Evaluation and Validation of Distraction Detection Algorithms on Multiple Data Sources

Abstract

Many researchers have developed algorithms to detect distraction, but they have yet to be validated on multiple data sources. This study aims to evaluate these algorithms by comparing their ability to detect distraction and predict event likelihood. Four algorithms that use measures of cumulative glance, past glance behavior, and glance eccentricity were used to understand the distracted state of the driver and were validated on two separate data sources: naturalistic and experimental data. Results showed that there was a higher likelihood of event detection when cumulative glances were considered. Glance eccentricity was best for predicting distraction. Future research can use these findings to design mitigation systems that give drivers feedback in instances of high crash likelihood.

Looking out or Looking Away?—Exploring the Impact of Driving With a Passenger on Young Drivers’ Eye Glance Behavior

“…The eye glance data were coded and analyzed for this study by primary and secondary coders who were knowledgeable about the process of identifying a glance location. The eye glance data was also analyzed in the context of evaluating distraction detection algorithms, which is included in a separate study (Mehrotra et al, 2018).…”

“…Past research has found positive influence due to passenger intervention (Buckley & Davidson, 2013), as well as acceptability of negative norms by peers resulting in negative influence (Chein et al, 2011). Research by Zhang et al, (2018) examined this degree of association through measures of stimulating companionship. These studies point toward the degree of relationship between driver and passenger as having a variable impact on behavior.…”

“…An external audio system to the simulator provided environmental sounds like traffic, passing vehicles, and road noise, as well as pre-programmed voice commands. Vehicle data were collected and analyzed in previous studies (Zhang et al, 2018(Zhang et al, , 2019.…”

“…Relatedly, analysis of the same dataset evaluated algorithms designed to detect drivers’ distraction. The research found that eccentricity (deviation away from the center of the field of view), and the duration of the glance away from front view may result in higher crash likelihood (Mehrotra et al, 2018).…”

“…Relatedly, analysis of the same dataset evaluated algorithms designed to detect drivers' distraction. The research found that eccentricity (deviation away from the center of the field of view), and the duration of the glance away from front view may result in higher crash likelihood (Mehrotra et al, 2018). Pradhan et al (2014) examined how young drivers scanned for hazards when driving with passengers by considering confederates (actors) that represented different passenger properties.…”