ANN ARBOR, Mich. -- Imagine for a moment the comfort of having your car driven for you by a finely tuned, meticulously programmed computer, while you relax in the passenger seat, reading a novel, doing the crossword puzzle or taking in the scenery without a care.
Now imagine that instead of a computer, it's a New York cabbie driving. Or your grandfather. Or your teenage daughter. Or my teenage daughter.
The car is still being driven for you, but suddenly, that passenger seat isn't so comfortable.
As drivers, we're never as comfortable as when we or someone with the same risk tolerance and driving style is at the wheel. In that sense, even a well-trained computer might frighten us by taking a curve faster than we'd like, or exasperate us by hewing too blindly to the speed limit.
This is something the pioneers of the autonomous vehicle think hard about. Their work is motivated by the pressing need to make our roads and the interactions between vehicles safer overall. But they also know that for their vision of safe, self-driving cars to take hold, each "operator" must feel comfortable with the decisions those cars make on when to brake, how fast to accelerate, when to pass or how fast to take a turn.
At the map-making company Here, a Nokia subsidiary and major auto supplier that's just been sold to an auto industry consortium, they call this challenge "humanized driving" -- making automated driving comfortable and acceptable for humans, in order to avoid unsettling surprises.
"One of the reasons to build this humanized driving is to make these vehicles not behave very differently from other vehicles on the road," said Ogi Redzic, head of connected driving for Here, in an interview on the sidelines of last month's Automated Vehicles Symposium. "One part is to make humans inside comfortable, but just as big a reason to do something like this is to make people in other vehicles comfortable."
So in developing the next generation of maps that will guide automated vehicles, Here isn't just taking high-res pictures of the world. It's also combing through 15 years of traffic and GPS data that it has collected and recorded to study how real drivers behave in certain situations, such as braking on an exit ramp, or taking a sharp curve. The results of that massive analytics project will be translated into the algorithms that automakers and suppliers use to program the behavior of vehicles.
"At some level, you'd call it a map," said Redzic. "A map not of roads, but of human behavior."