Acoustic Search and Detection of Oil Plumes Using an Autonomous Underwater Vehicle

Abstract

We introduce an adaptive sampling method that has been developed to support the Backseat Driver control architecture of the Memorial University of Newfoundland (MUN) Explorer autonomous underwater vehicle (AUV). The design is based on an acoustic detection and in-situ analysis program that allows an AUV to perform automatic detection and autonomous tracking of an oil plume. The method contains acoustic image acquisition, autonomous triggering, and thresholding in the search stage. A new biomimetic search pattern, the bumblebee flight path, was designed to maximize the spatial coverage in the oil plume detection phase. The effectiveness of the developed algorithm was validated through simulations using a two-dimensional planar plume model and a 90-degree scanning sensor model. The results demonstrate that the bumblebee search design combined with a genetic solution for the Traveling Salesperson Problem outperformed a conventional lawnmower survey, reducing the AUV travel distance by up to 75.3%. Our plume detection strategy, using acoustic sensing, provided data of plume location, distribution, and density, over a sector in contrast with traditional chemical oil sensors that only provide readings at a point.