The gadget that could make you a better driver.

How we get from here to there.
Nov. 30 2009 12:32 PM

It Knows If You've Been Speeding

Can a device that tracks when I swerve, accelerate, and brake make me a better driver?

Illustration by Mark Alan Stamaty. Click image to expand.

The everyday act of driving suffers from a feedback gap.

Sure, there are immediate signs when you're not doing something right, like when the rumble strips jolt you awake as you distractedly drift to the side of road. But more often, our only feedback comes in the form of run-ins with the law. I recently picked up a traffic ticket—my first in almost a decade—for driving 68 mph on the 45 mph Goethals Bridge (which, as I now know, is radar-controlled). Running late for the airport, I knew I was I driving faster than I should have been but never would have guessed by how much. My feedback came in the form of a $200 ticket and the officer's gentle reproach: "You were a little hot on the gas pedal there, Thomas."

Without consistent feedback, it can be tough for a motorist to answer the question: How good a driver am I? In fact, we tend to be overconfident behind the wheel. A risky driver may go for a long time without a crash out of sheer luck, the same way a person who eats a terrible diet may live for decades with no apparent ill effects. And when external feedback does arrive, in the form of a honk or comment from another driver, it is likely to spark cognitive dissonance (What's their problem?) in the face of our carefully constructed sense of self-esteem. 

But what if you had something in your car that was monitoring your behavior (and I'm not talking about your spouse)? What would you learn about your own driving? Would it change the way you act?

I was in pursuit of answers to these questions when I recently installed a ROVER in my car. The ROVER, from a Colorado company called Cartasite, is a small black box that consists of a three-dimensional accelerometer—which can measure rapid acceleration, hard braking, and erratic lane shifts—as well as a GPS and the ability to send data over cellular networks. Basically, it's like an iPhone that happens to be plugged into your car's Onboard Diagnostics port, that small interface (usually found beneath the steering wheel) that mechanics use to reference car trouble and which also, it turns out, collects a wealth of ongoing data—about things like engine load or the amount of carbon in the engine—that is usually wasted.

The arrangement was simple: I would drive, and David Armitage, CEO of Cartasite, would monitor my behavior and provide some relative benchmarks (the device is also installed in several thousand vehicles in Colorado). A few words about my driving. Having absorbed, in the course of researching my book Traffic,scads of dry analyses of the various ways things can go wrong on the road—not to mention the traumatic biomechanical consequences that can follow—I tend toward caution (with the aforementioned exception). Having a 6-month-old daughter and being well out of my 20s also tends to neuter those need-for-speed impulses. I generally treat braking as losing—some kind of admission I haven't adequately anticipated molecular disturbances in the traffic flow—and find nothing more satisfying than a perfectly timed approach to a light changing green. I also live in New York City, where on some weeks my entire driving profile is summed up thusly: 1) Tuesday morning: Start car, double-park car across street. 2) Leave car parked for one hour. 3) Return car to freshly street-swept curb. 4.) Leave car there until following Tuesday.

It just so happened, however, that during the trial I was doing more driving than usual, enough at least to provide a suitable sample. What I can report is that although the device was not visible to me and was not emitting any sort of signals, I would not have been more aware of its presence if it had been a giant eye suspended from the rearview mirror. I began, in fact, to anthropomorphize it a bit; it became "David," (i.e., Cartasite's CEO, whom I envisioned as having a window open on his desktop monitoring my progress with day-trader intensity). When, on I-95 South, a florist's van changed lanes in front of me without warning, necessitating a firm press on the brake pedal, I thought, "David's going to see that one, but how will he know it was the van's fault?" When I handed over the keys to the valet at the Foxwoods Casino (I was there for a planning conference, I swear), I briefly hesitated, with visions of a Ferris Bueller-style attendant throwing off my results. And when I brought the car in for routine maintenance, the puzzled mechanic came over to ask if I knew about the strange device in my OBD port. This led me into a brief sort of John Le Carré reverie, as if I were some West Berlin station chief who's just learned he's being watched: which I was.

