Self-driving cars, or driverless cars, are becoming the future of driving without manual intervention. Currently, the speed of around 25 kilometers per hour (km/h) has been achieved for self-driving cars; however, the research in this direction is progressing to achieve more than 100 km/h. Automatic braking, lane-keeping assistance, and collision warning and avoiding are a few examples of automated systems being installed in cars.

The authors present an automatic driver monitoring system to monitor the uncertainty of a driver’s biological state and behavior in real time. They developed “a robust driver monitoring platform that consists of automotive on-board diagnostics sensors to capture the real-time information of the vehicle and driving behavior.” Further, it contains “a heterogeneous wearable body sensor network that collects the driver’s biometrics using electroencephalography, electrocardiogram, electromyography, and Galvanic skin response synchronized to record speed, acceleration, steering, and fuel consumption.” The system has many features: it is minimally intrusive, comprehensive, user-friendly, responsive in real time, ubiquitous, and available remotely.

For the experiments, the authors employed four drivers on a route comprising a mixture of urban roads and highways in Dearborn, Michigan. The drivers were asked to drive twice, at 11 AM and 6 PM, when there was less traffic and in rush hour, respectively. The system monitored the drivers’ behavior while driving and while conversing with passengers, and extracted the drivers’ biological state information using data mining, machine learning, and statistical analysis.

The authors generalize the results of less-complicated driving environments for automated driving scenarios. They provide insights into having a statistical measure of the biometrics for an easy and safe switching of “control between the driver and the automated vehicle when necessary.”