Making roads safer by reading drivers' moods

Researchers at Stanford University are exploring the relationship between drivers and the car of the future, and how sensory technology can make better drivers.

Carl-Gustav Linden
4 min read

Your next car might know you better than most of your friends or family do.

Stanford's Junior
Stanford conducts a wide array of automotive research. A team representing the university has competed in DARPA's Urban Grand Challenge, an event that showcases breakthroughs in self-driving cars. Stanford's robot, Junior, was the first to cross the finish line in the competition last November. Stefanie Olsen/CNET News.com

If current research pans out, the car of the future could figure out not only where you drive, what sort of music you listen to, news preferences, what you like to eat, or whom you are calling--but it might also know how your mood affects your driving. And eventually, it could turn into the ultimate backseat driver, taking full control if it's not satisfied with the way you're manning the steering wheel.

It might sound like something from the distant future, but researchers at Stanford University are working on all manner of technological improvements to the automobile. They're hoping features like camera detection of face movements, voice analysis, and sensors in the steering wheel will result in cars that can accurately detect a driver's mood and make appropriate adjustments if it's affecting their driving.

Clifford Nass, a professor at Stanford and director of its Communication between Humans and Interactive Media (CHIMe) Lab, believes autonomous driving will not be limited by the technology itself, but rather how much responsibility people are willing to outsource to their cars. We have already seen them give up some control. Antilock brakes and stabilization systems, for instance, are already standard features, and the next step could manage everything from imposing speed controls for lead-footed drivers to using sensors trained on road surfaces to guide the car by itself.

The lab is also looking at ways in which cars can improve someone's driving by giving them audible feedback. Some drivers are already having conversations with their cars every day, interacting with navigation or voice-operated music systems, for instance. But that's still a relatively new technology.

"Talking directly to your car is not all that common...but increasingly, the car is inviting you to talk. What is going to happen next is cars talking to you. There is a great interest in how cars can teach you to be a better driver," Nass said in an interview with CNET News.com at the university campus.

Researchers are busy exploring how drivers behave in simulated situations, and they've found that feedback from the dashboard isn't always welcome. In trials, a synthetic car voice might tell people, "You're not driving very well and you need to pay more attention." That message actually tended to worsen people's driving habits as they got angry, and with more sincere warnings like, "You really need to be more careful," driving deteriorated even more until the voice insisted that the driver pull over--the driver getting so furious that the trial ended with an accident, (fortunately, just a virtual one).

The tone of the car's voice is important. Studies show that happy people drive best when they get advice from a happy sounding voice (listen to MP3), while the performance of drivers who are upset improves by 40 percent if the voice is more subdued flat (listen to MP3).

For drowsy drivers, some intellectual challenges can work out well. Stanford tested people by playing them Swedish language learning tapes and found that the ones who repeated the sentences stayed more alert.

"A small amount of talkback makes people stay more awake," said Nass, who thinks it could be beneficial for a car to engage its driver in a little conversation. There are cultural differences between countries that also need to be addressed. Some years ago, the German car maker BMW had a product recall of its 5-series because German male drivers could not stand the female voice in the navigation system. In driving simulations in Japan, people got upset if the mood recognition system used in simulations told them they were sad, while that was not a problem for people doing the same tests in the U.S.

Another goal is to develop cars that help elderly people, since graying populations all over the world continue to drive, but might be aided by warning systems for red lights, stop signs, or pedestrians. Stanford has teamed up with the American Association for Retired People for this development project.

Inexperienced young drivers have a whole other set of problems, as it takes 10 years to become a good driver, Nass said.

"One possibility is to let the car take over; like when entering a highway, the car can hit the gas for you," Nass said.

Two issues likely arise down the road are advertisements (would you want your car to deliver commercial messages?) and personal integrity (do you want insurance companies to monitor your driving behavior?).

"What the car industry is going to do (with all these technology advances) is a great question," Nass said.

Car manufacturers all over the world are following the experiments with great interest. After all, Silicon Valley has, in a very short time, become the R&D center for many car-tech companies as cars get new interface technology and become computing devices. "Everybody's opening up facilities here. This is the place now for car research," Nass said.

Last year, Volkswagen gave $5.7 million as a gift to Stanford's new CarLab, an effort to pool all car-related research into one place. The CarLab will be formally opened in September. One important focus will be the transition from non-autonomous cars to a vehicle that can operate itself.

"How willing will people be to let the car take over for you? That is one of the main missions with CarLab," Nass said.