The cross-sectional width of highways is a major factor that affects the construction cost of engineering projects.With the increasing demand for intensive highway construction,research on dedicated lanes or roadway f...The cross-sectional width of highways is a major factor that affects the construction cost of engineering projects.With the increasing demand for intensive highway construction,research on dedicated lanes or roadway for cars has attracted research attention.The lateral oscillation value of vehicle’s trajectory is the direct factor that affects the lane width;however,relevant research is relatively limited,and the characteristics are not yet clear.Therefore,this study utilized an integrated radar shorter video system(IRVS)to collect real high-precision trajectory data,obtaining 24697 datasets.Statistical methods were used to reveal the lateral oscillation value and determine the effective lane width of cars for safe driving.The research results are as follows.1.The lateral oscillation value of cars varies across different lanes.Vehicles on the inner two lanes tend to drift leftward,whereas those on the outermost lane tend to drift rightward.2.When the operating speed ranges from 80 km/h to 110 km/h,the lateral oscillation value presents a statistical rule,which obeys a normal distribution for the left and right margins,i.e.,D_(left)~N(0.87,0.15)and Dright~N(0.72,0.15).This lateral oscillation value is insensitive to fluctuations in the operating speed(80 km/h~110 km/h)from the standpoint of field data.3.The fitted expressions of effective lane width for cars were proposed,and the effective lane width at the 95th,90th,and 85th percentile is 3.2 m,3.0 m,and 2.8 m,respectively.These results can provide technical support for dedicated lane width for cars,and have practical significance for the intensive construction of road infrastructure in highly urbanized areas.展开更多
基于激光雷达的车道线检测目前使用最多的是基于雷达扫描点密度的检测方法,但它的抗干扰能力差.为此,本文利用激光雷达的回波脉冲宽度对于车道线与路面的区分度进行特征提取,提出一种特征提取方法,分两步进行——基于脉冲宽度动态阈值...基于激光雷达的车道线检测目前使用最多的是基于雷达扫描点密度的检测方法,但它的抗干扰能力差.为此,本文利用激光雷达的回波脉冲宽度对于车道线与路面的区分度进行特征提取,提出一种特征提取方法,分两步进行——基于脉冲宽度动态阈值的种子点提取和基于高斯核加权搜索的区域生长.然后引入FCL(fuzzy C-means of line)算法识别车道线(以线为中心进行聚类),最后通过最小二乘法拟合车道线.通过实车在6个不同的道路场景下进行实验,都能够准确检测出车道线,同时具有较高的检测精度.展开更多
基金supported by the National Key R&D Program of China(Nos.2017YFC0803907,2017YFC0803900,and 2014BAG05B02).
文摘The cross-sectional width of highways is a major factor that affects the construction cost of engineering projects.With the increasing demand for intensive highway construction,research on dedicated lanes or roadway for cars has attracted research attention.The lateral oscillation value of vehicle’s trajectory is the direct factor that affects the lane width;however,relevant research is relatively limited,and the characteristics are not yet clear.Therefore,this study utilized an integrated radar shorter video system(IRVS)to collect real high-precision trajectory data,obtaining 24697 datasets.Statistical methods were used to reveal the lateral oscillation value and determine the effective lane width of cars for safe driving.The research results are as follows.1.The lateral oscillation value of cars varies across different lanes.Vehicles on the inner two lanes tend to drift leftward,whereas those on the outermost lane tend to drift rightward.2.When the operating speed ranges from 80 km/h to 110 km/h,the lateral oscillation value presents a statistical rule,which obeys a normal distribution for the left and right margins,i.e.,D_(left)~N(0.87,0.15)and Dright~N(0.72,0.15).This lateral oscillation value is insensitive to fluctuations in the operating speed(80 km/h~110 km/h)from the standpoint of field data.3.The fitted expressions of effective lane width for cars were proposed,and the effective lane width at the 95th,90th,and 85th percentile is 3.2 m,3.0 m,and 2.8 m,respectively.These results can provide technical support for dedicated lane width for cars,and have practical significance for the intensive construction of road infrastructure in highly urbanized areas.
文摘基于激光雷达的车道线检测目前使用最多的是基于雷达扫描点密度的检测方法,但它的抗干扰能力差.为此,本文利用激光雷达的回波脉冲宽度对于车道线与路面的区分度进行特征提取,提出一种特征提取方法,分两步进行——基于脉冲宽度动态阈值的种子点提取和基于高斯核加权搜索的区域生长.然后引入FCL(fuzzy C-means of line)算法识别车道线(以线为中心进行聚类),最后通过最小二乘法拟合车道线.通过实车在6个不同的道路场景下进行实验,都能够准确检测出车道线,同时具有较高的检测精度.
