Driver drowsiness detection based on non-intrusive metrics considering individual specifics.

Area of Science:

• Road safety
• Human-computer interaction
• Transportation engineering

Background:

• Drowsy driving is a major cause of road accidents.
• Current drowsiness detection methods often fail due to a lack of personalized assessment.
• Effective drowsiness detection is crucial for preventing related crashes.

Purpose of the Study:

• To develop an advanced drowsiness detection model.
• To incorporate individual differences in drowsiness effects on driving.
• To improve the accuracy of real-time drowsiness detection for drivers.

Main Methods:

• Nineteen driving behavior and four eye-tracking variables were collected.
• Participants drove a simulator after an overnight shift, reporting drowsiness levels.
• Multilevel ordered logit (MOL), ordered logit, and artificial neural network models were employed.

Main Results:

• The MOL model demonstrated superior drowsiness detection accuracy.
• Individual differences were found to significantly enhance detection capabilities.
• Key predictive variables included eyelid closure, pupil diameter, and steering control metrics.

Conclusions:

• Integrating individual variability into drowsiness detection models boosts accuracy.
• Personalized drowsiness detection systems can improve driver safety.
• Further research into individual differences can refine drowsy driving prevention strategies.