Cars that talk to you are annoying; cars that talk to each other prevent accidents

U. Michigan gets grant for live-action road test of car-to-car communications system

Cars that drive by themselves are cool, but I don't know anyone with any expertise in remote control or artificial intelligence that would stand in front of one when its engine was running.

Cars that think for themselves are hard to build and harder to trust. Cars that talk to each other while they're on the road in the same way printers or servers on the same network talk, on the other hand, could make highway driving much more efficient and a lot safer.

Car-to-car communications is about to get its first large, real-world test in Ann Arbor, Mich., where the University of Michigan's Transportation Research Institute will be building a huge test-bed using a $14.9 million grant it just got from the U.S. Dept. of Transportation.

Car-to-car blind-corner collision warning

Car-to-car communications is a kind of compromise between two opposite but almost equally impossible alternative approaches to smart individual transportation systems.

One is the DARPA robot self-driving vehicle that is smart enough to sense its surroundings, understand where it's going, what it has to do to avoid obstacles and how to know when it's gotten where it's going or hit something important (the owner, or a pedestrian) on the way there.

A lot of the DARPA competition autonomous cars are really cool; some are pretty smart, for cars. I wouldn’t trust any of them to pull out of the driveway on their own if the road hadn't been closed to traffic first.

It will be ten years before they're competent enough to be used in enclosed environments like military bases, zoos, corporate campuses and the like.

If they're smart enough to hit the real road with real drivers in less than 20 years I'll be shocked. Machines are simply not that good at predicting or reacting to human behavior. (Neither are humans, but we don't get permanent dents or have all our sensors knocked out permanently if we hit something at 20 MPH.)

DARPA autonomous vehicle challenge

The other approach – smart roads -- is more efficient in one way, unbelievably more ambitious in another.

The idea behind smart roads is to provide the interconnectivity and control infrastructure along with the physical infrastructure itself.

The roads have sensors buried in the tarmac that can identify individual vehicles as well as provide precise GPS location signals – precise enough to help a computer change lanes, not just find Peoria.

Wireless networks, traffic sensors and traffic-control substations would line the roads themselves, tracking the number, speed and control mode (Manual or Automatic) of the cars within its area.

Substations communicate with each other to monitor potential traffic tie-ups, and give instructions to the cars within their areas including how fast to go, which lane to take, how close to approach the cars in front or behind, and which cars are being controlled by humans rather than computers, so traffic on autopilot can avoid them.

The cars would also need to include control interfaces – both computers and networking gear to giver remote control to a segment of the road when necessary, or just take instructions and control the vehicle themselves in the same way a pilot takes instructions from the tower, but doesn't let the Autopilot always fly the plane.

Smart roads would save lives by preventing crashes, save time by preventing traffic jams and conserve fuel by reducing the average number of delays and number of times a vehicle has to change speed.

Smart roads which, at best, would include only major highways or turnpikes to offer a little autopilot relief to drivers, are completely impossible given the economics, logistical scale of the build-out and state of technology.

Smart-car technology somewhere between an autonomous vehicle and a smart road is already on the highway, however. Luxury and sport-luxury cars including models of BMW and Mercedes have radar or sonar sensors to detect cars in the driver's blind spot, prevent the driver from tailgating unsafely, wake up a driver who's nodding off and brew up a spot of tea for Jeremy and the boys on Top Gear when they're out doing road tests.

The next major step is to create communication and computing standards that will not only warn a driver of danger from another car, but warn the other car as well.

By communicating via short-range radio using systems referred to as either V2V or C-to-C, cars can exchange GPS location, speed, direction, destination (and likely exit from the highway) to give the cars around you enough information about your status and plans that they can warn their own drivers of impending risks. That guy getting ready to pass you on the right might get a warning from his Therapy Porsche that your exit is a quarter mile up on the right and you'll be making a panicky lane-change and sharp deceleration any second.

You might not realize why he's scowling at you in the rear-view just then, but you'll forget all about him in the excitement of trying to miss the Porsche and not the exit.

DoT reports predict as many as 80 percent of serious accidents among unimpaired drivers can be eliminated or reduced by a little car-to-car negotiation, or early warning that a sedan three cars ahead just hit the brakes even though you can't see it through the giant SUV directly in front of you.

Car-to-car warning: oncoming vehicle

The Univ. of Michigan will use the $14.9 million grant to put as many as 3,000 cars equipped with short-range radio on the roads of Ann Arbor, Mich. during the next few years.

The cars will create a real-world test environment for communications, control and automation systems being developed by consortia of automakers and academics. In the U.S. the main group is the Vehicle-to-Vehicle Communication Consortium (based as Univ. Michigan rival Ohio State University). In Europe it's the CAR 2 CAR Communication Consortium – a semi-coordinated association of eight automakers, most of which actually started working together on C2C systems almost 10 years ago.

DoT predicts some level of practical C2C safety systems will become standard technology in as little as 10 years.

Read more of Kevin Fogarty's CoreIT blog and follow the latest IT news at ITworld. Follow Kevin on Twitter at @KevinFogarty. For the latest IT news, analysis and how-tos, follow ITworld on Twitter and Facebook.

ITWorld DealPost: The best in tech deals and discounts.
Shop Tech Products at Amazon