Using computer vision algorithms for the automatic land- ing of unmanned aerial vehicles

Authors: Kosova A. E., Korikov A. M.

Annotation: A block diagram of the control system for a mobile unmanned aerial vehicle (mobile UAV) is proposed - a quadrocopter and computer vision algorithms for automatic landing of mobile UAV are selected. The computer vision algorithms adapted to the problem of automatic landing of the mobile UAV allow to effectively solve this problem. The algorithm contains three stages: the definition of the target area using the Camshift algorithm; detection of attributes in the area of the marker of the landing pad on the basis of the SURF algorithm; matching the selected landing marker template and the target area in the frame. The results of the experiments and theoretical analysis indicate the effectiveness of computer vision algorithms for automatic landing of mobile UAV. Experiments confirm that the mobile UAV can realize autonomous navigation without using a global positioning system GPS or GLONASS.

Keywords: unmanned aerial vehicles, computer vision algorithms, sift, surf, camshift, automatic uav landing