GA-XGBoost Driving Distraction State Classification Model Based on Non-Driving Related Tasks

Abstract

At present, the man-machine control switch does not consider that the driver in different driving behavior states and lead to certain differences in the takeover efficiency, in order to study the relationship between the physiological rhythm signal and the driving behavior state when the driver is in different driving behavior state, this paper proposes to establish a classification model of the driving behavior state through the physiological rhythm signal. Firstly, based on PreScan, the simulated driving environment is established, and the physiological instrument is used to monitor the physiological signals of the driver during the test, the physiological rhythm signal data set under different driving states is established, and the corresponding labels are assigned to different driving behavior states, the physiological data is used as the input of the model, and the driving behavior state is used as the output, and then the parameters of the single XGBoost model are optimized by genetic algorithm to generate a GA-XGBoost driving behavior state classification model, and the results show that The optimized model accuracy increased by 12.5%. Therefore, the classification model proposed in this paper can effectively judge the driving behavior state through the physiological rhythm signals of the human body, and use the changes of some physiological indicators of the driver to achieve real-time monitoring of driving distraction.