Developing a system to reliably detect driver drowsiness using physiological signals
Hassan, Ahnaf Rashik 1, 2 ; Keshavarz, Behrang 2, 3 ; Yadollahi, Azadeh 1, 2
1. Institute of Biomaterials and Biomedical Engineering, University of Toronto; 2. Toronto Rehabilitation Institute-University Health Network; 3. Department of Psychology, Ryerson University
Background: In Ontario, 20-25% of all car accidents are related to drowsy or fatigued driving, making it one of the top five causes of accidents on Ontario's roads. Existing systems for drowsy driving detection are typically expensive, dependent on various parameters such as vehicle type, lighting conditions, or road geometry, or are often inconvenient for the driver. Therefore, a reliable drowsiness detection system is yet to emerge. This research aims to fill this gap by developing a convenient and wearable device that can reliably detect drowsy driving and can be used to alert vehicle drivers who are at risk of falling asleep while driving, thus minimizing car accidents.
Method: Canada’s most advanced driving simulator is located at Toronto Rehab (DriverLab) and will be used for this study. Twenty healthy individuals will stay awake at night, and perform three driving tests at 8PM, 12AM, and 3AM with no sleep in-between. Each driving test will contain a 45 minutes long, monotonous highway scenario that will promote sleepiness. A wearable device developed in our lab, called The Patch, will be used to record cardiorespiratory signals and head movements of the subjects during the driving tests. In addition, data from current gold-standards to detect drowsiness, including electroencephalogram (EEG), eye tracking, and steering wheel movements will be recorded during driving. Using the cardiorespiratory signals and head movement data recorded by the Patch, features such as heart rate variability, respiratory rate, and head position will be extracted. Pattern recognition tools, such as partial least squares regressions, will be used to develop models that will detect drowsy episodes. The results of The Patch will be compared to those obtained from the current gold-standard to validate The Patch to detect drowsy driving.
Expected results: We anticipate that drowsiness will be associated with reduction in respiratory and heart rate, reduction in respiratory and heart rate variability, decrease in respiratory intensity, and abrupt movements of the head. These alterations in the cardiorespiratory signals will be captured by The Patch, which, in turn, will be used by the drowsiness detection model to detect drowsy episodes. Unlike most of the existing drowsiness detection systems, The Patch will be inexpensive, non-intrusive, easy-to-apply, independent of lighting conditions, and convenient for the driver.
Significance: Upon successful validation of the Patch for drowsy driving detection, this research will help to develop a reliable, wearable, and convenient device that will increase road safety.