网联环境下驾驶人响应机制——一种交叉路口接近行为定性效应分析与定量驾驶模式提取案例  

Driver Response Mechanism in Connected Environment—A Case Study of Qualitative Effect Analysis and Quantitative Driving Pattern Extraction of Intersection-approaching Behavior

作　　者：张海伦 许庆[1] 高博麟 王建强[1] 李克强[1] ZHANG Hailun;XU Qing;GAO Bolin;WANG Jianqiang;LI Keqiang(School of Vehicle and Mobility,Tsinghua University,Beijing 100084)

机构地区：[1]清华大学车辆与运载学院,北京100084

出　　处：《机械工程学报》2024年第10期22-39,共18页Journal of Mechanical Engineering

基　　金：国家自然科学基金(52131201);北京市自然科学基金(3244031)资助项目。

摘　　要：智能网联技术的发展为交通安全和效率的提升上提供巨大机遇。然而现有研究未能深入阐述网联环境下驾驶人对环境的认知响应机制,缺乏网联环境下驾驶模式定量分析。提出研究交叉路口接近行为过程和响应机制的方法,探索驾驶人在网联环境下的交通行为学机理。在驾驶模拟器中设计两种驾驶场景,分别为基准传统环境和对照网联环境,在对照网联环境组中,向驾驶人提供交通灯相位和当前相位状态剩余时间。采集34名驾驶人视觉交互信息、车辆运动学和驾驶人操作行为特征等参数。分析人机交互界面的交互频率和累积时间百分比、首次交互时间和响应时间,以及驾驶人接近交叉路口的行为特征。建立贝叶斯非参数方法结合文本聚类算法的驾驶模式提取模型,实现对驾驶模式定量描述。结果表明,在红灯和绿灯相位下的人机交互特性存在显著差异,首次交互时间和响应时间高度相关,网联环境可以显著提升交叉路口通行效率,改善驾驶行为。所提出的驾驶模型可以有效描述路口接近行为的6种驾驶模式,网联环境可以降低23.7%的加速行为,平稳驾驶概率提升25.0%。The development of intelligent connected technology has provided great opportunities for the improvement of traffic safety and efficiency.However,the existing research fails to elaborate the driver’s cognitive response mechanism to the environment in the connected environment,and lacks the quantitative analysis of driving patterns in the connected environment.A method for studying the intersection-approaching behavior process and response mechanism is proposed,and the traffic behavior mechanism of drivers in the connected environment is explored.Two driving scenarios are designed in the driving simulator,namely the benchmark traditional environment and the controlled connected environment.In the connected environment,the driver is provided with the traffic light phase and the remaining time of the current phase state.Parameters such as visual interaction information,vehicle kinematics,and driver operating behavior characteristics of 34 drivers are collected.The interaction frequency and cumulative time percentage of the human-machine interface,the first interaction time and response time,and the behavioral characteristics of drivers approaching intersections are analyzed.A driving pattern extraction model based on bayesian non-parametric method combined with text clustering algorithm is established to achieve quantitative description of driving patterns.The results show that there are significant differences in the human-machine interaction characteristics under red and green light phases,and the first interaction time and response time are highly correlated.The connected environment can significantly improve the efficiency of intersection traffic and improve driving behavior.The proposed driving model can effectively describe the six driving patterns of intersection-approaching behaviors,and the connected environment can reduce the acceleration behavior by 23.7%,and increase the smooth driving ratio by 25.0%.

关 键 词：汽车工程 驾驶行为 人机交互 驾驶模式 智能网联汽车 

分 类 号：B842[哲学宗教—基础心理学] U461[哲学宗教—心理学]