Research On Driver’s Driving Capability Evaluation Method Based On Vehicle States

Chinese automobile industry as a primary industry of national economy has developed rapidly and is gradually moving to automotive intelligence.In-depth study of the driving behavior and characteristics of human drivers is of great significance in promoting the development of intelligent vehicles to highly automated driving.In the process of intelligent,there is a long time of human-machine cooperative driving.Aiming at the problem of driving right transfer in the process of Man-machine co-politing,this paper establishes a comprehensive evaluation model of driving ability based on vehicle driving states,and puts forward the driving right switching strategy based on the change of driving ability.The main research contents of this paper are as follows:1)Based on Unreal Engine 4(UE4),CarSim automobile Dynamics model and Logitech G29 Vehicle Control Feedback Kit,a driving simulation platform is established,a typical driving scene and working conditions are designed,and the synchronous collection of data in simulated driving experiment is realized.A driver information acquisition system based on WinForm framework in Visul C # platform is set up,and the synchronous collection and analysis of driver’s basic information and questionnaire results are completed.2)The driving experiment and stochastic N-Back experiment were designed,and the changes of driver’s physiological index and vehicle state index were studied in the driving process with different degree of distraction.The difference between driver’s physiological data and vehicle state data under different degree of distraction was tested by Mann-Whitney U test method.The results show that the steering wheel angle entropy,steering frequency,distance fluctuation and other indexes accurately reflect the driver’s state change,thus constructing the vehicle State index set which characterizes the driver’s ability.3)Starting with the vehicle state,the correlation between vehicle state indexes is analyzed.Based on the subjective and objective comprehensive driving ability evaluation algorithm of Fuzzy Analytic Network Process(F-ANP)combined with entropy weight method,this paper establishes the driving ability evaluation model based on vehicle state,and uses dynamic time window to discretize the data,which improves the timeliness of the evaluation of driving ability.At the same time,the driving capacity evaluation model based on vehicle state was verified by using questionnaire survey to evaluate and sort driving ability offline.4)BP Neural Network is used to realize the real-time quantification of driving capability,and the change characteristics of driver’s driving ability during long driving time are analyzed,the results show that the driving ability always fluctuates up and down within a certain range,and with the increase of driving time,the driving ability shows a decreasing trend,This paper puts forward the driving right switching strategy based on the change of driving ability.