AUTONOMOUS QUADROTOR UAV SYSTEMS FOR DYNAMIC PLATFORM LANDING WITH ONBOARD SENSORS

摘要

For autonomous landing objective of an unmanned aerial vehicle(UAV), environmental interaction can be improved with pose estimation and motion planning. In this paper, using an onboard camera, an autonomous dynamic landing system is designed with respect to a UAV platform, which contains a comprehensive framework and is integrated on an onboard computer. Specifically, pose estimation for the monocular camera is conducted based on visual-inertial localisation algorithms, and target detection algorithms are utilised for landing target recognition. A pose transformation algorithm is designed and target information is conversed, and landing trajectories are generated and even regenerated with respect to moving targets by motion planning with provided state information. An outer-loop geometric tracking controller for the UAV is integrated to complete landing tasks by calculating the required thrust and attitude. To construct a capable and extendible experimental platform, a comprehensive flight system is designed above the robot operating system (ROS) including the landing objective, and the feasibility of the proposed method is verified by experiments for both static and dynamic landing targets.