UT Austin study finds some ADAS warning systems invoke bad driving behavior

Some ADAS warning systems could be worsening driver behavior, according to a new study from the University of Texas at Austin. 

Ashish Agarwal, associate professor of information, risk and operations management at the university’s McCombs School of Business, said automakers could reduce risky driving behaviors by knowing how humans react to different kinds of warning signals. 

At the request of an unnamed vehicle manufacturer, Agarwal worked with Cenying Yang of the City University of Hong Kong and Prabhudev Konana of the University of Maryland to study how people reacted to driving features. 

The study analyzed data from the vehicle manufacturer for cars sold in 2018 and 2019, a release from the college says. Data collected from the onboard sensors included trips, speeds, and acceleration rates for 195,743 vehicles. Some of the vehicles had ADAS features, and some did not. 

According to the release, the team’s main question was: Do these systems affect the general driving behavior of users over time, and, if so, how? 

The team focused on hard braking and speeding rates, and how both coincided with blind spot detection and land departure/forward collision warnings. 

“We compared cars with blind spot detection with the ones that don’t have that to see how it influences speeding and hard braking behavior,” Agarwal says in the release. 

The data showed that the two kinds of warning signals had opposite effects on driving behavior. 

Blind spot detection reduced the daily number of hard braking events by 6.76% and speeding events by 9.34%, compared to vehicles that didn’t have the feature. 

However, lane departure/forward collision warning led to a 5.65% increase in hard braking and a 5.34% increase in speeding. 

The research found that these effects were magnified over time. 

For each extra month that a driver used blind spot detection, they sped 0.40% less, while they sped 0.32% more for every added month that they used lane departure warning. 

According to the release, the two safety features prompt different behaviors in line with a psychological concept called dual process theory. 

The urgent warnings require a driver to correct course immediately, the release says. This causes System 1, or reactive thinking, which is automatic and unconscious. 

“It triggers risk compensation behavior, which impedes your learning and makes your behavior worse,” Agarwal says in the release. 

Blind spot detection does not demand an instant reaction, the release says. It allows time for deliberation, or System 2 thinking. 

Agarwal says in the release that automakers should consider these findings as they design next-generation ADAS features. 

“For example, a car could repeat a warning signal after a lane departure as well as during,” the release says. “That might encourage System 2 thinking, helping a driver reconsider their risky behavior.”

Agarwal says that for learning to take place, a user needs to be in System 2 mode. 

“That means that you learn, and your behavior improves over time,” he said. 

The research, “General Behavioral Impact of Smart System Warnings: A Case of Advanced Driving Assistance Systems,” was published in Production and Operations Management of Sage Journals.

IMAGE

Photo courtesy of metamorworks/iStock