基于可变视场角的空中加油锥套位姿精确测量方法  

Accurate Measurement of Refueling Drogue Pose Based on Variable Field Angle

作　　者：王宏伦[1,2] 阮文阳 王延祥 吴健发 左芝勇[5] 康荣雷[5] Wang Honglun;Ruan Wenyang;Wang Yanxiang;Wu Jianfa;Zuo Zhiyong;Kang Ronglei(School of Automation Science and Electrical Engineering,Beihang University,Beijing 100191,China;Science and Technology on Aircraft Control Laboratory,Beihang University,Beijing 100191,China;Beijing Aerospace Institute for Metrology and Measurement,Beijing 100076,China;Shenyuan Honors College of Beihang University,Beijing 100191,China;The 10th Research Institute of China Electronics Technology Group Corporation,Chengdu 610036,China)

机构地区：[1]北京航空航天大学自动化科学与电气工程学院,北京100191 [2]北京航空航天大学飞行器控制一体化技术重点实验室,北京100191 [3]北京航天计量测试技术研究所,北京100076 [4]北京航空航天大学高等理工学院,北京100191 [5]中国电子科技集团公司第十研究所,成都610036

出　　处：《战术导弹技术》2020年第4期135-143,共9页Tactical Missile Technology

基　　金：国家自然科学基金项目(61673042)。

摘　　要：针对无人机自主空中加油近距视觉高精度导航问题,提出了一种基于可变视场角的空中加油锥套相对位置和姿态精确测量方法。首先构建了由多组不同固定视场角相机组成的双目视觉测量系统,提出了基于非线性滞环特性的相机组切换策略,有效克服了随着距离增加锥套目标在相机图像中比例逐渐减小而导致测量精度降低的问题。在此基础上,使用深度学习YOLO v2算法对首帧图像进行检测与识别,截取感兴趣区域(ROI)并进行图像处理,提取标志灯质心坐标。然后根据双目视觉原理进行三维重建,以计算出标志灯的空间位置。最后根据解析几何关系解算得到锥套相对相机的位置和姿态信息。空中加油对接过程视景仿真和地面实物试验表明,本方法在设定的测量范围内均能达到较高的相对位置和姿态测量精度,且满足实时性要求。Aiming at the problem of autonomous aerial refueling(AAR)visual navigation for UAVs,a real-time and precise method for the drogue position and attitude measurement based on variable field of view is proposed.First,a binocular vision measurement system composed of several cameras with different fixed field angles is constructed,and a camera switching strategy based on the nonlinear hysteresis characteristic is proposed.The proposed system and strategy overcome the problem of the decreasing measurement accuracy as the proportion of taper sleeve in the image decreases with the increasing distance.On this basis,the deep learning algorithm,YOLO v2,is used to detect and recognize the first frame of image.Taking region of interest(ROI)and performing image processing to extract the centroid coordinates of the marker light.Then,according to the principle of binocular vision,three-dimensional reconstruction is performed to calculate the position of the marker light.Finally,the pose of the drogue is calculated according to the analytical geometric algorithm.Visual simulation and prototype tests of the docking process in AAR show that the proposed method has a high accuracy for measuring the relative position and attitude of the drogue in a certain range and meets the real-time requirements.

关 键 词：无人机 自主空中加油 锥套位姿测量 可变视场角 双目视觉 深度学习 

分 类 号：V249[航空宇航科学与技术—飞行器设计]