Hu-Moment-Based Autonomous Landing of a UAV on a Hemispherical Dome

Abstract

In this paper, a vision-based autonomous landing system for a fixed wing unmanned aerial vehicle (UAV) is proposed for landing on a three-dimensional structure, which acts as an arrested landing mechanism and provides a strong visual cue for the camera to be detected easily. In particular, a red-colored hemispherical inflated air-bag (dome) has been considered as the visual cue in this paper. Moment-based shape descriptor called Hu-moments are leveraged for accurate detection of the dome. Characterization of these moments with horizontal and vertical distance of the UAV from the dome that is used to reliably detect the dome even at large distances is performed using software experiments. The proposed algorithm needs only a monocular camera and a processing unit on-board and hence is cost-effective and also applicable in GPS-denied environments. The proposed algorithm is simulated in a combined environment of V-Realm builder and MATLAB. Simulation results are presented to validate the presented algorithm for autonomous landing. This algorithm is also easily extendable to different colors and shapes of 3D structures.