A data-driven approach for monitoring forward velocity for small and lightweight drone

For low altitude unmanned aerial vehicle (UAV) applications, because of the strong atmosphere turbulence, the flight track and velocity may seriously deviate from the predetermination. This may cause severe degradation for achieving the expected performance of the UAV system. In this paper, we mainly focus on UAV-Radar applications, and a novel approach for estimating the drone's forward velocity variation is proposed based on the raw radar data. By using sub-aperture map-drift (MD) method, the Doppler rate variation due to the drone's flight deviation is estimated first. Then, an adaptive filter is presented, which is capable of extracting the Doppler rate variation of the forward velocity, and simultaneously, that of the radial acceleration. Therefore, a time-varying forward velocity can be obtained for the drone without any airborne inertial navigation assistance. A raw data set collected by a UAV radar system also with canonical navigation information is applied for the verification of the proposed approach.