Vision-Based Precision Landings of a Tailsitter UAV

Abstract

In this paper we present a method used to land a tailsitter unmanned air vehicle (UAV) in hover flight, on a ground target that is detected using an onboard camera. This method does not use Global Positioning System (GPS) data to calculate the aircraft’s relative position to the target. Instead, video from an onboard camera, the aircraft’s orientation, and the aircraft’s altitude above ground level (AGL) are used to calculate the aircraft’s position relative to a ground target. Because it is important that the ground target stay in the camera’s view to calculate the UAV’s position, a method of calculating the aircraft’s range of attitudes that allow the ground target to remain in view is also presented. The method described in this paper has been implemented in simulation with successful results. Nomenclature XUAV Calculated x location of the UAV in the target coordinate frame YUAV Calculated y location of the UAV in the target coordinate frame qaircraft Quaternion describing the aircraft’s orientation expressed in the inertial frame qdesired Quaternion describing the desired aircraft orientation expressed in the inertial frame qerror Quaternion describing rotation from qaircraft to qdesired expressed in the body frame αaz Gimbal azimuth angle αel Gimbal elevation angle