Using Naturalistic Vehicle-Based Data to Predict Distraction and Environmental Demand

Using Naturalistic Vehicle-Based Data to Predict Distraction and Environmental Demand

Dina Kanaan (University of Toronto, Toronto, Canada), Suzan Ayas (University of Toronto, Toronto, Canada), Birsen Donmez (Department of Mechanical and Industrial Engineering, University of Toronto, Toronto, Canada), Martina Risteska (University of Toronto, Toronto, Canada) and Joyita Chakraborty (University of Toronto, Toronto, Canada)
Copyright: © 2019 |Pages: 12
DOI: 10.4018/IJMHCI.2019070104
Article PDF Download
Open access articles are freely available for download

Abstract

This research utilized vehicle-based measures from a naturalistic driving dataset to detect distraction as indicated by long off-path glances (≥ 2 s) and whether the driver was engaged in a secondary (non-driving) task or not, as well as to estimate motor control difficulty associated with the driving environment (i.e. curvature and poor surface conditions). Advanced driver assistance systems can exploit such driver behavior models to better support the driver and improve safety. Given the temporal nature of vehicle-based measures, Hidden Markov Models (HMMs) were utilized; GPS speed and steering wheel position were used to classify the existence of off-path glances (yes vs. no) and secondary task engagement (yes vs. no); lateral (x-axis) and longitudinal (y-axis) acceleration were used to classify motor control difficulty (lower vs. higher). Best classification accuracies were achieved for identifying cases of long off-path glances and secondary task engagement with both accuracies of 77%.
Article Preview
Top

Introduction

Driving is a complex task that mainly requires visual perception and manual control. The performance of this task is influenced by the driver’s ability to perceive and adapt to different environmental demands, which in turn is influenced by driver’s state. Distraction, a driver state that is defined as “the diversion of attention away from activities critical for safe driving toward a competing activity” by Regan, Lee, and Young (2008, p. 34), has been shown to be prevalent among drivers (Dingus et al., 2016; Young & Lenné, 2010), to impair driving performance (Caird, Willness, Steel, & Scialfa, 2008; Horrey & Wickens, 2006; Regan et al., 2008), and to increase crash risk (Dingus et al., 2016; National Highway Traffic Safety Administration, 2018).

Technological advances are generating smarter vehicles that aim to enhance traffic safety and the experience of driving by detecting driver state (McCall & Trivedi, 2004; Pentland & Liu, 1999; Yang, Lin, & Bhattacharya, 2010), the state of the driving environment (Fridman et al., 2016), and driver intent (Jain, Koppula, Raghavan, Soh, & Saxena, 2015; Martin, Vora, Yuen, & Trivedi, 2018). These technologies that are mainly still in development are promising for distraction mitigation. For example, upon predicting driving impairment due to distraction, the vehicle may provide feedback to the driver that could help direct the driver’s attention back to the driving task (Donmez, Boyle, & Lee, 2007, 2008) or may take over lateral and/or longitudinal control (e.g. Stanton & Young, 2005). Further, when high levels of driver distraction are detected, the vehicle can adapt user interfaces that are built-in or brought into the vehicle, such as by filtering information content, delaying notifications, and blocking access to certain actions. For example, Tchankue, Wesson, and Vogts (2011) developed a prototype adaptive user interface for an in-car communication system that blocked incoming phone calls and prevented drivers from sending text messages when distraction was detected. Such a system can also block phone calls when environmental demands are predicted to be high. In general, better system awareness of the driving environment can lead to more effective and intelligent distraction mitigation strategies.

Complete Article List

Search this Journal:
Reset
Open Access Articles
Volume 12: 4 Issues (2020): 1 Released, 3 Forthcoming
Volume 11: 4 Issues (2019)
Volume 10: 4 Issues (2018)
Volume 9: 4 Issues (2017)
Volume 8: 4 Issues (2016)
Volume 7: 4 Issues (2015)
Volume 6: 4 Issues (2014)
Volume 5: 4 Issues (2013)
Volume 4: 4 Issues (2012)
Volume 3: 4 Issues (2011)
Volume 2: 4 Issues (2010)
Volume 1: 4 Issues (2009)
View Complete Journal Contents Listing