The U.S. Army has demonstrated a machine learning-based sensor technology designed for fixed-wing aircraft that works with synthetic aperture radars to detect explosives from a designated safety point.
As part of the 15-month Phase I demonstration, the Army Combat Capabilities Development Command and the Army Research Laboratory (ARL) assessed the multisensor technology’s capacity to use artificial intelligence and ML to collect real-time target data, the service branch said Wednesday
The team used airborne SARs, electro-optical and infrared radars, a junction detection radar and Light Detection and Ranging technologies for ground vehicles and small unmanned aerial vehicles.
Detection algorithms used for the exercise involved geo-referencing and pixel-alignment techniques to enable threat detection through an augmented reality approach.
Lt. Col. Mike Fuller, program manager at the Defense Threat Reduction Agency, said the team used side-by-side comparisons of multiple modalities during the assessment to maximize the probability of threat detection while minimizing false alarms.
For Phase II, the Army intends to downselect a sensor system ahead of a final demonstration on various types of terrain. DTRA funds the three-year effort through the Blood Hound Gang Program.
