"It's okay! It's okay!" my passenger says. "The car will do it for you."

And so it does. Without my doing a thing, the Chrysler Concorde I'm driving slows down enough to leave a safe gap between me and the truck ahead.

"Whoa!" I gasp. "That was cool!"

While I'm exulting, my passenger is analyzing. The car's response, the distance between it and the truck, even my hovering foot all are of interest to Zevi Bareket, a senior engineering research associate with the University of Michigan Transportation Research Institute (UMTRI). The car we're in has been outfitted with an experimental "adaptive cruise control" system, and Bareket wrote the computer programs that control the system.

Like regular cruise control, this system maintains a cruising speed that the driver specifies. But it goes a step beyond, into the realm of "intelligent transportation systems," by allowing me to set a minimum distance between my car and the one ahead. If we start to get too close, my car automatically slows down, braking if necessary, to keep the distance I've selected. If the car ahead speeds up or changes lanes, leaving me a long stretch of unobstructed highway, the Concorde automatically speeds up to the cruising speed I've chosen.

Systems like these may make life a lot easier for drivers, but they're also making it a heck of a lot more challenging for engineers. When cars were simpler and stupider, engineers considered how drivers responded to certain types of instrument displays or how easily they could operate the pedals but paid little attention to the minute details of all the things people do while driving. Now, as cars take over more and more of the tasks that drivers used to do, engineers are realizing they need a deeper understanding of just what those tasks entail.

"We're moving from the study of a vehicle as a big hunk of steel with a driver inside to looking at the dynamic system comprising a driver, his or her vehicle and the nearby highway environment. That is, we're trying to understand how the driver drives," explains Robert Ervin, head of UMTRI's Engineering Research Division. And in doing so, researchers are asking new questions, gathering and analyzing data in novel ways and arriving at fresh insights. With approaches that borrow from psychology and sociology, as well as traditional engineering disciplines, they're forging what Ervin calls a new "science of driving."

Instead of just looking under the hood, researchers are peeking inside the head of the person behind the wheel. They're asking questions such as, How does a driver decide when and how much to brake? What cues does the driver use before braking-the sight of a car looming up ahead? the feel of the road surface? a glimpse of motion off to the side? What makes a driver decide to change lanes? How much weaving from one side of a lane to the other is typical?

Cruising in the Concorde with Bareket, another question occurs to me: How does a driver react to a partially automated car that takes over some of the work of driving? It takes me a while to get used to the car slowing down and speeding up by itself. But gradually, I begin to trust its judgment. And then I begin to trust it too much. Exiting at a ramp with no car ahead of me, I forget for a moment that the system sees no reason to slow down. It only knows to do that when my car gets too close to one up ahead; it can't read speed limit signs or understand that negotiating a cloverleaf at 70 miles an hour could be disastrous. This time, it's up to me to brake.

Clearly, drivers' notions—and misconceptions—about where the car's job leaves off and theirs begins are things that engineers must understand if they are to design safe and effective driver-assistance systems. The need to collect this whole range of information about how people drive led UMTRI researchers to undertake one of their most ambitious studies.

Over a period of 14 months, an UMTRI team trained 108 randomly selected southeast Michigan drivers to use 10 test cars equipped with both conventional and adaptive cruise control (ACC), then turned them loose to drive the cars as their own for two to five weeks. For the first week, drivers could choose to turn conventional cruise control on or off anytime they wanted. After that, the only choice was driving with or without ACC.

UMTRI researchers outfitted each of the 10 test cars with a data collection system, says Jim Sayer, an assistant research scientist with UMTRI's Human Factors Division. On each trip, onboard computers continuously collected and stored information about the car's speed and the gap between it and the vehicle in front of it. A global positioning satellite system collected data on the car's location, and a video camera mounted behind the rearview mirror recorded a view of the road ahead. A "concern" button on the dashboard was at hand for drivers to push any time they were concerned about or dissatisfied with ACC. The drivers also filled out questionnaires and participated in focus groups after returning their cars.

With 108 drivers spending a total of 3,049 hours on the road and traveling 114,084 miles, the UMTRI study yielded a huge collection of data that have already provided much valuable information.

It showed, for example, that drivers fall into several groups, classified by the strategies they use in traffic:
Hunters—Those aggressive folks who whoosh up behind you and tailgate until you move over. They drive fast and like to lead the pack.
Ultraconservatives—They are the opposite of hunters. They take it nice and slow, staying far behind the car in front of them.
Flow Conformists or Gliders—This type travels at about the same speed and following distance as other cars around them.
Planners—A shrewd bunch, they figure out how to drive fast without getting too close to the car ahead, a strategy that allows them to go for long stretches without touching the cruise control. (Bareket, who has logged more than 5,000 hours in ACC-equipped vehicles, falls into this category. Once, driving one of the cars to a technology show in the Upper Peninsula, he didn't touch the brake or accelerator even once between Ann Arbor and the Mackinac Bridge, a distance of around 300 miles.)

In the study, a given driver usually fell into the same category—hunter, ultraconservative, flow conformist or planner—whether on or off ACC. But the four types used the system differently. Flow Conformists, for instance, used ACC more often and set the system to allow longer distances between their vehicle and the one ahead. Hunters chose the shortest "headway" distance the system would allow but used ACC less frequently overall than did the other groups, possibly because it wouldn't let them tailgate.

Age was a factor, too. "Older people almost never used the 'close' setting, young people rarely used the 'distant' setting, and middle-aged people normally used the middle setting," says Paul Fancher, a senior research scientist with UMTRI's Engineering Research Division who directed the field test.

Like me, many drivers in the UMTRI study had to be reminded not to expect more of the car than it was capable of doing. They tended to use adaptive cruise control "when the world looked benign and there were fewer possibilities," Fancher says. "But they still did the tough stuff the way they always had."

Most drivers said they liked using the system, and an insight into its appeal came from data collected when they weren't using ACC. The researchers discovered that in normal driving, drivers press and release the gas pedal far more frequently than anyone would have guessed, and each use registered as a peak and valley on the UMTRI graph of their behavior.

"In an hour there can be a thousand peaks and valleys," says Fancher, "You can see why this would be fatiguing, even if people aren't aware of it. With ACC they don't have to work as hard."

A bound report nearly an inch thick details the project's initial findings. One UMTRI researcher already is sifting through it to try to learn whether time of day influences a person's driving speed and choice of roads. And Sayer is intrigued with the possibility of exploring relationships between personality type and driving style.

When drivers enrolled in the study, they were given the Myers-Briggs Type Inventory "personality test," often used in corporate personnel decisions. It classifies people by such factors as how they express themselves, evaluate other people and act on their feelings, and is correlated with scales of aggression, self-confidence and other traits.

As far as Sayer knows, no one has ever done a rigorous study relating driving style to Myers-Briggs, but there are clear reasons to take a look.

"You're hearing more and more about things like road rage on the news," he observes. "You can't help but wonder if there might be some relationship there."