Car manufacturers have been obliged since July 6 to incorporate new vehicle models that they want to sell in the EU a series of advanced driver assistance systems (ADAS). Among them, an intelligent cruise control, which connects the cruise control with a signal recognition system; a rear traffic detector that facilitates maneuvers or a black box that can help find out the causes of accidents. All these measures work thanks to different sensors (radar, lidar, ultrasound, external cameras) that process information about the environment surrounding the vehicle.
But data will also be collected on what happens behind closed doors. Brussels has included among the new mandatory measures another one that directly affects drivers. The vehicles that want to be homologated in the EU from this summer, the necessary procedure before being able to sell in community territory, must incorporate systems that measure the level of alertness, sleep and driver fatigue. If drowsiness is detected, an acoustic, visual or sensory alert will be sent to the user. These measures, which some brands such as BMW already voluntarily introduced in their models, will be mandatory for any new car sold from 2024 (Brussels gives the industry two years to sell vehicles that do not have these devices).
How can the car know if the driver shows symptoms of fatigue? The solution that manufacturers are opting for is to install eye tracking or facial recognition systems in cars, although without calling it by that name. In other words, small cameras placed at strategic points in the cabin from which to observe what is happening in the cabin. Bosch, for example, is in favor of installing cameras on the steering wheel that analyze the flicker of the driver.
Measuring the frequency of blinking, as well as the tilt of the head, is an effective method to infer whether or not there is drowsiness. “It is assumed that the more you blink, the more likely you are to fatigue. He has been experimenting with it for years. These systems require a certain calibration, that the person you are measuring is at a certain distance, but they are reliable”, explains José Miguel Fernández Dols, Professor of Social Psychology at the Autonomous University of Madrid and an expert in facial expressions and emotions.
Affectiva is the world’s largest provider of so-called driver monitoring systems. The American company, acquired in 2018 by the Swedish SmartEye, has contracts with 14 large car producers, although for reasons of confidentiality they can only speak openly about Volvo and BMW. Although this last manufacturer recently changed supplier, the Series 2 models have been incorporating SmartEye software since 2018 to detect the driver’s status, also present in the Series 4 and X3 to X7. “When the Urban Cruise Control function with traffic light detection is activated, the system checks if the driver is facing the road. If so, the vehicle automatically starts moving forward as soon as the light turns green. If not, an acoustic signal is activated to warn that you can now drive, ”describes a BMW spokesman.
“If you want to understand the behavior of drivers and make sure they are not tired or distracted, it is important to look them in the face,” says Gabi Zijderveld, director of marketing at SmartEye, by video call. The technology company markets systems that only focus on the driver and other more advanced ones that take into account the rest of the passengers. “We use machine learning and computer vision systems with cameras in the car to analyze the faces of passengers and detect objects to see if someone is talking on the phone. We can identify several drivers of the car, so that each one is offered their desired configuration, and detect if a child is in the back seats to warn if they stay inside”, describes the executive. Its systems have been trained with more than 40,000 hours of driving by some 6,000 different people that have led to some 500,000 different data entries.
At the moment they only use cameras, says Zijderveld, but they do not rule out incorporating additional sensors in the future to measure passenger sweating or audio from inside the cabin. That is the direction in which the automobile sector is heading, he says. “The industry has realized that if you put cameras in the car and use artificial intelligence you can collect a lot of very complex data to detect human behavior that evolves over time. And that can then be monetized.”
