A New Vision-based Method of Autonomous Landing for UAVs

Abstract

This paper investigates the fundamental solution of vision-based autonomous landing for UAVs, aiming to reduce error and improve landing accuracy. Combined with visual navigation, a new landing pad and recognition algorithm based on ArUco code is designed. We apply the feature point coordinates solving into attitude estimation to enhance the anti-interference ability of the system. The error model is designed and fused to further improve the accuracy of landing. Extensive simulations and real-world experiments verify the superior performances of our method. The designed landing pad and new recognition algorithm shorten the time of target recognition and greatly improve the accuracy of autonomous landing.