Research On The Driver Workload Similarity Model Of Comparative Experiment In Real And Simulated Expressways Based On EEG Data

In the closed loop system composed of people,cars and roads,the driver is the most crucial part.The driving process is actually a process in which the driver perceives the environment,makes decisions,and then controls the vehicle.Therefore,today’s traffic safety situation has become increasingly severe.Research on driver behavior plays a decisive role.Even with the rapid development of artificial intelligence,various technologies of intelligent networked automobile have made rapid progress,and the role of the driver is gradually weakening.However,before full-automated driving technology matures,vehicles must accept more or less control or supervision from the driver,so the driver still plays an extremely crucial role.In the field of driver behavior research,driver workload is one of the most important research directions.Accumulation of driver workload can cause driver fatigue,and driving fatigue can endanger traffic safety and people’s lives.Most of the existing researches about driver workload are based on driving simulators.However,even if the driving simulator has a high degree of simulation,there is also a difference between simulated driving and real driving in driving experience,which results in errors in the research based on driving simulators.In this paper,a comparative study of the driver workload between the road experiment and the driving simulator experiment is carried out,and a data compression method based on the integral recombination method is originally created,and the similarity model of driver workload in real and simulated highways is established.The specific research content is as follows:(1)Design and implementation of the experiment scheme: A comparison experiment of the driver workload in real and simulated expressways was designed,and a comparative experiment platform was set up.The road experiment and the simulator experiment were completed separately,and the driver’s EEG data,reaction time data,and subjective self-evaluation of the driver were collected.(2)Selection of driver workload evaluation index: First,this paper preprocesses the acquired experiment data,then compares the reaction time data with the EEG data,and seeks the Pearson correlation coefficient between them to select the largest EEG factor as the evaluation index of driver workload.Through calculation,this paper chooses α/β value as evaluation index.(3)Classification of driver workload levels: This paper selected K-means clustering method that is one of the unsupervised learning methods to initially classify driver workload levels and divides the driver workload into normal,medium,and high load states.Then combined with the subjective evaluation of the driver,the results of the classification of the driver workload are verified with accuracy,and the results show that the classification result is reliable.(4)Contrastive analysis of driver workload in real and simulated expressways: This paper analyzes the change of driver workload in real and simulated expressways from the time domain.And then this paper carried out comparative studies about changes of driver workload in these two scenarios and focused on different points.(5)The similarity model of driver workload in real and simulated expressways based on integral recombination method: This paper invented a data compression method based on integral recombination method,which unified the duration of driver workload data in real and simulated expressways,and then established a similarity model that focused on the changing law of driver workload.