A Study of Deep Learning Model for Driver Takeover Intention
With the Automated Vehicle (AV) technology gradually becoming stable enough to assist human drivers to do the driving tasks, drivers could divert their attention to non-driving related tasks while using the AV. Existing studies have developed models to predict drivers’ takeover performance upon requests from AVs, however, few work developed predictive models to predict drivers’ takeover intention. In order to design a robust automated driving system, it is important to predict drivers’ takeover intention in AVs and potentially alter the driving styles of AVs to reduce unnecessary takeover behavior from human drivers. In this work, we developed a multiple input Convolutional Neural Network (CNN) model to predict drivers’ takeover intention in automated vehicles with data from 29 human drivers with the time window ranging from 1-4 seconds were utilized the time series data. The results showed that the developed CNN model predicted the active takeover intention with accuracy up to 96.7% using the windows size of 4 seconds. The developed model could be further applied to AV systems to help AV systems to predict drivers’ intentions before a takeover action and optimize the driving styles of AVs to fulfill drivers’ needs in human-AV interaction.
Advisor: Yiqi Zhang, Assistant Professor, The Harold and Inge Marcus Department of Industrial and Manufacturing Engineering
Penn State Only
Files are only accessible to users logged-in with a Penn State Access ID.
|Work Title||A Study of Deep Learning Model for Driver Takeover Intention|
|License||In Copyright (Rights Reserved)|
|Work Type||Research Paper|
|Publication Date||December 31, 2021|
|Deposited||March 22, 2021|
This resource is currently not in any collection.