Predicting drowsy driving in real-time situations: Using an advanced driving simulator, accelerated failure time...

Area of Science:

• Road safety
• Human factors in transportation
• Machine learning for predictive modeling

Background:

• Driver drowsiness is a significant factor in road accidents.
• Existing drowsy driving detection systems often rely on fixed parameters.
• Predictive models integrating real-world driving data are needed.

Purpose of the Study:

• To identify key factors influencing driver drowsiness.
• To develop a real-time drowsy driving probability model.
• To leverage virtual Location-Based Services (LBS) data for drowsiness prediction.

Main Methods:

• Conducted a driving simulation experiment with 32 participants.
• Collected continuous driving time and virtual LBS data (temperature, time of day, speed, traffic, road type).
• Utilized an Accelerated Failure Time (AFT) model and demographic data (nap habit, age, gender, experience).

Main Results:

• Drowsiness occurred less during the day and at lower temperatures.
• Drivers with a nap habit were more susceptible to drowsiness.
• Higher average speeds, prolonged traffic jams, and freeway driving increased drowsiness risk.

Conclusions:

• Environmental and demographic factors significantly impact driver drowsiness.
• The developed AFT model enhances understanding of drowsiness predictors.
• The model can power real-time LBS-based drowsy driving warning systems.