Advanced Machine Learning Target Recognition in Munitions

Polaris Sensor Technologies proposes a gun-launched intelligent munition concept for both fixed and moving targets which uses small, low cost sensors. It exploits sensor fusion and polarization measurement to obtain object detection candidates at extended ranges. An Automatic Target Recognition algorithm is used to confirm that the correct candidates are targeted, and employs a new classifier architecture which allows switching target sets in just minutes. Targets may be specified by map coordinates or relative to local terrain features and buildings. Moving targets are detected over the area of uncertainty and their motion is tracked as the munition approaches. The system is can be used in most weather conditions, day or night, and operates in the absence of GPS.