Driving Behavior Recognition Based on EEG Data From a Driver Taking Over Experiment on a Simulated Autonomous Vehicle

Abstract

Driving behavior recognition is a critical part of the driver safety system. Most of the popular researches have utilized the driver’s questionnaire data and driving data to recognize driving behavior. But few studies have used physiological data to recognize different driving behaviors, such as EEG data. In this paper, a new method was presented to recognize different driving behaviors based on EEG data. The driving data and EEG data were collected when participants took over the autonomous vehicle. Through K-means, the driving data were classified into two groups, and the classification results were utilized as inputs to generate a k-Nearest-Neighbor model to recognize driving behavior groups via EEG data. The results showed that the average recognition accuracy was 80.6%, and the highest recognition accuracy was 84.2%, indicating that there is an obvious relation between EEG data and driver taking over behavior.