The Toyota’s been modified past the point of street legal and it took a few blocks to get the feel of the accelerator. But now, comfortable behind the wheel, I casually wonder whether we should run down some pedestrians.
“No, I wouldn’t recommend that,” advises Omar Ahmad from behind the driver’s seat. His sensible voice matches his judicious wire rims and slicked coif. “You could do it — people have — but we don’t learn anything.”
A blank-faced man at the next intersection takes a step off the curb with a pompous disregard for the flow of traffic and the green light over his shoulder. Just a love tap with the bumper to remind him of his pedestrian responsibilities, that’s all I want. No crippling, just a bruised pelvic bone, some light trauma around the spleen and pancreas. So at least he will learn something.
“Right,” I agree. “What if I just point us over that grassy hill? How about booming through the college campus? Or that pharmacy?”
“Stick to the roads,” Ahmad says.
Fine, fine. Maturity is the burden of progress.
The temptation to go on a rampage is constant once behind the wheel at the National Advanced Driving Simulator in Iowa City, though this may be a symptom of gaming. This $50 million simulator is the world’s most advanced, save maybe the exact replica built on a slightly larger scale at Toyota’s labs in Japan. The car is real, rigged with cameras and a computer keypad to distract you with timed commands. The motion system gives you 400 square meters of horizontal and longitudinal travel and roughly 360 degrees of rotation. Hidden hydraulic actuators vibrate to mimic the feel of the road, and the entire contraption rests a yaw rig capable of rotating the dome on its vertical access in any direction. The only thing visible outside is a virtual world that appears on eight liquid crystal display projectors. The graphics are PlayStation 2 quality, and as a result the effect is like being plopped inside a game of Grand Theft Auto while driving the world’s most expensive controller. The major difference: The results of this round may help researchers keep you from dying on America’s highways.
Ahmad, who helped design the software for the simulation — or at least the limited intelligence directing the virtual traffic and pedestrians — has been with the program almost since its inception.
As I cruise through a close approximation of Iowa City’s business district, a computer demands I type words into a keypad fixed above the gear shift. It’s a test for handling distraction. The simulator’s research projects are a mix of commissions for government agencies and private business — public sector work, more often than not — and in 2005 it was used in a federal study on people’s ability to drive while distracted by cell phones.
The graphics are important, Ahmad tells me, because so many people have an expectation of what simulation should look like, and he needs to lull my brain into accepting the reality he’s presenting. He tells me that I was handling the machine stiffly at first, but now I was driving like most people do on real streets.
Also at issue is a slight lag between the visuals and the feel of the car. When I take a right turn, the hardware beneath me gives the physical sensation of motion, but it takes a barely perceptible fraction of a second for the virtual world to catch up. Ahmad mentions maybe 30 percent of all people are susceptible to this gap, and only 2 percent of test subjects who make it through the screening process have a problem. I have not been screened. I ask Ahmad if it’s normal for the machine to give someone a case of the spins.
“You should stop then,” he says, “because it’s just going to get worse.”
Free of the world’s most expensive Camry, I manage a full 30 minutes before I start vomiting. I really wish I’d hit that jaywalker.
There was nothing like the NADS before it was built. A military flight simulator had been repurposed for driving, but it was like moving in third gear over a sheet of ice. Convincing your brain your were in a car had a unique set of challenges, compared to tricking it into believing you were flying. Engineers had to match the feeling of tires on road, the feeling of acceleration. And most people spend a lot more time in cars than they do airplanes. To replicate such a familiar experience in a compelling way is incredibly difficult.
Ahmad joined the team in 1998, after his software — part of his master’s thesis as an Iowa graduate student — was used in the development.
“If you want to run an experiment, and it’s essential to have control of all variables, a simulator is the tool of choice,” he says. “The moment you want to involve collisions, or drugs, legal or illegal, you’re not going to do that on the road. The disadvantage is that once people know they’re in a simulator their behavior is not one-to-one exactly as it would be in real life. So we have to make the experience as realistic as possible.”
