Research on flare path generation for improving automatic landing accuracy of fixed-wing UAV

Abstract

The flight consists of 5 phases : take-off, climb, cruise, approach and landing. The landing phase occupies the lowest flight operating time, but the highest accident rate and it is one of the most difficult part in flight. Most recently developed UAVs have an automatic landing capability to meet the level 3 automation level defined by DARPA. For manned aircraft, pilots are trained to prepare for emergencies during landing, but for UAVs, it should rely on solely on the flight control system. As a result, researches on automatic landing technique of UAV have been actively conducted for the past decade. In particular, intensive research has been carried out in areas of control techniques and navigation accuracy improvement using image. However, it is true that most of the studies have risks about applying to UAVs operating in actual field. Besides control techniques and vision based techniques, this thesis proposes other method to improve the accuracy of automatic landing of Fixed-Wing UAV. The proposed method is that generating a flare reference trajectory on-line fast using the current POSE information of the UAV and the geometrical information of the runway right before entering the flare phase. PID control technique used in most UAVs was used and additional navigation devices that enhance POSE information were not considered. Moreover, the proposed flare reference trajectory is applied to a UAV model similar to the tactical class UAV operating in military and compared with flare reference trajectory used in practice. In addition, the results for automatic landing accuracy of the proposed flare reference trajectory and flare reference trajectory used in practice were compared and analyzed in consideration of the error of the navigation sensors used in the automatic landing and the atmospheric disturbance conditions that may occur in the landing environment.