IIHS/MIT Studies Find that Drivers Learn to Avoid Restrictions on Partial Automation Systems

“Programming today is a race between software engineers striving to build bigger and better idiot-proof programs and the Universe trying to produce bigger and better idiots.” That was the observation of writer Rich Cook in his 1990 book, The Wizardry Compiled. “The Universe is winning,” he concluded.

If that was true in 1990 it is doubly so now, according to new research by the Insurance Institute for Highway Safety, which has found that drivers are more likely to engage in distracted driving when using partially automated advanced driver assistance systems (ADAS).

Working with the Massachusetts Institute of Technology’s AgeLab, IIHS researchers found that drivers actively skirt the measures automakers’ programmers put in place to prevent them from shirking their responsibility to attentive driving. They looked at Volvo’s Pilot Assist and Tesla’s Autopilot and the news was bad in both cases.

Tesla Autopilot drivers learned the interval of the attention reminder and made quick motions to head off those alerts even while using their phones. If ignored, those alerts lead to intervention by the computer that prevents the driver from continuing.

Drivers using the Volvo system were much more likely to engage in activities like checking their phone or eating a sandwich while the system was active than when it was turned off, the IIHS found. “These results are a good reminder of the way people learn,” said IIHS President David Harkey. “If you train them to think that paying attention means nudging the steering wheel every few seconds, then that’s exactly what they’ll do.”

Related:IIHS Finds Significant Increase in Pedestrian Fatalities from Tall, Boxy Vehicles

“In both these studies, drivers adapted their behavior to engage in distracting activities,” Harkey said. “This demonstrates why partial automation systems need more robust safeguards to prevent misuse.”

Volvo

To study the Volvo system, the team looked at how the driving behavior of 29 volunteers driving a 2017 Volvo S90 sedan that was equipped with Pilot Assist evolved over a four-week period. This was how they documented how likely drivers were to engage in nondriving activities when using the automation on highways and charted how this changed during the test.

Over the course of the three-year study, the Volvos received two software updates that improved the lane-centering feature and attention reminders.

Researchers separated the drivers into three groups with one group doing all their driving before either update, a second group that drove the vehicles after the changes to the lane-centering feature, and a third group that did their driving after both the lane-centering feature and the attention reminder updates.

Related:IIHS Finds Drivers Overestimate Partial Automation Systems

In all cases, the Pilot Assist system required the driver to keep their hands on the steering wheel. All three groups of drivers engaged in distracting visual-manual activities, including eating, grooming, and using electronics. They did this even more often when using Pilot Assist than while driving without it and this was the case for both infrequent users and power users of the system.

“We saw some differences in how the three groups used Pilot Assist, but we couldn’t say for sure they were related to the software changes,” said IIHS Senior Research Scientist Ian Reagan. “The takeaway for me was that the technology was linked to more distraction for all three.”

Tesla

For the Tesla Autopilot study, a different group of IIHS and AgeLab researchers studied how the driving behavior of 14 Autopilot newbies changed over a month behind the wheel of a 2020 Tesla Model 3. They watched for how often the drivers triggered the system’s initial attention reminders, escalated warnings, emergency slowdown, and lockout protocols.

The Model 3 used a torque sensor in the steering column to judge whether the driver was paying attention. When the system thinks the driver’s hands aren’t on the wheel for a short time, it issues an initial attention reminder in the form of a gray hands-on-wheel icon that appears on the central display, along with a flashing blue light and a written message instructing the driver to apply a slight turning force to the steering wheel.

Related:IIHS Flunks Half the Class in Nighttime Collision Avoidance

The driver can dismiss the alert by making a slight steering adjustment, tapping the turn-signal stalk, or toggling a dial on the steering wheel. If the driver fails to respond, Autopilot begins a series of escalating visual and audible alarms and alerts. Eventually, Autopilot will even slow the vehicle to a stop, deactivate the driver assistance, and deny the driver access to the feature for the rest of the trip.

Over more than 12,000 combined Autopilot miles drivers triggered 3,858 attention-related warnings from the partial automation system. About half of those alerts were false alarms when they had at least one hand on the steering wheel but were apparently not moving it enough to satisfy the torque sensor.

Normally the initial reminder is enough, with drivers responding within about three seconds, usually by wiggling the steering wheel slightly. But 72 times the driver didn’t respond fast enough to prevent the alerts from escalating and 16 persisted to the point of the driver being locked out of the system. There is no information, however, on how many of these might have been intentional tests by the driver to learn the car’s responses.

During the test, the rate of initial attention reminders per thousand miles traveled with Autopilot increased by 26 percent, while the rate of escalations fell by 64 percent. The percentage of time that drivers were disengaged in the period surrounding the alerts also increased, even though the duration of each alert was shorter.

This was because drivers did nondriving secondary activities, looked away from the road, and had both hands off the wheel more often during the alerts, and in the 10 seconds before and after them as they learned how the attention reminders worked. The longer they used the system, the less time it took them to take their hands off the wheel again once the alerts stopped.

“These results show that escalating, multimodal attention reminders are very effective in getting drivers to change their behavior,” said IIHS Senior Research Scientist Alexandra Mueller, the study’s lead author. “However, better safeguards are needed to ensure that the behavior change actually translates to more attentive driving.”

It sounds like the ball is back in the software engineers’ court in their never-ending battle with the Universe’s infinite supply of bigger and better idiots.