Mover and Alert Detection and Hardware Accelerated Target Tracking using Efficient Resources (MAD-HATTER)

The Navy is developing and deploying day/night Wide Area Airborne Sensor (WAAS) payloads for tactical UASs, which can enable intelligence, surveillance, and reconnaissance (ISR) tasks to provide valuable situational awareness over large areas. However, the size, weight, and power constraints of tactical UAS platforms make onboard WAAS ISR processing impossible currently. Our Mover and Alert Detection and Hardware Accelerated Target Tracking using Efficient Resources (MAD-HATTER) solution is an integrated hardware/software system designed to process WAAS imagery in real-time onboard a UAS. MAD-HATTER employs a revolutionary computing hardware architecture which uses slightly reduced precision arithmetic to offer several orders of magnitude greater efficiency over conventional processors in terms of throughput vs. power consumption. In Phase I, we determined that MAD-HATTER will run mover detection on Singulars accelerator with three orders of magnitude improvement in speed/power compared to a CPU. In testing, MAD-HATTER achieved high probability of detection and low false alarm rate using both floating point and reduced precision arithmetic. During Phase II, we will develop a full-scope MAD-HATTER prototype using emulated hardware and optimized tracking software. We will extend MAD-HATTERs ISR capabilities to include classification and multiple target tracking.