Considerations When Calculating Percent Road Centre From Eye Movement Data in Driver Distraction Monitoring

Ahlstrom, Christer; Kircher, Katja; Kircher, Albert · 2009 · Crossref

DOI: 10.17077/drivingassessment.1313

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study investigates methodological considerations for calculating Percent Road Centre (PRC), a performance indicator used to monitor driver distraction via eye movement data. While PRC is traditionally defined using fixation data extracted from eye-tracking recordings, it has been suggested that raw gaze data (including both fixations and saccades) could serve as a viable alternative. The research aims to determine if PRC calculated from raw gaze data yields comparable results to fixation-based calculations and to assess the impact of gazes directed at the centre rear mirror on the accuracy of the indicator. The analysis utilized naturalistic driving data from a field operational test involving seven participants who drove a Saab 9-3 equipped with a SmartEye Pro 4.0 remote eye tracker. Participants drove for approximately 10 days in a baseline phase and three weeks with an active distraction warning system, accumulating an average of 4,350 km. Data were processed offline using MATLAB, excluding sessions with less than three minutes of active driving or speeds below 50 km/h. PRC was calculated using a sliding four-second window, defining the road centre area as an 8° radius circle centered on the most frequent gaze angle. Fixations were identified using a velocity and duration-sensitive algorithm, while raw gaze data included all recorded points. The study compared PRC time traces derived from both methods using correlation and wavelet-based semblance analysis. Additionally, the researchers evaluated the effect of excluding gazes intersecting with the centre rear mirror. Results indicated that PRC time traces based on gaze data and fixation data were highly similar, with an overall correlation coefficient of 0.95 and an average wavelet semblance of 0.84. However, fixation-based PRC values were significantly lower in amplitude than gaze-based values, showing an average absolute difference of about 8%. The distance between the calculated road centre points for the two methods was negligible (average 0.02 radians). Regarding the centre rear mirror, the study found that for three of the seven drivers, the road centre area intersected with the mirror’s location. Approximately 2% of gaze cases within the road centre area were directed at the mirror. Excluding these mirror gazes altered the road centre point in 4% of trips and reduced the average PRC by about 1%, confirming that mirror gazes can negatively influence the indicator's correctness. The findings suggest that PRC can be calculated directly from raw gaze data without segmenting it into saccades and fixations, simplifying the implementation of real-time distraction monitoring systems by removing the need for complex fixation detection algorithms. However, because of the amplitude difference between the two methods, threshold settings for warning systems must be carefully adjusted. Furthermore, the study concludes that gazes directed at the centre rear mirror should be excluded from PRC calculations to ensure accuracy, particularly in systems capable of identifying specific vehicle components. This approach enhances the robustness of PRC as a metric for driver distraction monitoring.

Key finding

PRC time traces calculated from raw gaze data are highly similar to those calculated from fixation data, allowing the omission of fixation segmentation, though gazes directed at the centre rear mirror should be excluded to prevent calculation errors.

Methodology

field_study

Sample size: 7

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• gaze based attention detection
• visual
• eye movements scanning
• attention allocation
• temporal
• peripheral attention

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: behavioral performance data
• Methodological Resource: tool software, measurement protocol