Machine Learning for UAV Classification Employing Mechanical Control Information

Range-Doppler images are widely used to classify different types of Unmanned Air Vehicles (UAVs) because each UAV has a unique range-Doppler signature. However, UAV's signature depends on its movement mechanism. This is why classifier's accuracy would be degraded if the effect mechanical control system UAVs wasn't taken into consideration, which may lead non-unique while in-flight. In this paper, full-wave electromagnetic CAD tool investigate systems two quadcopters, hexacopter, and helicopter their signatures. A Mechanical Control-Based Machine Learning (MCML) algorithm introduced four UAVs. Different (ML) algorithms were applied generated datasets that considered information The Convolutional Neural Networks (CNN) provided robust performance reaching an higher than 90%.