摘要
在UAV着陆的导航与控制中引入了机器视觉方法,提出了机器视觉辅助的UAV半自主着陆系统方案,该系统由图像获取与处理、导航解算和地面平台控制模块组成。图像获取模块使用多光谱摄像机和激光测距仪,分别获得UAV的图像和深度信息。图像处理模块使用目标分割和运动分析方法,测量UAV位置。导航解算模块使用Kalman滤波器抑制测量噪声,输出滤波结果用于导航,预测结果控制地面平台跟踪UAV。实时地实现了上述机器视觉算法,并结合虚拟现实环境建立了仿真系统。使用反射内存网络连接该仿真系统与各外部模块,进行了着陆过程联合仿真。结果表明,该系统能够满足UAV半自主着陆导航任务的要求。
Machine Vision was introduced into the navigation for UAV landing, and then a vision-aided semi-autonomous landing system was proposed. It was composed of four modules, performing image acquisition, image processing, navigation calculation, and platform control respectively. A multi-spectral camera and a laser range finder acquired the image and depth of the UAV. The image processing module calculated the position of the UAV with image segmentation and motion analysis methods. Kalman filtering in the navigation calculation rejected noises in position measurements, while simultaneously output filtered position for navigation and forecast position for visual tracking. Machine Vision algorithms were realized in real-time and a simulation system was established with virtual reality environment. Using reflective memory network, the simulation system was connected with other modules to fulfill distributed simulation of the landing procedure. In conclusion, the distributed simulation proved that the vision-aided semi-autonomous landing system was valid for UAV landing.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2010年第A01期66-69,共4页
Journal of System Simulation
关键词
机器视觉
半自主着陆
虚拟现实
联合仿真
machine vision
semi-autonomous landing
virtual reality
distributed simulation
