基于无轨电车避撞的雷达目标识别和跟踪技术研究

发布时间：2018-01-10 18:30

  本文关键词：基于无轨电车避撞的雷达目标识别和跟踪技术研究　出处：《北京交通大学》2017年硕士论文　论文类型：学位论文

  更多相关文章： 无轨电车 激光雷达 栅格地图 扩展卡尔曼滤波算法/最近邻算法组合 信息采集软件 实时显示软件

【摘要】：近年来,许多大中型城市大力建设公共交通系统,随着“绿色交通”理念的提出,无轨电车作为新能源公共交通工具,逐渐有复兴之势。但无轨电车有车体宽,车身长,转弯半径大等缺点,导致车辆有很多视觉死角,容易发生交通事故。车载激光雷达可以探测车前情况,识别车前障碍物的大小并判断障碍物的运动状态。将激光雷达技术与无轨电车结合,研究无轨电车目标识别与跟踪技术有助于提高车辆行驶安全性。本文主要目的研究可应用于无轨电车避撞的激光雷达目标识别与跟踪的技术方案,通过相关算法完成对目标识别与跟踪功能的实现。(1)本文分析了无轨电车的国内外发展状况,并对障碍物目标识别和目标跟踪技术分别进行阐述。接着分析无轨电车障碍物目标识别和目标跟踪的功能需求和系统总体设计,重点说明了目标的分割聚类和目标跟踪,分割聚类采用栅格算法,对激光雷达散点进行聚类处理,建立车前目标分类,目标跟踪采用扩展卡尔曼滤波算法和最近邻算法组合,判别聚类数据结果的状态,对其中的目标障碍物进行跟踪。(2)本文阐述了无轨电车避撞系统的结构,并对其中的信息采集模块、目标识别模块和目标跟踪模块进行了说明。对激光雷达信息采集软件和激光雷达实时显示软件分别进行软件的功能和结构设计,采用C++编程语言完成软件。(3)对无轨电车障碍物识别和目标跟踪系统进行实车测试与验证,并对实验结果进行分析。通过实车采集信息与测试方案信息对比,验证了栅格算法和扩展卡尔曼滤波算法和最近邻算法组合的有效性。通过对软件各功能的测试,实现了车前环境信息采集、车前环境实时雷达显示、车前环境实时视频显示,车前障碍物的聚类处理,验证了软件的可用性。实验结果表明,论文提出的无轨电车障碍物识别和目标跟踪系统能够准确采集车前环境信息,能够实现障碍物识别和目标跟踪,实时准确的为驾驶员提供车辆预警信息。
[Abstract]:In recent years, many large and medium-sized cities have made great efforts to build public transport system. With the concept of "green transportation", trolleybus as a new energy public transport, gradually has the potential to revive, but trolleybus has a wide body. Long body, large turning radius and other shortcomings, resulting in a lot of vehicle visual dead-angle, prone to traffic accidents. Vehicle lidar can detect the car in front of the situation. Recognize the size of the obstacle in front of the vehicle and judge the motion of the obstacle. Combine lidar technology with trolley bus. The research on target recognition and tracking technology of trolleybus is helpful to improve the safety of vehicle running. The main purpose of this paper is to study the technical scheme of lidar target recognition and tracking which can be applied to collision avoidance of trolleybus. This paper analyzes the development of trolley bus at home and abroad. Then the functional requirements and the overall system design of obstacle target recognition and target tracking for trolleybus are analyzed. The segmentation clustering and target tracking are emphasized. The raster algorithm is used to cluster the scattered points of lidar, and the target classification in front of vehicle is established. Target tracking uses extended Kalman filter algorithm and nearest neighbor algorithm to judge the status of clustering data results and track the target obstacles. 2) this paper describes the structure of trolley bus collision avoidance system. The information acquisition module, target recognition module and target tracking module are described. The function and structure of the software are designed for the laser radar information acquisition software and the laser radar real-time display software. C programming language is used to complete the software. 3) to test and verify the obstacle recognition and target tracking system of trolley bus. The results of the experiment are analyzed. The information collected by the real vehicle is compared with the information of the test scheme. The validity of the combination of grid algorithm and extended Kalman filter algorithm and nearest neighbor algorithm is verified. Through testing the functions of the software, the acquisition of environment information in front of vehicle and the real-time radar display in front of vehicle environment are realized. Real time video display in front of vehicle environment and clustering processing of obstacles in front of vehicle verify the usability of the software. The experimental results show that. The obstacle recognition and target tracking system of trolley bus proposed in this paper can accurately collect the environment information in front of vehicle, realize obstacle recognition and target tracking, and provide early warning information for driver in real time.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U482.2;TN958.98

【相似文献】

中国期刊全文数据库 前5条

1 叶炜W，

本文编号：1406378