论文部分内容阅读
本文主要研究了在无人机空中自主加油过程中,利用安装在无人机上的摄像机得到的视觉图像确定无人机与加油机之间相对位置关系的算法。本文以经典的OI算法为基础展开研究,文章主要分为三个部分:首先简要介绍了OI算法的基本实现方法,指出了在OI算法中,旋转矩阵R的初值的选择对于位姿估计的精度有一定的影响。其次主要介绍了一种线性算法用于获取R初值的方法,同时针对空中加油过程中无人机与加油机之间的相对姿态角变化较小的特点,提出了可以将旋转矩阵R的初值设为单位矩阵。论文最后在仿真环境下对相关算法展开实验研究,同时研究了在高斯噪声环境下特征点在加油机表面的分布对位姿估计结果的影响,论文得到如下的结论:1、针对空中加油过程中无人机与加油机之间的相对姿态角变化较小的特点,在高斯噪声环境下,以单位矩阵作为旋转矩阵的初值得到的位姿估计的结果好于以线性算法为初值的结果。2、在相同的高斯噪声环境下,特征点之间的距离越大时得到的位姿估计结果的精度越高。3、特征点的数量对相对位姿估计的精度没有明显影响。
This paper mainly studies the algorithm of determining the relative position between UAV and tanker by using the visual images obtained from the cameras mounted on the UAV in the process of autonomous air refueling in the UAV. In this paper, based on the classical OI algorithm, the article is divided into three parts: Firstly, the basic implementation of OI algorithm is briefly introduced. It is pointed out that in the OI algorithm, the selection of initial value of rotation matrix R for pose estimation Accuracy has a certain impact. Secondly, the paper introduces a linear algorithm for obtaining the initial value of R, and at the same time aims at the small change of the relative attitude angle between the unmanned aerial vehicle and the tanker during air refueling, The value is set to the unit matrix. At the same time, the paper studies the influence of the distribution of feature points on the surface of the dispenser under Gaussian noise environment. The conclusions of the dissertation are as follows: 1. In the process of air refueling In the Gaussian noise environment, the result of pose estimation based on the unit matrix as the initial value of the rotation matrix is better than the result of the linear algorithm as the initial value . 2, in the same Gaussian noise environment, the greater the distance between the feature points, the higher the accuracy of the pose estimation result. The number of feature points has no obvious influence on the relative pose estimation accuracy.