无人车的道路边界检测研究

发布时间：2018-01-20 16:26

本文关键词： 无人车边界检测激光雷达聚类双阈值　出处：《西安工业大学》2016年硕士论文　论文类型：学位论文

【摘要】：近年来车辆技术的发展使得无人驾驶技术成为国内外研究的热点,无人车的道路边界检测是无人驾驶技术的重要研究方向之一,道路边界检测就是要识别出道路上的障碍物,判别出道路边界,提取出无人车在道路的可行区域,保证无人车可以在未知的道路环境中安全行驶。本文首先根据无人驾驶智能车系统的要求搭建了无人车的硬件平台,安装了所需的各类传感器。然后对所安装的激光雷达传感器和GPS/INS惯性导航的通讯协议以及数据进行了比较详细的介绍说明。最后,实现了工控机与无人车上安装的各传感器之间的通讯,从而得到激光雷达、GPS/INS惯性导航的检测数据。其次,结合现实道路环境的具体情况,在现有的三维激光测距系统使用场合、特点等基础上,分析了本文所采用的三维激光测距系统是由激光雷达配合无人车的行驶来进行扫描,并且建立了系统所用的激光雷达局部坐标系、无人车局部坐标系和无人车全局坐标系这三种坐标系。文中提出一种在无人车行驶中基于激光雷达的道路边界检测方法。确定雷达数据点的搜索原则,通过最近邻域算法对激光雷达检测的数据进行聚类；再经过双阈值结合中位值平均滤波算法对激光雷达采集的道路边界信息进行处理,最后通过最小二乘法拟合出道路边界,进而估算出道路车道线信息,为无人车的行驶提供安全可靠的保障。最后通过Visual Basic软件平台结合无人车的运行对道路边界检测算法做了相应的测试实验。实验结果表明：激光雷达能够对道路边界数据进行有效的聚类,同时实现了对无人车道路边界和侧方障碍物的准确检测。
[Abstract]:In recent years, with the development of vehicle technology, driverless technology has become a hot spot at home and abroad. Road boundary detection of unmanned vehicles is one of the important research directions of driverless technology. Road boundary detection is to identify the obstacles on the road, identify the road boundary, and extract the feasible area of the road. Ensure that the unmanned vehicle can drive safely in the unknown road environment. Firstly, the hardware platform of the unmanned vehicle is built according to the requirements of the unmanned intelligent vehicle system. Then the communication protocol and data of lidar sensor and GPS/INS inertial navigation are introduced in detail. The communication between the industrial control computer and the sensors installed on the unmanned vehicle is realized, and the detection data of GPS / ins inertial navigation are obtained. Secondly, combined with the actual road environment. On the basis of the application and characteristics of the existing 3D laser ranging system, it is analyzed that the 3D laser ranging system used in this paper is scanned by the laser radar combined with the driving of the unmanned vehicle. And the local coordinate system of lidar used in the system is established. In this paper, a method of road boundary detection based on lidar is presented in this paper, and the search principle of radar data points is determined. The nearest neighborhood algorithm is used to cluster the data of lidar detection. Then the road boundary information collected by lidar is processed by double threshold and mean median filter algorithm. Finally, the road boundary is fitted by the least square method, and then the road lane information is estimated. To provide a safe and reliable guarantee for the driving of the unmanned vehicle. Finally, through the Visual. The Basic software platform is used to test the road boundary detection algorithm combined with the running of the unmanned vehicle. The experimental results show that the lidar can effectively cluster the road boundary data. At the same time, it realizes the accurate detection of the road boundary and side obstacles of the unmanned vehicle.
【学位授予单位】：西安工业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：U463.6;TN958.98

【相似文献】