCEN.]