How Google's robo-cars mean the end of driving as we know it

Editors' note: Be sure to catch the other stories in this package: on self-driving cars bristling with sensors, on real-world experiments with platoons of connected cars, and on smart transport grids.

Google's self-driving car initiative is moving into a new phase: reality.

Three years after first showing the world what it was up to -- rolling out a Toyota Prius with laser-scanning hardware awkwardly perched on the roof -- Google is moving its big idea out of the lab and into the real world.

Consider recent developments: A spokesman confirmed to CNET that the company was in what were described as productive talks with automakers involving Google's self-driving technology. Separately, Google is reported to be crafting a partnership with auto supplier Continental. And there's even the possibility of Google-powered robo-taxis sometime in the future.

Cars and computers: A look at the future of autos (pictures)

+15 More

See all photos

Where is this all leading? The answer, naturally, is complicated. After all, we're talking about an effort to force the biggest change in the auto industry since the first Model A drove out of Henry Ford's factory a century ago. And the automakers aren't about to let Google plow through their turf unless it helps their businesses. They have their own plans with self-driving cars, and some execs scoff at Google co-founder Sergey Brin's wide-eyed remarks about making self-driving cars commercially available by 2017.

"We do not think someone will have a fully autonomous production vehicle that soon," said Daniel Flores from General Motors' advanced technology group. "Vehicles that can drive themselves are years -- maybe decades -- away. The technology will develop in steps to allow the vehicle to do more and act incrementally as sensors get more robust and costs come down."

That might be, but such talk isn't about to diminish Google's ambitions. Google is leading the charge on this one. It's prodding the auto industry -- not to mention regulators and the insurance business -- to push the whole effort at a Silicon Valley pace.

Already Nevada, Florida, and California have legalized driverless-car testing on public roads, with a lobbying nudge from Google. And although plenty of other efforts are under way -- from GM to Nissan and Audi -- Google has the highest profile, something that surely rankles the old guard.

"Google's embarrassed the car industry by getting out ahead on this and getting all the attention," said Roger Lanctot, an automotive analyst with Strategy Analytics.

Let's not lose sight of something else: Anyone who's watched Google's swift rise in its 15-year history knows there's also a big potential side benefit. Google talks about making the roads safer, but the company's core business has plenty to gain from freeing up drivers from that task of, well, driving. How much? Americans on average spend 18.5 hours a week in a car, which adds up to a lot of time they could be checking Gmail, editing Google Docs, watching YouTube videos, and clicking ads.

"They're trying to free people's time so they can dedicate themselves to the Internet and Googling around," said Alberto Broggi, a professor of computer engineering at the University of Parma in Italy named an IEEE senior member for his autonomous vehicle work. "They're motivated by great innovation, but the bottom line...is business."

The computerization of cars
Motivation aside, a big change is under way, and it requires us to start thinking of cars very differently. Vehicles that drive themselves are the clearest example of what happens when cars transform into full-scale, general-purpose computing systems. But it's not the only example. Self-driving and connected cars will bristle with sensors, negotiate with traffic lights, talk to each other about safety conditions, join into train-like platoons, and become members of intelligent urban transit networks.

Historically, the car industry has focused on passive safety -- items like seatbelts and airbags. With the arrival of active safety technology that lets vehicles take pre-emptive action, cars will use data to help them decide what to do when drivers aren't paying attention or don't know what to do.

A lot of this boils down to communication. Sure, cars are getting more computing smarts, but so much of what's promised requires more data, and that requires ways to transmit it. How will that happen? As with personal computing, it will be a bit of a mess, with multiple networking technologies for multiple needs.

Computing today uses a handful of networks for different circumstances -- Wi-Fi, 3G, 4G, and Bluetooth among them. When automotive computing becomes a facet of personal computing, you can expect those standards to carry over. But then you can add some new network technologies designed to serve vehicles. The biggest are the 802.11p and the accompanying higher-level dedicated short-range communications (DSRC) standards, which govern how vehicles communicate with each other (V2V) and with infrastructure (V2I).

That technology, which rides the 5.9GHz frequency range for radio communications, can be used for things like collision avoidance, managing traffic at intersections, and linking cars into coordinated, fuel-efficient groups called platoons. Carmakers, however, are worried about interference on the 5.9GHz band if the US Federal Communications Commission permits other uses of the spectrum.

Traditional mobile networks
The current mobile-phone networks will find an eventual place in cars for passengers' entertainment and personal communications, but safety legislation will give the technology a head start in cars. That legislation will require emergency-call (e-call) technology in all vehicles in Europe by 2015 so cars can automatically dial for help after a crash.

