山区公路交叉口驾驶避险决策行为特性分析  

Analysis of the characteristics of driving risk avoidance strategic decision behavior at mountainous road intersection

作　　者：秦雅琴[1] 包丽馨 陈亮[1] 勾钰 王锦锐 QIN Yaqin;BAO Lixin;CHEN Liang;GOU Yu;WANG Jinrui(Faculty of Transportation Engineering,Kunming University of Science and Technology,Kunming 650500,China)

机构地区：[1]昆明理工大学交通工程学院,昆明650500

出　　处：《安全与环境学报》2024年第6期2295-2304,共10页Journal of Safety and Environment

基　　金：国家自然科学基金项目(71861016)。

摘　　要：为研究驾驶特征指标与驾驶避险行为决策的关联规则以提升驾驶安全,将驾驶避险决策行为划分为纵向“停车避险”和横向“转向避险”,并运用驾驶模拟系统构建12种山区公路交叉口交通冲突虚拟场景,招募38名驾驶人进行试验,采集车辆特征和驾驶人扫视、眨眼、注视等眼动特征数据。基于随机森林构建驾驶避险决策行为模型,然后引入沙普利加和解释(SHapley Additive exPlanation, SHAP)框架进一步分析车辆、眼动特征与驾驶避险行为之间的非线性关系。结果表明:模型对纵、横向避险行为预测的准确率分别为84.77%、94.70%;纵向速度标准差、扫视持续时间标准差、轨迹偏差标准差、侧向速度标准差与驾驶避险决策行为存在明显关联,如纵向速度标准差过大(约大于10 km/h),纵向“停车避险”可能性明显增加。This study aims to improve driving safety based on the association rules between driving characteristic index and risk avoidance behavior.This study divides driving risk avoidance strategic decision behavior into two types:longitudinal“braking avoidance”and horizontal“turning avoidance”,and uses a driving simulation system to construct 12 virtual scenes of traffic conflicts at mountainous road intersections.The author recruits 38 drivers for driving simulation tests and collects vehicle characteristics and eye-tracking data including scanning,blinking,and gaze.A behavioral model for driving avoidance decisions based on random forests is constructed.The SHapley Additive exPlanation(SHAP)framework is introduced to further analyze the nonlinear relationship between vehicle characteristics,eye movement characteristics,and driving avoidance behavior.The results of the strategic decision show that the accuracy of the prediction for the longitudinal avoidance behavior model is 84.77%and the accuracy of the prediction for the horizontal avoidance behavior model is 94.70%.The analysis results from the SHAP framework show that longitudinal speed standard deviation,sweep duration standard deviation,trajectory deviation standard deviation,and lateral speed standard deviation are significantly associated with driving avoidance strategic decision behavior.For example,when the standard deviation of longitudinal speed is approximately greater than 10 km/h,SHAP>0.Mountain road intersections can be potentially risky and drivers tend to make braking avoidance decisions;When the standard deviation of trajectory deviation is approximately greater than 0.45 m,SHAP>0.Drivers are very likely to encounter traffic risks at mountain road intersections and need to change the vehicle trajectory to avoid a collision,at which time drivers tend to take a turning risk avoidance decision.By analyzing the characteristic effect of 4 samples,it is concluded that the characteristic indexes shown in the model can well reflect the strategic dec

关 键 词：安全工程 驾驶避险 决策行为 驾驶模拟 分类预测 沙普利加和解释(SHAP) 

分 类 号：X951[环境科学与工程—安全科学]