Drogue Position Measurement of Autonomous Aerial Refueling Based on Embedded System
Drogue Position Measurement of Autonomous Aerial Refueling Based on Embedded System
复制标题
DOI：
10.1016/j.sna.2023.114251
复制
发表时间：
2023-02
期刊：
影响因子：
--
通讯作者：
Yang Gao;Shoubo Yang;Xiaolei Liu
中科院分区：
文献类型：
--
作者：
Yang Gao;Shoubo Yang;Xiaolei Liu
在自主空中加油中，受气流的影响，快速运动的锥管位置变化很大。机载位置的实时测量是加油部件成功对接的关键。针对上述问题，提出了一种融合图像检测、图像跟踪、预测和双目立体视觉测量算法的快速鲁棒定位方法。同时，搭建了现场可编程门阵列（FPGA）和数字信号处理器（DSP）的嵌入式终端，并对所提出的方法进行了部署。硬件系统集成了图像采集模块和图像处理模块。设计了异构多核平台上图像流的交互路径，并实现了并行加速算法。我们的算法和嵌入式系统经受住了恶劣的环境测试，并表现出良好的效果相比，最先进的跟踪器的无人机（UAV）的锥套视频。
In Autonomous aerial refueling (AAR), the position of the rapid moving drogue varies greatly by the influence of airflow. Real-time airborne position measurement is the key to the successful docking of refueling components. Aiming at the problems above, a fast and robust positioning method is proposed which fuses image detection, image tracking, prediction, and binocular stereo vision measurement algorithms. Meanwhile, we built an embedded end of field-programmable gate array (FPGA) and digital signal processor (DSP), and deployed the proposed method. The hardware system integrates image acquisition module and image processing module. We designed interactive paths of image streams on the heterogeneous multicore platform, and implemented parallel acceleration algorithm. Our algorithms and embedded system have withstood harsh environmental test and performs good results on the drogue videos of unmanned aerial vehicle (UAV) compared with state-of-the-art trackers.