Driver awareness systems in autonomous vehicles easily cheated, study finds
Driver awareness systems are used in vehicles with autonomous abilities to detect when a driver is not paying attention to the road, but a new study from the American Automobile Association (AAA) has found these systems can be easily tricked.
Driver monitoring systems typically use either a camera or steering wheel to detect when a driver is distracted when a vehicle is running autonomously. Currently, all vehicles with a level of autonomy are required by law to have an alert driver ready to take the wheel at any moment.
However, real-world testing by AAA found that the systems using a driver-facing camera are the best at keeping motorists focused on the road, but the technology is not foolproof and a driver determined to cheat the system can still defeat it.
“The key to a safe, active driving-assistance system is effective driver monitoring that can’t be easily tricked,” said Greg Brannon, automotive engineering director at AAA.
“Vehicle technology has the potential to improve roadway safety, but the last thing we want are ineffective features in the hands of uninformed or overconfident drivers.”
Active driving-assistance systems are widely available and often called semi-autonomous because they combine vehicle acceleration with braking and steering.
Since their introduction, drivers have found ways to cheat the systems by watching videos, working, sleeping or even climbing into the backseat. This behaviour can go undetected by the vehicle and, in some cases, result in deadly crashes.
One such incident occurred in April 2021 that caused the death of two men who were being transported in a Tesla running autonomously. Neither passenger was in the driving seat at the time of the fatal accident.
Camera-based systems watch the driver’s face to detect alertness, while an alternate system is used to monitor steering wheel movements. AAA test-drove four popular makes and models in real-world conditions on a California highway to evaluate these systems’ effectiveness.
The study found the camera-based systems disengaged drivers 50 seconds sooner than steering wheel systems and were more persistent than those detecting steering wheel movement when the driver was looking down with their head facing forward, but their hands off the wheel.
On average, the percentage of time the test drivers were engaged was approximately five times greater for camera-based systems than for steering wheel systems.
Steering wheel monitoring required only minimal input to prevent system alerts, allowing up to 5.65 continuous minutes of distraction (equivalent to over six miles of disengaged driving at 65 mph).
In comparison, camera-based systems allowed 2.25 minutes of distraction during the ten-minute long test drive.
Even after issuing multiple warnings of inattentive driving, both systems failed to disable the semi-autonomous features and force the driver to take the wheel and pay attention.
“Regardless of brand names or marketing claims, vehicles available for purchase today are not capable of driving themselves,” said Brannon. “Driver monitoring systems are a good first step to preventing deadly crashes, but they are not foolproof.”
Sign up to the E&T News e-mail to get great stories like this delivered to your inbox every day.