Driver glance behaviors and scanning patterns: Applying static and dynamic glance measures to the analysis of curve driving with secondary tasks

Abstract

Performing secondary tasks (or non‐driving‐related tasks) while driving on curved roads may be risky and unsafe. The purpose of this study was to explore whether driving safety in situations involving curved roads and secondary tasks can be evaluated using multiple measures of eye movement. We adopted Markov‐based transition algorithms (i.e., transition/stationary probabilities, entropy) to quantify drivers’ dynamic eye movement patterns, in addition to typical static visual measures, such as frequency and duration of glances. The algorithms were evaluated with data from an experiment (Jeong & Liu, 2019) involving multiple road curvatures and stimulus‐response secondary task types. Drivers were more likely to scan only a few areas of interest with a long duration in sharper curves. Total head‐down glance time was longer in less sharp curves in the experiment, but the probability of head‐down glances was higher in sharper curves over the long run. The number of reliable transitions between areas of interest varied with the secondary task type. The visual scanning patterns for visually undemanding tasks were as random as those for visually demanding tasks. Markov‐based measures of dynamic eye movements provided insights to better understand drivers’ underlying mental processes and scanning strategies, compared with typical static measures. The presented methods and results can be useful for in‐vehicle systems design and for further analysis of visual scanning patterns in the transportation domain.

Improving accuracy of gaze‐control tools: Design recommendations for optimum position, sizes, and spacing of interactive objects

“…However, for differently abled or other individuals who cannot conveniently use these devices, modern human–computer interaction systems are not user‐friendly. Recently, eye‐related data have been widely used in interface assessment (Goldberg & Kotval, 1999), ergonomic analysis of behaviors (Jeong et al, 2019; Niu et al, 2020; Yang et al, 2018), and cognitive load assessment (Ahlstrom & Friedman‐Berg, 2006). Further, through the processing and calculation of eye‐tracking data, eye movement has gradually been used as a new interactive method.…”

“…of behaviors (Jeong et al, 2019;Niu et al, 2020;Yang et al, 2018), and cognitive load assessment (Ahlstrom & Friedman-Berg, 2006).…”

New Measures of Visual Scanning Efficiency and Cognitive Effort

“…The direct eye tracking indices, which are generated based on fundamental fixation and saccade, cannot be a proxy for the driving task performance. This is because that these indices are basically simple statistics and only applicable for specific applications, so as not to behave generally and robustly as an estimator of visual scanning efficiency [7].…”

“…Fixation and saccade are classic indices for this purpose. But, direct and indirect usages of these indices (for examples, rate/duration of them and their simple combinations) are more applicable in specific application scenarios [12], than in the general efficiency definition. In addition, pupil dilation and blink rate are two widely used eye tracking indices for the study of cognition and psychology [13].…”

Entropy-Based Effect Evaluation of Delineators in Tunnels on Drivers’ Gaze Behavior

“…The results showed that the entropy method could achieve good classification performance in driving fatigue studies [ 25 ]. Jeong et al stated that compared with typical static measures, the Markov-based entropy method can quantify eye movement and provide a better understanding of individual switching patterns [ 26 ].…”