By Paul Horrell
Adjust font size:
(PopSci.com) -- I'm driving through eastern France, the blip-blip of the lane markers zinging backward through my peripheral vision at about 90 mph. I check the mirrors: nothing there.
Pretending to doze off, I let the car drift gently to the left. Just as it begins to veer over the dotted line, the left side of my seat vibrates, activated by an infrared sensor looking at the road paint. Meander right, and it's my right thigh that gets the warning.
If this really had been a case of inattention rather than journalistic inquiry, I can assure you that the buzzing seat would have jolted me back to the job at hand. The car I'm driving is a prototype from the French automaker Peugeot CitroŽn, but a showroom-ready copy isn't many months away.
Flash back five months: I'm at a test track at the base of Mount Fuji in Japan, in a Lexus fitted with a pre-crash safety system. I drive down the track at about 40 mph toward a rubber post.
Instinct and education make it hard to keep my foot off the brake, but a group of earnest engineers insist that I aim their $70-grand sedan straight for the post. At the last fraction of a second, when the car's radar sensors and microprocessors have determined that the idiot at the wheel really isn't going to steer around the obstacle, the safety system shoots into action.
Seatbelt pre-tensioners cinch up, and the front and rear suspension dampers stiffen. As soon as I touch the brake pedal -- better late than never -- the car's brake-assist tugs them on at max effort. It isn't enough to avoid the crash, but the impact speed is about half what it would have been without the new system.
The next generation of environment-sensing cars will use more than just radar and infrared sensors to watch for signs of trouble. Video cameras will look for stoplights that have turned red and for children who are running toward the road. Distance-sensing lasers will check for vehicles in the driver's blind spot and the passing lane.
These sensors won't do anything that a vigilant driver can't already do, but what if they could? What if your car could sense road conditions and traffic problems that are out of your sight? That's coming too.
The next giant leap in sensing will be radio networking that enables cars to exchange information.
"Communication [between cars] will be like an additional sensor," says Ralf Herrtwich, director of vehicle IT research at DaimlerChrysler.
Car-to-car communication will ensure that your automobile is impeccably informed about road conditions ahead. And this extra "sensor" will have almost unlimited range, because information can be instantaneously relayed from one vehicle to the next, to the next, and so on.
No one doubts the extent of information-gathering and communication features that will be built into the networked cars of the future. Some of these features will merely assist the driver by, for example, pointing out a patch of black ice around the next bend.
But what about the driver who fails to act on the warning? Should his car be empowered to "take the wheel"? Some automotive experts foresee a day when our cars will be so well informed that we'll be better off leaving some of the driving to them.
Time for another demonstration. It's Berlin, and I'm in a Smart car, DaimlerChrysler's tiny two-seater that has become a familiar sight in Europe's cramped city streets.
But this Smart is different: It's smart. On the dashboard, a flashing display warns me of an accident two streets away, and the navigation system suggests a detour. My car, outfitted with a GPS position finder and an off-the-shelf wireless local area network (WLAN) communication system, was tipped off by another car carrying the same gear.
If knowledge is power, then the intellectual-horsepower rating of tomorrow's vehicles is going to be tarmac-shreddingly high. Say just one car's stability-control system is activated at an unusually slow speed on a highway off-ramp. It'll send out a slippery-road warning. All WLAN-equipped cars in the vicinity then get the message, but they'll warn their drivers only if they're headed for the same off-ramp.
The system will also provide traffic information on a need-to-know basis. Imagine there's a truck unloading in the next street on your route. It would never make the radio reports, but you could be trapped fuming for 10 minutes. WLAN -- "traffic radar," as Herrtwich puts it -- will let you know and reroute you. What's really new here is the way traffic will behave almost biologically, like a swarm of bees, a self-educating network.
This is a killer app, because it doesn't require expensive infrastructure. No traffic-control center or information exchange. No need for roadside beacons that the authorities would have to install. Instead cars will seamlessly set up ad hoc networks, passing information from car to car.
Virtually all of the necessary hardware is already on the shelf at companies such as Bosch, Delphi and Samsung. What's needed now is the software to tie everything together: sensors, wireless radio networks and GPS navigation systems.
Together these technologies create a system that provides immediate warnings of delays, accidents, temporary speed restrictions and road conditions -- the everyday hazards that lie in wait just around the corner.
And because the system knows exactly where each driver is, it won't drown drivers in a running commentary about what's happening on the other side of town (unless the other side of town is the destination they have programmed into their navigation systems).
Of course, the system will not work well if there aren't enough vehicles outfitted with the gear. The question is how to reach critical mass. Fortunately, WLAN networks are good for more than just traffic radar; they're also useful for downloading entertainment.
"Ten years from now, we're talking about a radically changed way of listening to music, watching TV and videos," Herrtwich says. Once WLAN is adopted for in-car entertainment, the technology could also be used for road-safety and traffic-networking functions. Herrtwich puts the cost at much less than the navigation system it will be paired with. Call it a couple hundred of today's dollars.
But don't hold your breath. Technical standards are still a few years away -- Herrtwich predicts 2008. If the first networked cars roll out in 2010, such features won't be standard for at least another decade. This glacial rate of advance might give us time to settle some of the arguments that have begun to brew over the limits of driver-assist technologies -- the point at which the car, having sniffed out a veritable wealth of information about the world around it, switches (perhaps unbidden) to autopilot.
