01876012

Component

https://doi.org/10.1177/03611981231157401

Driver monitoring systems are growing in importance as well as capability. This paper reports on drowsy driving detection models generated from multiple sources of driver monitoring data. Behavioral (driver) data were provided by a camera-based production-type driver monitoring system manufactured by Aisin Technical Center of America (from the Aisin Group). Vehicular data were recorded from the National Advanced Driving Simulator’s large-excursion motion-base driving simulator. Forty participants drove the simulator for up to 3 h after being awake for at least 16?h. Periodic measurements of drowsiness were made every 10 min using both observational ratings of drowsiness by an external rater and the self-reported Karolinska Sleepiness Scale. A novel application of sequence analysis with clustering and hidden Markov models resulted in models that tracked well with the subjective drowsiness measures. The area under Receiver Operating Characteristic curves evaluating the models ranged from 0.85 to 0.87. By allowing for many distinct patterns observed in driving sequences, the hope is that the method will offer a robust way to accommodate pattern variability that naturally occurs over time and among drivers.

Chris Schwarz https://orcid.org/0000-0003-0291-7157© National Academy of Sciences: Transportation Research Board 2023.

English

0000-0003-0291-7157

pp 553-562

2023-8

Transportation Research Record: Journal of the Transportation Research Board

Web

References (19)

Detection and identification; Driver monitoring; Drowsiness; Human factors; Sequential analysis

Karolinska Sleepiness Scale

Highways; Safety and Human Factors

TRIS, TRB, ATRI

Mar 20 2023 3:02PM