Research On The Validation Of Vehicle Driving Simulator Based On Driver’s Traffic Characteristics

It has gradually become an important development direction in the field of road traffic research to apply vehicle driving simulator for simulation study based on the road traffic ‘people-vehicle-road-environment’ system,which brings more and more urgent requirements for systematic evaluation of the effectiveness of vehicle driving simulator.Drivers play a dominant role in road traffic system.However,from the perspective of man-machine ergonomics,comprehensive analysis of the effectiveness of vehicle driving simulators based on complex road traffic conditions and risky road traffic environments is still insufficient.Based on the ergonomics theorystarting from the perception and behavior characteristics of drivers,aiming at the comprehensive evaluation of the simulator,the comprehensive evaluation system of the effectiveness of the vehicle driving simulator is constructed according to the four elements of the road traffic system "people-vehicleroad-enviroment".It focuses on solving the problems of insufficient research on complex road conditions,risky traffic environment and comprehensive evaluation in simulator effectiveness evaluation.The evaluation system takes the quantitative evaluation of effectiveness as the reasearch focus,carries out theoretical research,experimental design and case verification,and obtains the comprehensive conclusion of simulator effectiveness evaluation through systematic analysis,subjective and objective weighting and model calculation of the typical extracted multi-dimensional and multi-attribute parameters.The major of the study involves:1)Evaluation of basic effectiveness of simulator under the man-machine ergonomics visual threshold.According to the three elements of cockpit,static imaging,and basic dynamic characteristics,the evaluation of the basic perception effectiveness of the driving simulator vehicle is divided.The effectiveness of the cockpit is studied based on the ergonomics hard point theory as the evaluation standard.In terms of the fidelity of static imaging,a three-dimensional virtual image calibration model of a driving simulator is constructed.And for vehicle dynamics characteristics,the real vehicle and Carsim are selected for comparative experiments,and the calibration correlation validity was used to evaluate.2)Simulator effectiveness evaluation for interaction between road perception and driving behavior.Curves with different curvature radii are selected as evaluation scenes,and vehicle speed and trajector are taken as evaluation indexes.The α reliability coefficients,Spearman-Brown coefficients,Guttman half-coefficients and R reliability coefficients are used to verify the consistency of the experimental speed.The validity coefficient and Cohen d index of effect are used to evaluate the reliability and effect of the experimental vehicle speed.The cosine similarity is used to evaluate the lateral deviation of the trajector.The simulator curve trajectory planning model(MUL-BILSTM)is established based on the idea of humanoid manipulation data-driven.The model results can replace the expected trajectory of vehicle driving under complicated road conditions and reduce the cost of real vehicle parameter acquisition.3)Effectiveness evaluation of simulators in risky traffic environment.Taking the risk degree of the expanding TTC(Time-To-Collision)theoretical model as the calibration parameter,and physiological and psychological characteristics of drivers under risk environment as evaluation indexes,designing risk conflict scenario such as without block signals cross intersection,shedding no cross intersection,low flow no signal without sunscreen,cross intersection,high flow no signal without sunscreen,DTW(dynamic time adjustment)is introduced to analyze the similarity between TTC and drivers’ electrocutaneous signals(EDA)and electromyographic signals(EMG).And combined with ECG and EMG sample entropy and complexity analysis in different risk environments,the effectiveness of driving simulator in risk environments is evaluated.4)Comprehensive evaluation of simulator effectiveness based on human factors.In order to comprehensively consider the effectiveness of the driving simulator,the driving simulator parameters with nonlinear and uncertain characteristics are considered,and deep learning model has good capability of nonlinear fitting with the characteristics of deep excavation features.The Mul-Bayes-LSTM model for comprehensive evaluation is further introduced according to the driving simulator in experimental psychology indicators via constructing the vehicle,road and environment feature variables of the driving simulator as the input layer and the two-dimensional feature matrix of the human factor feature variables as the output layer.Bayes hyperparameter optimization was used to better match the driver’s human factor characteristics and reflect the actual experimental results of the driving simulator.Based on the model results,entropy method and CRITIC(Criteria Importance Though Intercrieria Correlation)analysis method were used to weight the evaluation indexes and then substituted into the evaluation model to complete the comprehensive effectiveness evaluation of the simulator.In summary,a comprehensive evaluation system for the effectiveness of vehicle driving simulators applied in the traffic field is constructed,from the perspective of man-machine ergonomics and aiming at filling in the deficiencies of complex road conditions,risky traffic environment and comprehensive evaluation in the effectiveness evaluation of vehicle driving simulators.In the study of the independent and partitioning evaluation of the effectiveness of the simulator on complex road conditions,a curve trajectory planning model(MUL-BI-LSTM)driven by humanoid manipulation data is proposed.In order to objectively reflect the experimental effects of the simulator in a complex environment,a comprehensive effectiveness evaluation model of the simulator based on drivers’ psychological characteristics was proposed,which provides theoretical and methodological support for exploring the behavior mechanism of traffic human factor and evaluating and optimizing the comprehensive operation of the driving simulator.