Try this thought experiment: Your smarty-pants car is fitted with something called intelligent speed adaptation. It knows the speed limit everywhere -- permanent limits are on the navigation system, temporary ones are sent through the wireless link. And it punctiliously obeys, its gas pedal falling limply ineffective as you approach the limit.
Also, your vehicle can upload its position to road authorities, so they can use variable road tolls as a traffic management tool -- raising the price on busy stretches during rush hours. How do you like the notion that someone somewhere always has the position (and speed) of your car logged?
Meanwhile your vehicle can sense highway lane markers and the position of the car ahead. It can even put on the brakes if it senses a hazard. Who's in charge here? Sure doesn't seem to be you, the so-called driver.
Some researchers insist that drivers simply can't be trusted and that external intervention is the only way to save them from themselves.
In 1997 Oliver Carsten, a professor of transport safety at the University of Leeds in England, coordinated a major European research project on intelligent speed adaptation. He concluded that if drivers were physically prevented from breaking speed limits (including temporary ones imposed because of bad weather or traffic), there would be a 59 percent reduction in fatal accidents.
But other experts insist it's the controls, not the drivers, that can't be trusted. Raymond Freymann, managing director of BMW's research and technology group, gets decidedly agitated at the mention of speed restrictions, and not just because customers are unlikely to pay top dollar for a hobbled Ultimate Driving Machine.
All the critical systems in a modern car -- engine, transmission, brakes, stability controls -- are linked to a single electronic system called the Controller Area Network bus, or CAN bus. If the CAN bus can be programmed to control the speed of your car, Freymann says, it could also be vulnerable to hackers -- which, at 70 mph, is chilling to contemplate. A roadside assassin, he notes, could disable your brakes by remote control.
"If somebody wanted to kill you, that's just the way he would do it."
German technocrat though he is, Freymann also has doubts about the reliability of automatic collision-avoidance systems. He argues that with antilock brake and stability systems, 100 percent reliability is possible because they merely survey signals from within the car: wheel speed, pressure applied to the brake pedal, and so on.
"The moment we go outside the car and look at the environment, it becomes very complicated. There is no time to react. All the decisions must be taken within a second."
Ford's chief of environmental and safety engineering, Sue Cischke, is similarly cautious.
"We could do collision mitigation by reducing collision velocity. But we have to give such systems lots of time under the development curve. You need perfect knowledge of an impending accident, and that's what we don't have."
Despite these limitations, Ford and other automakers are moving ahead with experimental systems that could help cars anticipate accidents.
That Lexus pre-crash safety system is no more than a toe-in-the-water stage in a worldwide rush of R&D effort to support -- or usurp -- the driver in moments of danger. Video-processing algorithms will soon be powerful enough to recognize another vehicle on a collision course. And if the driver is oblivious to the peril posed, the vehicle can apply its own brakes in time to stop.
But oddly enough, no automaker yet has public plans to fit such a fully autonomous system to a showroom vehicle. There are two reasons: One, drivers like to think they're in charge. Two, if something goes wrong, drivers like to sue. And they tend to "sue the money," the money being deep-pocketed automakers.
"People say they never want the car to take control," Cischke remarks, though she adds that with the hindsight of having had an accident avoided, they'll probably take a different view.
When antilock brakes were new, some people feared the technology would lengthen stopping distances, but today most drivers wouldn't be without them. Collision avoidance, however, poses a greater likelihood of litigation. As VW's former head of research Ulrich Eichhorn puts it, "People will sue the hell out of you -- sue if the system doesn't stop you, and sue if it does on a false alarm and your grandmother in the back breaks her neck."
Americans are more litigious than Europeans, which is one reason why most experiments with "driver support systems" are happening in Europe. Cramped European city centers -- many were laid out centuries before the automobile was invented -- also provide incentive for adopting technology that can help unsnarl traffic. And European culture is generally more tolerant of restrictions on individual freedom.
A case in point: In a European Airbus, if the pilot pulls the stick back so far that the plane is about to stall, the stick effectively locks up. If an American-made Boeing jet is about to go outside the envelope, a cockpit alarm sounds, but the stick still responds.
A Boeing-like system may be the model for future American autos. For example, a car could gently increase the gas-pedal spring pressure when the speed is too high -- the navigation system knows how tight that upcoming curve is, remember -- or apply a light twist to the steering wheel if the car is drifting off course. The actual driving, though, would still be left to the driver.
Nor would drivers have to sacrifice their privacy, advocates say, because there's no need for cars to broadcast their identity in order to share useful information.
"We need to know speeds to know if there is a traffic jam," BMW's Freymann says. "But we are not interested in which car is [sending the message] or who is driving it."
Sensing and networking technologies will most likely premiere on luxury cars -- their buyers tend to bear the cost of putting novel features into mass production. But before long, all this onboard intelligence will undoubtedly benefit a new crop of small city cars. Advanced networking will make them superhandy for negotiating urban traffic snafus. And collision-avoidance systems will provide an extra margin of safety for the driver of a small car who's worried about being crushed by a larger vehicle. Light on muscle, maybe, but heavy on intelligence.
The development path of the technologies isn't at issue. Neither are the motivations behind them: improved safety and convenience. The trick will be working out how we pesky humans will react. Will we fall into line behind the bureaucrats who think we deserve nothing more than to be controlled? Or will we assert our inalienable right to louse things up, each in our own individual way?