The market trend seems clear. The European New Car Assessment Program (EuroNCAP), a body supported by several EU governments and the main manufacturers in the sector that issues reports and tests of the vehicles that go on sale, speaks of driver monitoring systems as a technology that will become as necessary as the seat belt security. “These types of systems will be very important to improve both the safety and the driving experience of the BMWs of the future,” they say from the German brand.
security vs. privacy
It may be surprising that the community institutions advocate the installation of devices of this type in vehicles. Brussels has shown very restrictive with biometric recognition technologies precisely for dealing with extremely sensitive personal data. Just as the fingerprint can be intentionally erased or burned, the iris of human beings is the same from the age of three onwards. The patterns that describe the face are also unalterable: not even aging is capable of modifying the unique disposition of each individual’s facial features. The EU’s approach to biometrics is, for the time being, conservative. The legislation establishes that you must be very careful when collecting and using this data (only when its use is proportionate to the risks involved). The devastating potential of this technology if it falls into the wrong hands was demonstrated last year in Afghanistan: the taliban used biometric data of former officials of the State to persecute and assassinate them, according to international media reports.
But there are nuances. “The so-called EU General Safety Regulation difference between biometric information and physiological metric. The first encompasses technologies capable of personally identifying individuals, such as the retina reader or facial recognition; the second has to do with other parameters, such as the measurement of pulsations, sweat, blinking, electrodermal activation or breathing”, he maintains Lorena Jaume-Palasí, expert in ethics and philosophy of law applied to technology and advisor to the Government of Spain and the European Parliament on issues of ethics of artificial intelligence.
According to this nomenclature, the researcher highlights, biometric technologies can be passed off as something else. “Physiological metrics data biological aspects and process these data based on statistical measurements. And that is biometrics,” she stresses. “This regulation seems like a strategy to accommodate biometrics in the field of mobility without using that word,” she says. European Regulation 2019/2144, which regulates the innovations that cars must incorporate from this summer, makes it clear that “any of these security systems must work without using any type of biometric information on drivers or passengers, including facial recognition ”. However, Jaume-Palasí insists that it is difficult to place the measuring stick correctly. “Depending on the artificial intelligence technique used, something is not specified in the regulation, we can talk or not about facial recognition. Since we don’t know what the algorithm is looking at, you can’t say for sure that there is facial recognition,” she concludes.
The use of these technologies poses a series of risks or questions: Where and for how long are user data kept? Who can access them? What happens if the police ask for access to this data for security reasons? What if these data are not 100% reliable and lead to error?
The articles of the Regulation itself establish that “drowsiness and loss of driver attention warning systems (…) must not continuously record or store any data that is not necessary for the purposes for which the data was collected. , or otherwise treated, within the closed-loop system”. The data must remain in the same car and cannot be processed by third parties. That’s what Affectiva executives say they’ll do. “Our systems go in the car and don’t need to be in the cloud. We do not store images either: we use the video collected by the cameras in real time to understand what is happening in the vehicle”, describes Zijderveld.
According to sources from the Spanish Agency for Data Protection (AEPD), as it is a European Regulation, the national authorities are not competent to carry out prior evaluations of this technology. “Those responsible who want to implement facial recognition technologies must establish if their treatment could fit into any of the exceptions specified by the General Data Protection Regulation (RGPD)”, these same sources warn. Or put another way: biometric data can only be processed if the need to do so is justified.
Is it necessary to study the face?
According to a study published in 2015, there are simpler methods than biometrics to determine driver fatigue: it is enough to know how you are driving. Taking data on how and how much you brake, accelerate or turn provides sufficient and reliable information to conclude if you are paying due attention to driving. It would also be a non-intrusive method, says Jaume-Palasí, as it does not involve the installation of cameras.
Using facial expression to detect sleepiness can be misleading. “The information on muscular movements is interesting but of doubtful reliability in a face in constant movement and position changes. And the attribution of emotion is directly science fiction”, concludes Fernández Dols. him and his team have shown that it is impossible to infer emotions from facial expressions alone of people.
Its potential impact on the safety of vehicle passengers is also unclear. “Monitoring distraction does not solve the problem of distraction”, highlights Jaume-Palasí. “How does that improve the safety of the driver and his companions?” Professor Fernández Dols also asks. Or is it part of another great experiment in the style of Cambridge Analytica?”.