基于跑道平面结构化线特征的固定翼无人机视觉导航算法

针对固定翼无人机在着陆阶段的位姿估计的问题，提出了一种基于跑道平面结构化线特征的无人机视觉导航方法。结构化线特征包括跑道平面上的跑道边缘及共面的垂直于跑道边缘的单条合作标志线。利用单台固连在无人机上的前视相机对跑道区域进行成像，自动提取结构化线特征。在无人机降落前期利用完整的结构化线特征配置解算出无人机的六自由度位姿参数(偏航角、俯仰角、滚转角、纵向位置、横向位置、高度)，并在无人机降落到较低高度时，利用退化的结构化线特征(跑道边缘)解算出无人机的关键位姿参数(偏航角、俯仰角、横向位置、高度)。三维实景仿真实验证明，在距离机场距离为200m处，无人机的距离参数精度为<0.5m，角度参数精度为<0.1°。本文的方法充分考虑到无人机自主着陆过程中的成像特点，具有自动化程度高、工程实用性好的优点。

A vision-based landing method using structured line features of runway surface for Fixed-Wing UAVs

For the estimation problem of pose and attitude of fixed-wing UAVs in the terminal landing stage, an vision-based landing method using structured line features is proposed. Structured line features in this paper are the runway edges and a single cooperation line perpendicular to the runway edges. One forward looking camera equipped in the fixed-wing UAVs is used to capture multiple pictures of the structured line features and these features are extracted automatically. 6-dof pose and attitude parameters (pitch angle, yaw angle, roll angle, longitudinal position, lateral position and altitude) are calculated using geometric constraint of full configuration structured line features in the earlier stage of the landing, the key parameters (pitch angle, yaw angle, lateral position and altitude) are calculated uisng degenerate configuration structured line features (only the runway edges) in the final stage of the landing. In the 3D stimulation experiment, accuray of distance is less than 0.5m, accuray of angle is less than 0.1° when fixed-wing UAVs are landing in the 200m away from runway. The proposed algorithm takes accout of the scene character in the landing stage of the fixed-wing UAVs, and has the advantages of high-automatic and practical.