The reports came a few weeks later. Among the variables tracked by the ROVER are hard braking, rapid acceleration, overspeed, nighttime driving (as the report noted, "fatigue, poor lighting and the possibility of other drivers being impaired may place you at much higher risk when driving at night"), mpg, and idling. In the first report, I was pleased to see I had no hard braking and no rapid acceleration, though I did have 22 minutes of idling and an average of 1.1 overspeed events per hour. I had actually thought this would be higher, as I was definitely exceeding 55 on certain marked sections of I-95—as was everyone else—but it turns out ROVER doesn't, for the moment at least, track speed in relation to actual posted speeds; in part, Armitage told me, because speed limit data are so spotty. By the next week I had pulled something of a minor miracle, at least for Cartasite: a perfect 100 score. Over 7:39 hours and 217.92 miles of driving, I had no hard braking events, no rapid acceleration, a mere four minutes of idling (the device begins recording idling events after three minutes), and no nighttime driving.

I am, admittedly, a bit of an outlier when it comes to this sort of thing; having received my perfect score, it occurred to me that I should have placed another device in, say, a New York City taxi. But thanks to reports Cartasite shared with me (names omitted), you can get a sense of how steep the curve is: One driver, who notched a score of 72, clocked more than 30 hard braking events per hour and half as many rapid acceleration moments (Armitage suspects the driver may have been on a cell phone). In one fleet vehicle report that came in, the driver was reported to have been idling for seven hours—even though he had driven only 11 miles that day.

As with "smart metering" systems for domestic electricity use, the idea here is that feedback—by giving people a sense of what other people's behavior is, or what the ideal behavior should be—can make people's behavior change. Idling is a perfect example: Since the costs are essentially invisible (particularly for company drivers) and the behavior seemingly innocuous, there's little incentive to stop. (People also misunderstand when it makes sense to idle: Studies have shown that drivers falsely believe idling is more efficient than stopping and starting the engine for pauses up to three minutes long, when it's reality only more efficient to idle for less than a minute.) But as Michael P. Vandenberg and Anne Steinemann have argued, the societal consequence of mass idling—while small in the context of overall driving—is still huge: "[I]f Americans could be induced to idle for only thirty seconds from a cold start and ten seconds while waiting, 12.8 billion pounds of carbon dioxide emissions could be eliminated per year. This behavior change also would generate savings of roughly 640 million gallons of gas." Put another way, Americans simply sitting in their car produce more emissions than the sum total emissions (including nondriving activities) of countries like Costa Rica.

Better feedback could help us curtail driving behaviors that are linked to both increased emissions and the likelihood of crashes. Things like hard braking and speeding consume more fuel—an Australian study found that larger cars driven conservatively can get superior fuel efficiency than small cars driven aggressively—and they also seem to be predictive of crash involvement. As that same study noted in relation to fleet vehicles, "the case study found that the fuel consumption rate of crash-involved vehicles was higher than that of vehicles not involved in crashes." The things that make for efficient driving—lessening unnecessary braking by anticipating future events, avoiding higher speeds, etc.—are the things that make for good driving in general.

How effective can driving feedback be? The yearlong trial program in Denver called "Driving Change" (which employed ROVERs) recorded a 35 percent reduction in drivers' total idling time. A study involving a trial group of teenage drivers in the United Kingdom, using an in-vehicle data recorder from another company, GreenRoad, found, according to New Scientist, that "the number of unsafe driving events undertaken by each driver halved after the warning lights were turned on." The simple presence of a device seems to change driver behavior without any further intervention (one study found the device reduced crashes by 20 percent). Insurance companies, not surprisingly, have been experimenting with IVDRs—for those data would surely be more predictive of risk than the crude measures currently employed—although for now they have largely been used to measure only basic things like mileage. The question is: Would the average driver accept having her behavior monitored if the promise was lower insurance premiums? For now, the IVDR action is hottest in situations in which the driver is not driving his own car: livery drivers (whose bosses are looking to reduce fuel costs and liability claims) and teens. DriveCam, for example, sells its in-vehicle monitoring technology (which includes video snippets of unusual events) to parents of teen drivers. (American Family insurance customers can get it for free).

A number of other questions still remain: Is it more effective to give drivers real-time or retrospective feedback? How do you present that feedback? Do you need to provide positive, as well as negative, feedback? Will feedback without further incentives (or disincentives) do any good? In any case, it's clear that driving doesn't always allow for the "better angels of our nature," and the road to greater self-awareness—and safer driving—may run through your vehicle's onboard computer.

Tom Vanderbilt is author of Traffic: Why We Drive the Way We Do, now available in paperback. He is contributing editor to Artforum, Print, and I.D.; contributing writer to Design Observer; and has written for many publications, including Wired, the Wilson Quarterly, the New York Times Magazine, and the London Review of Books. He blogs at howwedrive.com and lives in Brooklyn, N.Y. You can follow him on Twitter at www.twitter.com/tomvanderbilt.