Feature Extraction and Evaluation for Driver Drowsiness Detection Based on Thermoregulation

Abstract

Numerous reports state that drowsiness is one of the major factors affecting driving performance and resulting in traffic accidents. In the past, methods to detect driver drowsiness have been developed based on physiological, behavioral, and vehicular features. In this pilot study, we test the use of a new set of features for detecting driver drowsiness based on physiological changes related to thermoregulation. Nineteen participants successfully performed a driving simulation, while the temperature of the nose (T nose ) and wrist (T wrist ) as well as the heart rate (HR) were monitored. On average, an initial increase in temperature followed by a gradual decrease was observed in drivers who experienced drowsiness. For non-drowsy drivers, no such trends were observed. In addition, HR decreased on average in both groups, yet the decrease in the drowsy group was more distinct. Next, a classification based on each of these variables resulted in an accuracy of 68.4%, 88.9%, and 70.6% for T nose , T wrist, and HR, respectively. Combining the information of all variables resulted in an accuracy of 89.5%, meaning that ultimately the state of 17 out of 19 drivers was detected correctly. Hence, we conclude that the use of physiological features related to thermoregulation shows potential for future research in this field.

Challenges of Driver Drowsiness Prediction: The Remaining Steps to Implementation

“…The main physiological measurement methods include Electroencephalography (EEG) [90]- [92], Electrocardiography (ECG) [93]- [95], Electromyography (EMG) [83], [96] and Electrooculography (EOG) [34], [97], [98]; moreover, some studies have also explored Galvanic Skin Response (GSR) [33], [96], Ballistocardiography (BCG) [99], Seismocardiography (SCG) [99] and Skin Temperature (ST) [36]. Photoplethsmogram (PPG) has also been applied with similar drowsiness correlations to ECG [32], [100], [101].…”

“…Hybrid methods are relatively new in the literature as the individual methods had to be established before the combined approaches [120]. Thus, most studies of hybrid technologies have reported better results [31]- [36], indicating that hybrid approaches are the best method for detecting drowsiness in terms of lowering intrusiveness levels, accuracy, and ability to function with data loss. This is due to a combination of different techniques being able to help overcome the negative components of specific techniques [13].…”

“…Studies have suggested that multi-modal (physiological) and hybrid approaches produce greater accuracy compared to a mono-signal or single drowsiness detection method [30]- [36]. Despite this, fewer studies have focused on hybrid models.…”

Special Issue on “Human Health Engineering”

“…Contributions can focus on sensors, wearable hardware, algorithms, or integrated monitoring systems. We organized the different papers according to their contributions to the main parts of the monitoring and control engineering scheme applied to human health applications, namely papers focusing on measuring/sensing of physiological variables [24][25][26][27][28][29][30][31], contributions describing research on the modelling of biological signals [32][33][34][35][36][37][38], papers highlighting health monitoring applications [39][40][41][42], and finally examples of control applications for human health [43][44][45][46][47][48]. In comparison to biomedical engineering, we envision that the field of human health engineering also covers applications on healthy humans (e.g., sports, sleep, and stress) and thus not only contributes to develop technology for curing patients or supporting chronically ill people, but also for disease prevention and optimizing human well-being more generally.…”

“…Gielen and Aerts [41] used physiological variables linked to thermoregulation to develop a drowsiness monitor for drivers. Since the process of falling asleep is accompanied by a shift/decrease in body temperature, online estimations of heat loss and heat production of drivers' can be used to monitor drowsiness.…”