Analysis of Drivers' Head and Eye Movement Correspondence: Predicting Drivers' Glance Location Using Head Rotation Data

Abstract

Summary:The relationship between a driver's glance pattern and corresponding head rotation is not clearly defined. Head rotation and eye glance data drawn from a study conducted by the Virginia Tech Transportation Institute in support of methods development for the Strategic Highway Research Program (SHRP 2) naturalistic driving study were assessed. The data were utilized as input to classifiers that predicted glance allocation to the road and the center stack. A predictive accuracy of 83% was achieved with Hidden Markov Models. Results suggest that although there are individual differences in head-eye correspondence while driving, head-rotation data may be a useful predictor of glance location. Future work needs to investigate the correspondence across a wider range of individuals, traffic conditions, secondary tasks, and areas of interest.

Investigating the correspondence between driver head position and glance location

“…The video was recorded by a camera mounted below the rearview mirror. A previous brief report from our group (Muñoz et al, 2015) showed a number of differences in the distributions of head rotations associated with glances to the road and center cluster between the static and dynamic samples. Consequently, the data from the 22 participants from the dynamic trials make up the focus of this analysis since actual on-road behavior is our primary interest.…”

“…The performance measures reported in the result section were computed using this normalization method. Furthermore, to discount any potential bias inherent in how subjects are sampled, a Monte-Carlo sampling technique was used (Muñoz et al, 2015;Muñoz et al, 2016). For each of 50 iterations of this sampling approach, training and validation test sets were generated as above.…”

“…Practically speaking, the only elemental parameter of an HMM is the number of hidden states used to model the temporal data. Muñoz et al (2015) provides additional detail as to how the parameters for each model were set.…”

“…has been seen as key to understanding driver interaction with in-vehicle devices and to estimate potential crash risks. For example, several studies show that crash and near-crash risk increased as the duration of off-road glances increased (e.g., Liang, Lee & Yekhshatyan, 2012;Victor et al, 2015). However, traditional technologies for automated eye-tracking have been susceptible to data quality issues (Ahlstrom et al, 2012;Sodhi, Reimer & Llamazares, 2002) and difficult to reliably use in production systems, especially for on-road experiments and naturalistic driving studies.…”

“…The present paper presents expanded analytic detail and findings from an effort that was developed to further explore whether head-rotation data can be used as a surrogate for eye-glance behaviors in an on-road environment where eye-tracking is more challenging compared to laboratory experiments (Muñoz et al, 2015). This analysis took advantage of glance and head rotation data drawn from a study conducted by the Virginia Tech Transportation Institute (Transportation Research Board, 2013); the glance data were manually coded for glance region and temporal points where glance orientation transition from one location to another by two independent coders (and a senior research associated mediated if disagreement occurred between two coders), and head rotation data were estimated from manually extracted facial landmarks (details are described in the 'Method' section).…”