Take-over Characteristics And Risk Analysis Of Human Driver In The Cooperative Driving System

With the rapid development of computer technology and artificial intelligence,the automotive industry is ushering in a great change that has not been seen in a century.In this paper,the classic single driver dual control man-machine co driving simulation environment was built through the traffic scene simulation software SILAB and human factors engineering test vehicle.The driving tasks to be taken over were evaluated based on resource occupation,and the driving task requirements in the whole process was quantified.By analyzing the characteristic data of driver take-over process,extracting measurable key factors,a take-over risk assessment model based on gradient lifting decision tree was constructed,which measured the risk from three dimensions: vehicle handling,eye movement representation and supplydemand balance.Specific research contents were as follows:(1)An experimental scheme to obtain the characteristics of driver take-over was proposed.Human factors engineering test vehicle,traffic scene simulation software SILAB,DG3 eyeglasses eye tracker,Mobileye driving data acquisition instrument,D-Lab driving behavior analysis system and other equipment are used to build three typical man-machine joint driving and take-over environments under six lane expressway,including front open space,rear end collision of front vehicles and lane change of front vehicles,Through the questionnaire test,20 drivers were selected to carry out the driving takeover test,obtain the whole process eye movement characteristics,vehicle operation data and subjective evaluation data of the subjects’ driving task needs,and preprocess the data.(2)The need for driving tasks to be taken over was quantified.Based on the multiplechannel theory of multiple-resource theory,a driving task demand evaluation model suitable for typical man-machine driving environment was established to quantify the driving task demand in the whole process of taking over.The subjective and objective comprehensive verification was carried out for typical take-over task scenarios.The significance level of the model was p = 0.04934(< 0.05),and the demand assessment model was reliable.(3)The characteristics of driver takeover behavior in man-machine driving environment was analyzed and compared.The driver’s driving characteristics under the two driving modes of take-over and traditional driving in the man-machine driving environment were compared and analyzed,and the significant relationship between various factors and driving risk was analyzed based on ANOVA.The research shows that the steering wheel angle value,brake pedal depression degree,accelerator pedal depression degree,gaze duration,saccade duration,saccade amplitude,pupil area and other data have a strong correlation with driving risk.(4)The driver take-over risk in the man-machine driving environment was analyzed.Because it was difficult to obtain the data source of conventional driving risk discrimination indicators,this paper proposed to comprehensively evaluate the driver take-over risk from three dimensions: vehicle handling,eye movement representation and task demand.Based on the above calculated driving demand and factor correlation analysis results,an assessment model for takeover risk in man-machine driving environment was established based on gradient lifting decision tree and supply-demand balance.The example showed that the accuracy of risk identification is as high as 89.2%.