Automatic landing assistant system based on stripe lines on runway using computer vision

Abstract

This research proposes the design and implementation of an automatic landing assistant system of fixed-wing unmanned aerial vehicle (UAV) using computer vision. In our research, the system was designed when radar or other sensing devices could not be used. For example, GPS-based guiding would be in effective in warfare. The procedure relies on a camera and an orientation sensor, combined with existing lane detection algorithms, etc. Novelties are among how to interpret the acquired data from detection to suit with a military research institute's demand/specification on the automatic landing task. UAVs have to navigate using stripes of runway for landing using only computer vision. Edge detection and Hough line transform are key algorithms for vision step. Then geometrical interpretation would be shown : stressing on relationship between a center line of runway and the plane. The experiment shows detection from a UAV and runway positioning in real time.