After that, it's an easy next step to other wireless services that carmakers could charge for: "c-calls" for concierge-like help finding something like a nearby hotel and "b-calls." Another perk of e-call technology: It can be used to track the location of stolen cars.

Whether people's smartphones or the car itself will provide links to the network is an open question. "That's the debate going on at every car manufacturer," said Cyril Deschanel, head of machine-to-machine communications technology for Vodafone in southern Europe.

Adding mobile networks to cars raises the payment problem. Will people sign up for new data plans for their cars, or will they add new usage to existing mobile contracts by letting their cars use their phones' network? Are you ready to pay for a second, third, or fourth data plan? Infrequently used technologies like e-call aren't much of a burden to carriers, but somebody's going to have to pay if kids are streaming videos into the backseat.

One thing is all but certain: smartphones and cars will have to cooperate. The Car Connectivity Consortium's MirrorLink, for example, standardizes the interface so that a vehicle's large, bright, conveniently placed screen can act as a display for software running on the phone.

The big three apps make obvious sense for such cooperation: Those for navigation, playing music, and handling phone calls. If the technology spreads, it could lead to a super-personalized world. Bringing your phone into a rental car, for instance, would customize it immediately so you'd know how to operate it. A big perk of the approach: phones, apps, and mobile networks change fast, so relying on them helps sidestep the problem of slow-changing cars falling behind the technology curve. Everything would update the way an app does on your phone now.

Sigfox: long range, short messages
Other network technologies could have a place, too, as companies try to find the best use of electromagnetic spectrum.

One comes from a French startup called Sigfox, whose ultranarrowband communications technology can cover long distances so that gaps in network coverage aren't a problem. But it works only with very small amounts of data -- 123-byte messages at a maximum rate of 140 per a day.

Thomas Nicholls, a Sigfox marketing executive, said the network is useful for insurance companies locating stolen cars and for shipping companies tracking goods. The city of Moscow uses Sigfox's technology to monitor whether parking spaces are taken.

Another shortcoming is that Sigfox only can communicate with things moving at 19 mph or slower, but Nicholls said that's a short-term problem. "There's no technical obstacle to support higher speeds, and it is already in our roadmap," he said. Sigfox is working to standardize the technology so it's easier for others to adopt, too.

Taking on the technology incumbent is tough, though, and conventional 2G, 3G, and 4G networks have big advantages over newcomers like Sigfox, said Alex Brisbourne, president of Kore Telematics, which sells services to let vehicles tap into those networks. Mainstream carriers pick up the huge expenses of such networks -- spectrum licensing, permission to build base station sites, and data hookups to the wired Internet -- and high volume is driving equipment costs down, he said.

Another network possibility is ZigBee, a frontrunner in the Internet of things, a futuristic shorthand phrase used to describe an emerging world in which everything from door locks to livestock link to the Internet. Zigbee's shorter-range technology isn't much use for moving vehicles, but it could be useful for an electric car that's parked in the garage and has to communicate with the house or with the power company about the best time to charge its battery.

"Cars are very scary to anyone in the power grid. The nightmare scenario is everyone comes home and plugs in an entire house worth of demand at 5:30 p.m.," said Daniel Moneta, CEO of startup MMB Networks, which sells software and hardware that other companies can use to build networked smarts into things like power meters.

It's not clear just how many communications technologies will be stuffed into cars of the future. Ultimately, automakers will respond to the same technological and economic forces that steer the computer industry toward widely used, lower-cost, flexible networks.

That's infotainment
The advent of connected, self-driving cars also augurs big changes in entertainment. A network connection means passengers can watch streaming video, listen to podcasts, or download games, and self-driving can cars let drivers have fun, too. More productivity-minded people can work on the way to or from the office.

The requirement for emergency-call technology in cars opens the door for these broader uses of a network connection, preducts Rudolf van der Berg, an economist at the Organization for Economic Cooperation and Development.

"Once that's in by default, a car company would be mad if they didn't make use of that," van der Berg said. "The car becomes an apps platform."

The emergence of a software auto-update era could extend to vehicles. A carmaker could send new firmware for engine or battery control systems over the network -- as long as the process could be made secure, according to Alan Ewing, executive director of the Car Connectivity Consortium.

Indeed, security in general is a major concern with connected cars. That's an area where Google, with a vast, global computing network, already has a lot of expertise.

Better safety, better efficiency
Google's self-driving cars have driven more than a half million miles -- so far with no accidents while the cars were under computer control. Google said it believes self-driving cars can cut accident rates since computers don't get sleepy or distracted, and already there are indications that self-driving cars can be safer than the average driver. Google expects the technology will significantly reduce the number of people killed in car accidents, currently 1.2 million people per year worldwide.

