Outdoor Autonomous Landing of a Quadcopter on a Moving Platform using Off-board Computer Vision

Abstract

 This paper presents a method that enables a quadcopter to perform autonomous landing on a moving platform using computer vision. In addition, the system implementation of the computer vision technique is presented. Unlike other researches, the camera is mounted on the moving platform instead of being installed on the quadcopter. Besides, the computer vision system is tested outdoor, and the results such as the performance and the accuracy are presented. In the stationary platform test, 5 out of 10 landings fall within 30 cm from the center. In the moving platform test, the maximum platform-moving speed for autonomous landing is 2 m/s. Hence, it is proven that this methodology is feasible. Lastly, the advantages and limitations of the computer vision technique proposed are discussed.