High-Precision Underwater Multi-Target Tracker (HUMTER)

We propose a forward-looking solution that uses machine learning to implement a tracker for Air Anti-Submarine Warfare (ASW) systems to improve tracking metrics when multiple targets (or target-like objects) are present in wide-area search missions. This approach will improve the processing capability on P-8A aircraft without requiring significant changes to how the Navy currently operates. Leveraging our expertise in machine learning model development for tracking systems, we propose an AI-powered tracking system with the potential to initiate tracks with just up to 4 detections while maintaining global false alarm rates well below 2%. Our solution will be packaged in a highly compressed, mostly hardware agnostic deployment format that in addition to being deployed on P-8A aircraft could in the future also be deployed at the tactical edge in higher resource-constraint environments such as at existing sensor nodes. Our proposed approach represents an adaptation to ASW and further improvement to a machine learning based, 3D tracking system that has been under development and testing by us for over three years for the US Marine Corps System Command.