Proactive guidance for accurate quadrotor-based UAV landing on dynamic platform

Abstract

The ever-burgeoning development of unmanned aerial vehicles (UAVs) within the past decades has provided people a promising solution in various applications, such as package delivery, autonomous infrastructure, and so on; although the many endeavors from engineers, the gap between theories and applications should be further bridged for real-world scenarios, where autonomous landing being one of the crucial aspects. Due to the complexity and uncertainty of the surroundings, the landing task of a UAV has been deemed to be the most critical and fragile stage of the whole flight mission. Hence, this thesis aims to improve the current state-of-the-art autonomous landing methods, in which pitch variation, visual-inertial based perception, path planning, and trajectory optimization are comprehensively studied. First of all, to enhance control performance and energy efficiency, a variable-pitch propeller (VPP) system and a novel control allocation method are introduced. The control allocation method was firstly verified in a simulation environment via mathematical models and was then implemented on a flight controller and experimented with in a motion-capture arena. The results show that the proposed method improves the yaw tracking performance and demonstrates an improvement in energy consumption through various pitch angles. In addition to the above, for the autonomous landing system of UAVs, a novel system configuration is presented. The proposed design, unlike most state-of-the-art UAV landing frameworks (that rely on UAV onboard computers and sensors), fully depends on the computation unit situated on the ground vehicle/marine vessel to serve as a landing guidance system. Such a novel configuration can lighten the burden of the UAV, whilst its computation power is enhanced. Specifically, a sensor fusion-based algorithm for the guidance system to perform UAV localization is utilized, whilst a control method based upon trajectory optimization is presented. Indoor and outdoor experiments are conducted, and the results show that precise autonomous landing on a 43 cm × 43 cm platform can be performed. Finally, the conclusion contains in-progress works and future opportunities for autonomous landing system improvement.