Richard Wallace, a researcher with the Center for Automotive Research, suggested that a crashless world could be within reach. "You can do radical things in automotive design," he said "Why do you need all this steel and zillions of airbags? A better solution is that the car can't crash."

Shucking steel safety cages would make cars vastly lighter. And that would dramatically reduce their fuel consumption, he said.

Google believes other efficiency gains could come by squeezing more cars onto existing roads and perhaps by fostering new ride sharing practices.

A world of networked cars would spur the rise of sophisticated urban transportation management in which computer control systems detect traffic trouble spots in the data and reroute vehicles accordingly. Some cities, including Eindhoven in the Netherlands, Lyon in France, and Singapore, and are already experimenting with this concept.

Will drivers willingly yield control to a computer, to peel away the security of being ensconced in thousands of pounds of metal? Maybe not for the current crop of drivers, but they won't be on the roads forever.

"The younger generation is not as car-centric as our parents were or we are," Wallace said.

Initially, self-driving cars must be self-reliant. But as communication technology spreads, cars can start communicating with each other and with infrastructure such as roads and traffic signals and plan detailed routes in advance.

"At some point all the traffic lights and stop signs go away, because if you've got full information on your flight path through that intersection, and it doesn't intersect in time and space with any others," then you just drive on through, Wallace said.

Is it safe?
The safety of the system will depend on communications technologies called V2V and V2I that link vehicles to each other and to road infrastructure.

With V2V, cars constantly broadcast location and speed information and gather that same information from other vehicles. That lets cars constantly evaluate potential threats, especially when combined with local road data such as whether a physical barrier separates the car's traffic lane from oncoming traffic. Even with human drivers, V2V could warn that it's not safe to pass or that cars in front are braking to avoid an accident.

Adding V2I into the mix means roads could warn drivers of school zones, accidents, or traffic signals that are about to turn red. A freeway could warn of accidents, cars in blind spots, or trucks merging into a driver's lane. V2V and V2I also offer self-driving cars an abundance of trustworthy, useful information.

"We really believe in car-to-car technology as the technology for the future," said Pim van der Jagt, managing director of the Ford Motor's research center Aachen, Germany.

Early versions of the technology might be broadcast-only devices that aren't even in cars -- key fobs and smartphones that help cars keep track of their surroundings, perhaps, Wallace said. He expects V2V-equipped cars to go on sale in five years.

Vehicular wireless communications use the 5.9GHz radio frequency range, which U.S. and European regulators set aside. The 802.11p standard forms a foundation for the higher-level dedicated short-range communications (DSRC) technology for V2V and V2I.

But the Federal Communications Commission is examining whether to open that frequency band up to any usage, a possibility that . stirs unease among automakers.

Here, another regulator enters the picture. The National Highway Traffic Safety Administration (NHTSA) will decide this year what to do about V2V and V2I safety mechanisms -- including the possibility of requiring it through regulations. The decision rests in part on 3,000-car pilot project in Ann Arbor, Mich., in which drivers are warned of hazards such as collision risks at blind intersections and when changing lanes. V2V and V2I also are in use in Singapore and in Eindhoven.

A NHTSA requirement for computer-controlled safety systems would likely keep the 5.9GHz band reserved for vehicle communications, van der Jagt predicted.

"If they really go for whole new fleet at a certain date, I think it'll stay," he said. "To get into safety applications like full emergency braking at an intersection if somebody runs a traffic light, we have to be pretty sure we have the latencies," meaning minimal communication delays.

And the NHTSA is worried about the consequences of opening up the spectrum. "NHTSA is concerned about this action on the potential safety gains that could be realized with vehicle-to-vehicle technology," the agency said in a statement.

New freedom
Solving the regulatory, technological, legal, and social problems won't be easy. Even if governments help spur the connected, self-driving car through requirements such as emergency-call support, car-to-car communications, and car-controlled automatic braking, it won't be easy to hammer out a new legal framework for liability. At the same time, two economic giants -- the automotive and computing industries -- won't join forces smoothly.

But it's just a matter of time before the change does come, said Hannu Laatikainen, executive vice president of the automotive business Murata, which sells accelerometers and gyroscopes to carmakers.

"All the direction is toward the driver sitting there and watching the TV, and the car is driving itself," Laatikainen said.

That's a huge cultural change for Americans trained by decades of advertising to relish the freedom that comes with a car's mobility. Google, though, has shown that consumers can dramatically change their habits when the right technology comes along -- Internet search, YouTube, Google Maps. And even people in the auto industry expect computing to profoundly change how and why we drive.

"There will always be the people who want the freedom of driving the car on a nice windy country road," said Ford's van der Jagt. "But there also are people who just want transport, to get to work. For them, as cities are getting bigger and traffic is getting worse, self-driving cars are freedom."