Near-Surface Whale Detection from Autonomous Ships Using Asynchronous Sensor Fusion System (ASSIST)

SSCI, Woods Hole Oceanographic Institute and Olin College of Engineering propose to develop, implement and perform initial simulation testing of the ASSIST (Asynchronous Sensor fuSIon SysTem) for near-surface whale detection. ASSIST is an information fusion approach which will integrate and process information from sonar, thermal imaging sensor, radar, and an information service such as the Right Whale Listening Network. The main project objective is to perform trade studies with different sensor combinations to arrive at a solution which is accurate, cost-effective and minimally invasive with respect to the existing ship electronics. The main innovative aspects of the project include: (i) Suitable modeling of sensor uncertainty, (ii) Data association to remove clutter, (iii) Judicious fusion of available information, (iv) Development of detection logic which increases detection confidence and minimizes false positives; and (v) Suitable performance metrics development and generation of ground truth data with respect to which different sensor combinations will be compared. The feasibility of the proposed approach will be demonstrated by leveraging SSCIs mature sensor fusion and selection technology, Woods Hole Oceanographic Institutes successfully-demonstrated whale detection approach based on thermal imaging, and Olin Colleges expertise in whale observation and sample taking using unmanned quadrotors.