The SPC RAS researchers, together with researchers from Austria, have developed a methodology for evaluating the effectiveness of information systems tracking the driver states potentially leading to road accidents: drowsiness, talking on the phone, malaise, inattentive driving, alcohol intoxication, and so on. Using the proposed method, scientists from the above two countries will run a series of experiments on a simulator that will significantly improve the accuracy of existing systems for monitoring dangerous human behavior while driving. The results of the study will be published in the scientific journal IEEE Access.
"Today, the topic of the road safety improving through various intelligent systems for monitoring the driver's condition is an important one that allows for understanding what means are the most effective and under what conditions. Therefore, together with Austrian scientists, we have developed a methodology for evaluating the effectiveness of such systems. To arrive to such a result, have been reviewed almost all existing methods intended for determining the driver's condition (over 200 extended articles). Also the results of the long-term research in this field run by us have been exploited" - says Alexey Kashevnik, Senior Researcher at SPC RAS.
It should be noted that St. Petersburg scientists are developing in the field for almost five years the approaches to monitoring the driver's conditions with the help of "computer vision". Previously, they created the Drive Safely software, allowing to store and analyze data about the driver's condition at the wheel: fatigue, concentration and aggressive driving behavior. The information is sent to the cloud storage from the driver's surveillance camera. At the same time, the system can work from a mobile phone. In addition, as part of creating a new methodology, Austrian scientists provided the results of studies of the driver's behavior in the car, equipped with various sensors to measure the characteristics of the car's movements (speed, turns, directions of movement, etc.) and the driver's well-being (breathing, pulse, heart rate, and some others).
"As a result, within the framework of the Russian Scienсe Foundation Project "Models and methods of intelligent driver support based on monitoring the situation in the cab of a vehicle", a methodology have been developed showing what sensors (or the result of their joint work) can most effectively detect a specific unsafe driver’s behavior. At that all types of distractions have been broken down into three large groups. The first one considered a case when a person "takes his hands off the steering wheel", that is, smokes, drinks or talks on the phone (manual distraction). When, for some reason, he takes his eyes off the road, even if his head is looking forward (visual distraction). Also, when he may be looking at the road and his hands are on the steering wheel, but he is immersed in himself and does not think about driving (cognitive distraction)"- notes Alexey Kashevnik.
Within these categories, scientists have considered a huge number of unsafe behaviors that can fall into different types of distractions and, depending on the situation, can be interpreted differently by the system. "For instance, the methodology evaluates the driver's behavior differently when he is driving on a highway with active traffic and when he is standing at a traffic light. In the first case, if he takes his eyes off the road, then the sensor, working according to our methodology, signals dangerous distracted behavior, in the second one - this is not so critical" explains Alexey Kashevnik.
In the near future, an international team of scientists with the help of the created methodology plans to test a comprehensive system for monitoring the driver’s behavior. The experiment will be conducted by almost 20 drivers who will work for 10 hours in special simulator cars equipped with a variety of sensors. The information obtained will allow scientists to establish the relationship between the probability of driving rules violations and the physical condition of the driver, his driving style, as well as the characteristics of the car. As a result, this will lead to a significant refinement of existing systems to ensure a safe driving. A group of researchers has already submitted a joint application for a grant from the Russian Science Foundation and the Austrian Science Foundation.
"The results of the research done will definitely find application in various companies that are associated with the transport sector, for example, in taxi companies. On the one hand, it is beneficial for carriers when drivers behave adequately, do not get behind the wheel if they do not get enough sleep, or they have health problems. On the other hand, passengers are interested in making sure that the trips take place without accidents. In addition, today we see a huge business interest in such projects: now two large companies will introduce pilot projects for the implementation of driver behavior monitoring systems created earlier in SPC RAS, and a newly proposed methodology will significantly improve their accuracy, and therefore make them safer," says Alexey Kashevnik.