Research On The Influence Of Emotion On The Takeover Ability Of Drivers Of L3 Level Autonomous Vehicles

With the rapid development of automatic driving technology,the automation level of vehicles has been greatly improved.At present,automatic driving technology can achieve automatic driving under conditions,but once it is difficult to make decisions or automatic driving fails,the driver still needs to manipulate the vehicle.Therefore,man-machine driving must become a necessary stage in the development of automatic driving,and the research on driving takeover is also carried out.In the process of taking over,the driver undertakes important functions of decision-making and execution.However,man is a complex individual,which will be accompanied by changes in mood,fatigue,consciousness,mental load and other states,and will have a great impact on the ability to take over.From the perspective of the driver’s own state,this paper explores the impact of the driver’s emotion on the takeover ability,and evaluates the driver’s takeover ability through the handling indicators of the driver’s takeover,including speed,acceleration,lateral offset,steering wheel angle,as well as the safety indicators of the takeover,collision rate and minimum TTC,so as to study the specific impact of emotion on the takeover ability of the driver.Firstly,based on the driving simulation platform,this paper carries out the automatic driving simulation takeover test,recruits driving volunteers,and induces the driver’s emotions by watching videos,including positive,negative and calm emotions.The takeover scene of lane change is designed,and the EEG,ECG and driving data of the driver are collected in real time,and then the effect of emotion induction is evaluated by the driver’s subjective questionnaire.Secondly,based on the collected physiological signals of drivers,this paper extracts multiple features of EEG and ECG signals,and constructs machine learning models of KNN,SVM,XGBoost and BP neural networks for driver emotion classification and recognition.Comparing the training results of various models,it is found that XGBoost model has stronger recognition ability,and the accuracy of classification and recognition of three emotions can reach more than 87%Finally,according to the traffic data,select the indicators to evaluate the takeover performance and takeover safety,and use the statistical analysis method for the takeover performance indicators and takeover safety indicators to explore whether there is a significant difference in the impact of emotion on the performance indicators,then calculate the weight of the takeover performance indicators through the analytic hierarchy process and entropy weight method,get the takeover performance score,and compare and analyze the differences of takeover performance under the three emotions.The results show that moderate positive emotion can promote the improvement of driver’s takeover performance,and too strong positive emotion will also bring negative effects.The negative emotion will seriously reduce the driver’s takeover performance,which is dangerous,and is prone to collision accidents,which increases the collision rate.The stronger the negative emotion is,the more the takeover performance will be reduced.To sum up,this paper analyzes the impact of driver’s emotional change on driver’s takeover ability,including takeover performance and takeover safety,and puts forward a machine learning model for driver’s emotional recognition,which is helpful to better monitor the driver’s state,avoid the negative impact of emotion and improve traffic safety.