Usually when government agencies come to the NADS, it’s either reacting to a problem or getting ahead of one. An early use of the simulator was testing how drivers react when they lose control to a blowout, back in the early 2000s when Ford and Firestone were issuing major product recalls. Now as more states legalize marijuana, drug policy makers want to know how stoners behave behind the wheel. “There’s surprisingly little information on it,” Ahmad says. “We have a specific [blood alcohol] number we’ve arrived at to say, if you’re over this you are not OK to drive, but there’s nothing like that for marijuana. How do we even determine that?” This year NADS completed a study into the combined effects of alcohol and marijuana funded by the Department of Transportation, the National Institute of Drug Abuse, and the Office of National Drug Control Policy.
Drug research is the kind of work tailored to the strengths of a simulator, in which you want control of every detail, down to the grip of the road, pedestrian behavior, and cloud coverage. The hard part is the drugs themselves. The feds send the cannabis, then store it in a locked room with a triple-alarm system that automatically notifies the police if it’s tripped, and the DEA performs random checks to make sure everything’s accounted for. A Volcano vaporizer, paraphernalia of choice for wealthy stoners and research scientists, is used to ensure participants are ingesting a perfectly controlled measure of blue smoke. Then there’s the problem of getting volunteers to drive high, which is harder than you might think.
“First off you have to get the right people,” Ahmad says. “Do I want people who’ve never had cannabis before? Do I want someone who’s stoned all the time? No. I want that average cannabis user, who maybe uses once a week. Once I find him now I’ve got to convince him that I can protect his privacy. And each time you have to stay in the hospital the night before to make sure you’re not taking any other drugs. There’s a needle in your vein checking your blood every hour. The next morning you come in, we dose you with a very specific amount of cannabis. You won’t know how much you’re getting. Then after the test is over, you can’t leave, you have to wait until it leaves your system so we can drive you home. So it’s losing a day of your life, five or six times over a matter of weeks.”
That research took more than a year, and only now are the results trickling out. Every agency that put a dollar in has its own people parsing the data. What they can say for sure now is that drivers on a cocktail of weed and booze tend to weave more on the virtual roads than a driver using either one alone. They also found that a driver’s oral fluids could give proof of recent marijuana use, but not enough evidence to tell if they were impaired.
The next round of testing focuses on nailing down a driver’s state, also commissioned by the government, this time with an eye on private tech advances as automakers invest billions into making a smarter car. Dodge might not want to reveal exactly how it makes an anti-lock brake to protect trade secrets, but the mechanism is so well understood that it doesn’t really matter. They’re similarly likely to keep trade secrets hidden when it comes to upgrading their safety features, but a line of software is more arcane. How do you interpret code?
“I can be drunk, or impaired, or drowsy, and they may be overlapping. People are rarely if ever in just one state at a time. So how do we figure it out without hooking you up to electrodes?” Ahmad explains. “So we develop mathematical algorithms to establish a driving signature. How you steer, how you use the accelerator, the brake pedal. In a lab situation we can detect 80 percent of the people who are drunk-driving without labeling anyone as drunk who isn’t drunk. We want to minimize false positives. Then we look at drowsiness. We can predict it six seconds before you had a lane departure if you’re drowsy. So now we’re looking at how we can distinguish the two.”
Whatever they find, it’s uncertain whether it’ll be applied to future car tech as more and more automakers rig cameras over the steering wheel. Once a study is finished, NADS presents its findings without recommendation for a specific course of action; its engineers aren’t in the advocacy business.
At some point, between starting the engine and ending the interview via a vomit-sesh, I stopped wanting to thrill-kill virtual pedestrians. As the computer prompts asked me to enter words into that floating keyboard (the word “magnanimous” appears in my notes, a word no one has ever texted, ever), the oblivious foot traffic kept coming. I started to actually worry about driving well. I didn’t want to gun the engine anymore. I wanted to do right by the road. When a woman blithely moseyed out between sedans without checking the traffic, my foot twitched over the brake.
When the objective isn’t a race, or some high-stakes drug trafficking in a replica of ‘80s Miami, you genuinely start to care about doing a good job. I thought about the near-misses I’d had in real life. The time a pick-up truck T-boned my Dodge only to have the driver get out and hee-haw, “Hey, man, I ain’t got no insurance. I got a buddy with a junkyard; we’ll fix that right up. Don’t call the cops, man,” then sped from the scene before I dialed 911. If days like that are what the NADS prevents, then I’m happy to play along. On real roads, shit can get scary real in a hurry.