Research On The Effect Of Multimodal Stimuli On Autonomous Driving Takeover Performance Based On Simulated Driving

As autonomous driving technology continues to develop around the world,the use of autonomous driving features in cars is gradually being adopted.However,traffic safety issues caused by autonomous vehicles are receiving increasing concerns.With current mainstream autonomous driving technology,it is still necessary to allow the driver to take over the vehicle if the vehicle is unable to perform the autonomous driving manoeuvre effectively,so the level of driver takeover in different situations can have a direct impact on traffic safety.By investigating the effect of different modalities of takeover request stimulation on driver takeover performance in different traffic scenarios,the level of driver takeover can be improved and thus traffic safety can be enhanced.Therefore,it is valuable to carry out a comprehensive evaluation and impact study of multi-modal stimuli on takeover performance.In this paper,the writer proposes a comprehensive quantitative evaluation method of takeover performance based on simulated driving data,using statistical analysis and data envelopment analysis,and further explore the specific effects of different types of modal stimuli on takeover performance.The details of the study are as follows.First,driving simulation experiments are designed and the experimental equipment,personnel and procedure are determined;the indicators of takeover behaviour in this study are identified by analysing the complete takeover behaviour process and their correlation with takeover performance is clarified;vehicle driving data and driver behaviour data are collected based on the driving simulation experiments,and data screening and segmentation are completed.Totally,360 takeover data samples were obtained.Second,by considering the entire process of each takeover as a whole,the writer integrates the behavioural characteristics parameters and propose a factor analysis-based takeover performance evaluation method,in which the scores of each parameter in the takeover performance are characterised by the combined factor scores;by considering the weight differences of parameters among independent samples in the takeover performance,a quantitative evaluation method of takeover performance based on data envelopment analysis(DEA)is proposed with excellent takeover behaviour as the benchmark;the implementation details of the two methods are clarified through case studies,and the advantages and disadvantages of each are analysed.Finally,the specific effects of multimodal stimuli on takeover behaviour are explored.Using the grey correlation analysis,the selected takeover behaviour indicators and the quantified takeover performance indicators are analysed together to determine the optimal performance values as the reference set,and then the correlation coefficient matrix and the final evaluation scores are calculated.The results show that there are differences in the specific effects of different modal stimuli on the driver takeover process,and there is a certain level of correlation and influence between the multimodal stimuli.The proposed methods for evaluating the takeover performance of autonomous vehicles and the specific analyses of the influence of multimodal stimuli on driver takeover behaviour can provide theoretical support and practical guidance for the setting and implementation of multimodal stimuli in different takeover scenarios and for improving driving safety as well.