Photo: General Motors

ALL STOP: Vehicle-to-vehicle ­communication like General Motor’s V2V technology keeps cars from crashing.

But how many cars and signposts must communicate to make a difference? Larry Burns, head of R&D for General Motors, says the company's modeling shows safety benefits even with less than 10 percent of vehicles outfitted with transponders. That number might be reached just by retrofitting all rental cars and other large fleets, while new cars with factory-installed transponders gradually raise the overall ratio. Even if only every 10th car is communicating, a vehicle in the fast lane might “hear” that someone just slammed on the brakes 15 cars ahead and start to slow well before the driver can see or react to the braking car ahead, says Burns.

A group of German automakers and component suppliers, along with Deutsche Telekom and several government ministries, is now writing requirements for a test in Hesse, in Germany's Rhine-Main region, that will equip more than 500 vehicles with transponders. Cars will communicate with each other but also with roadside units linked to central traffic control computers. The goal of the project, called SIM-TD, is to get real-world experience, including data that will help settle questions on what information is most useful. For example: Is a vehicle's relative trajectory adequate, or should it also transmit absolute position data from the navigation system? The first vehicles in this project are expected to hit the roads in 2009.

Photo: General Motors

LOOK OUT!: A GM system warns of cars you can’t see.

Five large automakers in North America—Chrysler, Ford, GM, Honda, and Toyota—are now defining a V2V message set; they plan to equip 50 vehicles and 20 intersections with communications technology and start on-road testing late next year. Similar efforts are underway in Japan as well.

Roughly a year ago, GM offered journalists a glimpse of the potential of its V2V research in a demonstration held at Camp Pendleton, Calif. Each of us was asked to drive a Cadillac sedan at 40 miles per hour (64 km/h) toward another Cadillac stopped ahead in the same lane. As the distance narrowed, a colored indicator on the dash turned from green to amber to red. A warning tone sounded steadily louder and faster as the cars calculated that a collision was imminent.

Shortly after the indicator turned red, the moving car braked itself, staying in lane and coming to a halt just a few car lengths behind the stopped vehicle. Forward-looking radar can do the same thing, of course. But the cost of radar transponders and image-processing software and circuitry is far greater than that of a short-range wireless transponder incorporated into each car—transmitting data already gathered by existing in-car sensors.

The great promise is that one day, a vehicle might—if the driver chooses—even drive itself autonomously. Sure, everyone loves driving down country roads on sunny Sundays. But suppose your car could handle the heavy parts of that grinding, stop-and-go, 40-km suburban commute while you answered e-mail or concentrated on that conference call. The car would speed up, slow down, and choose its routes to minimize fuel usage and emissions. What's more, it would keep traffic flowing smoothly and enable more cars to occupy the limited road space, adding freeway capacity without the need to lay more concrete.

We're a long, long way from that point, of course. But one thing is clear: even if DARPA gets its robotic vehicles in time to meet the Defense Department's 2015 deadline, the car companies that serve everyday drivers won't be adapting military technology for civilian use.