无人机视觉导航相对位置解算及二义性优化方法  

Relative Position Calculation and Ambiguity Optimization Method for UAV Visual Navigation

作　　者：齐艳霞 庄羽 李岳伦 梅志强 付英华 李冲 QI Yan-xia;ZHUANG Yu;LI Yue-lun;MEI Zhi-qiang;FU Ying-hua;LI Chong(Henan Jiuyu Bohui Fangzhou Consulting Development Co.,Ltd,Zhengzhou 450052;School of Automation,Beijing Institute of Technology,Beijing 100081)

机构地区：[1]河南九域博慧方舟咨询发展有限公司,郑州450052 [2]北京理工大学自动化学院,北京100081

出　　处：《导航与控制》2023年第2期68-75,共8页Navigation and Control

摘　　要：针对卫星拒止条件下无人机高精度相对导航的需求,开展了视觉导航相对位置解算方法研究。设计了具有同心圆特征的合作标识,以提升不同距离情况下合作标识检测的鲁棒性。通过椭圆模型将相对位置解算问题转换为高次方程求解问题,实现了像素坐标系下椭圆方程的拟合。提出了基于特征点的相对位置初始化方法和基于运动约束的相对位置跟踪方法,解决了视觉相对位置解算过程中存在的二义性问题,实验结果表明:近距离情况下,所提方法的相对导航解算精度达到厘米级;远距离情况下,所提方法能较好地跟踪相机运动,保持较高的解算精度和鲁棒性,克服了基于二维码ArUco方法失效的不足。Aiming at the requirement of UAV high-precision relative navigation under GNSS-denied conditions,the research on the relative position solution method for visual navigation is carried out in this paper.The cooperative marker with concentric circles is designed to improve the robustness of cooperative marker detection at different distances.The ellipse model is used to convert the relative position solving problem into a higher order equation solving problem,and the elliptic equation fitting in the pixel coordinate system is realized.In addition,the relative position initialization method based on feature points and the relative position tracking method based on motion constraints are proposed to solve the problem of ambiguity in the visual relative position solving process.The experiment results show that the relative navigation solution accuracy of the proposed method reaches the centimeter level in the near distance case.What's more,in the long distance case,the proposed method can better track the camera motion,maintain high solution accuracy and robustness,and overcome the deficiency of the method based on 2D code ArUco.

关 键 词：视觉相对导航 椭圆特征 位置解二义性 无人机 卫星拒止 

分 类 号：V249.3[航空宇航科学与技术—飞行器设计]