Evaluating the influence of driving styles on takeover responses of autonomous vehicles

In order to achieve personalized takeover of autonomous driving, the takeover responses of drivers with different styles under different non-driving related tasks were studied. A takeover test platform was built based on the driving simulator, and 28 participants were recruited to conduct tests under the focused state and the distracted states (watching news and playing games). Firstly, based on the data of takeover responses in the focused state, the driving styles were classified into aggressive, average and conservative using the Fuzzy C-means(FCM) clustering algorithm; Then, five indicators were selected, including manipulation reaction time, takeover duration, maximum brake pedal input, maximum steering wheel input, and lane deviation standard deviation, to analyze whether there were significant differences in takeover responses of drivers with different styles under the subtasks of watching news and playing games. The results showed that the differences in takeover duration and maximum brake pedal input were significant among drivers of different styles under the same subtask. Significant differences in manipulation reaction time and lane departure standard deviation were found between drivers of the same style under two different subtasks. In addition, Regardless of driving styles or subtasks, there was no significant change in maximum steering wheel input.