Analysis of the characteristics of driving risk avoidance strategic decision behavior at mountainous road intersection
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 decision behavior of driving avoidance.According to the association rules of the characteristic indicators,this study can anticipate upcoming risks at mountain road intersections and avoid them as much as possible.
万方数据
维普
