Prediction Of Takeover Time Under The Interaction Effects Between Mixed Traffic Flow And Conditionally Automated Driving Vehicles

Autonomous driving technology relies on the collaboration among artificial intelligence,visual computing,radar,monitoring devices and global positioning systems to operate vehicles automatically.However,existing autonomous driving technology is not yet mature and has many safety issues.Conditionally automated driving vehicles can decrease the safety problems by letting the drivers take over the vehicles’ driving authority and perform the corresponding driving operations.In this study,the time between the sending over of request from the automated driving system and the completion of the drivers taking back control of vehicles is defined as the takeover time.According to the existing researches,the takeover time can reflect the safety condition of the vehicles to a large extent.However,most scholars consider the variables related to driver themselves and the vehicles when predicting the takeover time,and rarely consider the fact that the conditionally automated driving vehicles are in the traffic flow environment.In other words,the safety of the conditionally automated driving vehicles is largely affected by the traffic environment.Therefore,the existing takeover time prediction models cannot capture the spatiotemporal characteristics of the conditionally automated driving vehicles in the mixed traffic flow environment well.Based on aforementioned discuss,the main research contents of this study are as follows:(1)Evaluation of the impact of takeover time for conditionally automated driving vehicles on the mixed traffic flow environment.This study conducts simulation model for traffic flow mixed with manual driving vehicles,fully automated driving vehicles and conditionally automated driving vehicles to analyze the impact of conditionally automated driving vehicles on heterogenous traffic flow.The results showed that different takeover time significantly affects the stability and safety of traffic flow under different takeover time conditions.A moderate takeover time allows the driver to complete the takeover quickly with sufficient observation of the surrounding traffic conditions and mitigate the adverse effects of conditionally automated driving vehicles takeover transition on traffic flow and improve traffic flow safety.Therefore,accurate prediction of the takeover time of conditionally automated driving vehicles will improve the safety of vehicles and traffic flow and reduce traffic risks.In addition,taking moderate takeover time(7s)as the given takeover time,we developed a traffic flow model,and it is found that increasing the total penetration rates of conditionally automated driving vehicles and fully automated driving vehicles or that of conditionally automated driving vehicles alone will expand the traffic flow stability area.Moreover,the improving effects of traffic flow stability increase with the value of both penetration rates.(2)Analysis of the impact of mixed traffic flow environment on the takeover time of conditionally automated driving vehicles.The takeover time of conditionally automated driving vehicles is affected by numeral factors,especially the mixed traffic flow environment.In order to analyze the influencing factors related to mixed traffic flow environment of the takeover time,this study conducts real vehicle takeover experiments based on the conditional automated driving vehicles and radar-video integrated machine provided by Jiangsu University.This paper uses various models to make preliminary predictions of the takeover time.And then constructs an analysis model to explore the effect of mixed traffic flow environment on the takeover time based on SHAP.On the one hand,the authors find that the prediction accuracy of Deep GBM model better than the other three models.The accuracy of Deep GBM model to predict the moderate takeover time is better than to predict the shorter and longer takeover time.On the other hand,when the driver takeover the vehicle in the intersection area,the driver pays more attention to the speed factor.While in the nonintersection area,the driver focuses closely on the longitudinal distance difference with the vehicle ahead.Further study results showed that the takeover process of conditionally automated driving vehicles is subject to significant lateral interference when taking over in intersection areas,which can have an impact on the driver’s takeover time and thus have a significant impact on vehicle safety.(3)Takeover time prediction considering mixed traffic flow environment.The assessment results of the takeover time impact on the mixed traffic flow environment showed the necessity of predicting the takeover time of conditionally automated driving vehicles.While the results of the mixed traffic flow environment impact analysis on the takeover time showed that the variables related to the mixed traffic flow environment must be considered for the takeover time prediction of conditionally automated driving vehicles.Therefore,this study predicts the takeover time considered the mixed traffic flow environment using various models.Built upon the comparison of prediction results for different models,this paper utilizing Stem GNN algorithm to predict conditionally automated driving vehicles takeover time in mixed traffic flow environment.In summary,the takeover time prediction model developed in this study can effectively predict the takeover time of conditionally automated driving vehicles in the real vehicle environment,which provides important theoretical support for the early commercialization of conditionally automated driving vehicles.