摘要
为研究驾驶人的跟车特性及探究可适用于不同风格驾驶人的跟车预警规则,为自动驾驶车辆开发可满足不同用户驾驶需求和驾乘体验的主动安全预警系统,选取50名被试驾驶人开展实车试验,采集驾驶人跟车行为表征参数并基于雷达数据确定跟车事件提取规则。选取平均跟车时距和平均制动时距为二维向量,使用基于K-means聚类结果的高斯混合模型将驾驶人聚类为3种风格类型(冒进型、平稳型、保守型)。通过分析3组驾驶人的跟车及制动数据,将不同类型驾驶人的制动时距分位数作为跟车预警阈值,结合实际预警数据及不同制动时距分位数对应的预警正确率,对现有跟车预警规则进行调整,以适应不同类型驾驶人的驾驶需求。研究结果表明:3组驾驶人的平均跟车时距和平均制动时距差异显著,冒进型驾驶人倾向于选择较小的跟车时距和制动时距,保守型驾驶人的跟车时距和制动时距则普遍较大;3组驾驶人的实际跟车预警次数为215次,驾驶人采取制动操作而系统未予以预警的次数为329次,系统整体预警正确率为21.9%,漏警率为87.5%,通过分析信息熵等判定当前预警规则并不合理;将每类驾驶人制动时距的10%分位数作为阈值时的预警效果较好,调整后的跟车预警规则能在一定程度上适应不同的驾驶人类型。
The research of early warning and control strategy considering driving style is a hotspot in the field of automatic driving.To study driver’s car-following characteristics and explore whether the warning rules can be applied for different driving styles and further develop active safety warning systems for self-driving vehicles to meet the needs of different users,50 participants were recruited to carry out a real road test.The characterization parameters for the analysis of car-following behaviors were collected and the extraction rules of the car-following events were determined based on radar data.The Gaussian mixture model method with the inputs of the K-means algorithm was used for clustering and drivers were classified into three types of driving styles(aggressive drivers,calm drivers,and conservative drivers)on the basis of twodimensional variables:average time gap and average time gap when braking.By analyzing the carfollowing data and braking characteristics of the three groups of drivers,combined with the real warning data and the warning rate of time gap when braking in different percentiles,the time gap when braking in the percentile format was suggested as the warning threshold afterward.Accordingly,the warning rules were adjusted to meet the needs of different types of drivers.The results show that there is a significant difference between the time gap and time gap when braking for three groups of drivers.Aggressive drivers tend to keep a smaller time gap or time gap when braking,while conservative drivers usually choose a larger value.The number of early warnings from three groups of drivers is 215,the number of times when the brake pedal is pushed by a driver while the system failed to give a warning was 329,the overall warning rate of the existing system is 21.9%,and the missed alarms account for 87.5%.Warning rules are unreasonable and should be reset considering driving styles after analyzing the information entropy.The result is satisfying when using a 10% percentile of time gap when braking
作者
刘通
付锐
马勇
刘卓凡
程文冬
LIU Tong;FU Rui;MA Yong;LIU Zhuo-fan;CHENG Wen-dong(School of Automobile,Chang'an University,Xi’an 710064,Shaanxi,China;Key Laboratory of Automobile Transportation Safety Technology,Ministry of Transport,Chang'an University,Xi'an 710064,Shaanxi,China;Modern Postal School,Xi’an University of Posts&Telecommunications,Xi’an 710061,Shaanxi,China;School of Mechatronic Engineering,Xi'an Technological University,Xi'an 710021,Shaanxi,China)
出处
《中国公路学报》
EI
CAS
CSCD
北大核心
2020年第2期170-180,共11页
China Journal of Highway and Transport
基金
国家重点研发计划项目(2018YFB1600500)
国家自然科学基金项目(51775053,51908054)
教育部长江学者和创新团队发展计划项目(IRT_17R95)
陕西省自然科学基础研究计划项目(2018JM5158)
陕西省教育厅专项科研计划项目(19JK0788)。
关键词
汽车工程
跟车预警规则
高斯混合模型
驾驶风格
驾驶人
高级驾驶辅助系统
automotive engineering
car-following warning rule
Gaussian mixture modeling
driving style
driver
advanced driving assistance system
