Driverless Cars on the Rise - Center for Technology Transfer and Enterprise Creation - Carnegie Mellon University

Sunday, January 5, 2014

Driverless Cars on the Rise

Thanks to advanced driver assistance systems (ADAS) ranging from adaptive cruise control to lane keeping assist and automated steering, today’s automobile is very nearly autonomous. The driver remains in control, of course, but built-in sensors, cameras and radars make it possible for the vehicle to take over much of the actual driving task. - See more at:
With new technologies, the fully autonomous automobile will one day be widely available. It’s just a matter of time, experts say, for the concept to move from a research and development exercise to full-scale production.

The Benefits Are Apparent

Human error is overwhelmingly to blame for the vast majority of automobile accidents today according to statistics cited by autonomous vehicle experts, and the economic cost of these accidents has been rising.
Thanks to advanced driver assistance systems (ADAS) ranging from adaptive cruise control to lane keeping assist and automated steering, today’s automobile is very nearly autonomous. The driver remains in control, of course, but built-in sensors, cameras and radars make it possible for the vehicle to take over much of the actual driving task.

For example, says Egil Juliussen, principal analyst for infotainment and ADAS at IHS Automotive in Chicago, more than 90 percent of traffic accidents are caused by human error, and in 2010, traffic accidents in the U.S. cost more than $300 billion, up from more than $230 billion in 2000. Moreover, Juliussen notes, this cost increase has occurred even as accident rates in the U.S. have been falling.

By contrast, self-driving cars are expected to save lives rather than put them at risk.

In October 2013, the Eno Center for Transportation, a non-partisan think tank based in Washington, D.C., released a research paper called Preparing a Nation for Autonomous Vehicles: Opportunities, Barriers and Policy Recommendationswhich estimates 1,100 lives would be saved and there would be 211,000 fewer crashes per year if 10 percent of vehicles on U.S. roads (12.7 million) were autonomous. With 50 percent, or 63.7 million autonomous vehicles, the estimates jump to 9,600 lives saved and 1.88 million fewer crashes. With 90 percent, or 114.7 million autonomous vehicles, the Eno Center projects 21,700 lives saved and 4.22 million fewer crashes.

Google, which has been testing driverless vehicles since 2009, now has about two dozen self-driving Lexus RX450h sport utility vehicles (SUVs) traversing roads nationwide. To date, they’ve collectively logged more than 500,000 miles, and none has caused an accident while in self-driving mode, the company says.

Additionally, self-driving cars are inherently more efficient than human-driven cars, and thus could yield time savings and environmental benefits, the Eno Center notes. Because a self-driving car is better at gauging and utilizing roadway space, it could help to reduce traffic congestion. So could the autonomous vehicles’ safety benefits alone help reduce congestion, the Eno Center says, citing Federal Highway Administration research that attributes 25 percent of traffic congestion to crashes and other “traffic incidents.”

Reduced traffic congestion would also lower fuel consumption, ultimately lowering the vehicle’s greenhouse gas emissions.

Self-driving cars also could offer new freedom to the elderly and disabled who are unable to drive a vehicle themselves.

So, with the benefits of self-driving cars clearly apparent, research and development is expanding. Automakers including Audi, General Motors, Hyundai, Mercedes-Benz, Nissan, Tesla and Volvo are working on self-driving cars. And auto industry suppliers such as Continental Automotive Systems are working to develop necessary technologies, too.

Advances Will Be Gradual

“For the foreseeable future, the human has to be in the loop,” says Raj Rajkumar, professor of electrical and computer engineering at Carnegie Mellon University in Pittsburgh. “Autonomous technology is a lot more mature than what people think, and it’s happening much faster than anyone predicted,” he says.

CMU researchers including Rajkumar have been working on self-driving cars since the mid-2000s. In 2007, a team from CMU won the DARPA Urban Challenge, an autonomous vehicle race organized by the U.S. Defense Advanced Research Projects Agency that involved driving 60 miles over a six-hour period on paved and unpaved roads at a decommissioned Air Force base. Of 125 initial teams competing, six completed the race, and their vehicles had to obey mock traffic rules in realistic scenarios—including red, green and yellow traffic lights at four-way intersections—divert themselves around blocked routes, and interact with fixed and moving obstacles such as other vehicles.

“It was a watershed moment,” Rajkumar says. “It was the first time it was demonstrated that vehicles could drive themselves, without anybody in them, following all the rules that you and I follow when driving—the first time in history autonomous driving in realistic conditions was no longer science fiction.”

Nevertheless, the auto CMU modified for the DARPA Urban Challenge was a 2007 Chevy Tahoe SUV that “looked like a science project,” Rajkumar says, with sensors conspicuously adorning its exterior and lots of additional electronic equipment inside the cabin.

Since 2011, the university has been working on a completely new platform, a model year 2011 Cadillac SRX crossover utility vehicle (CUV) that looks completely normal inside and out, having been modified with automotive-grade cameras, radars and light detection and ranging (LIDAR) systems that are like those now available factory-installed in many newer car models.

This was a technical milestone, too, Rajkumar says, because it proved that a normal-looking autonomous vehicle could operate in a high traffic density suburban environment with traffic lights and very complex intersections, as well as in a relatively simple highway environment.

Fully autonomous vehicles will be available in the 2020s, “when you could be reading a newspaper” while the car does all the driving, Rajkumar says.

“It’s a ‘when’ question, not an ‘if ’ question,” and the advance to full autonomy will come in multiple stages, says IHS’s Juliussen. In the first stage which extends through at least the next five years, he says, “the human can take over and drive.” But much later, around the year 2030, he speculates, there will come a stage when the human cannot take control, because the human-machine interface will have been removed, leaving only the machine to handle the driving task. Then everyone in the vehicle will be a passenger; there will be no driver.

By that time, he predicts, the advantages of self-driving cars will be self-evident: lower vehicle cost, improved safety, and greater convenience.

Juliussen notes that human-machine interfaces are expensive and that car prices will drop when you no longer need to add these complex systems to cars. “The cost of all the electronics and the software to perform the self-driving initially is going to be pretty high,” Juliussen says. But he adds that “as production increases and technology advances, that will drop fairly nicely.”

According to the Eno Center paper, though, it could take at least a decade for the cost of an autonomous vehicle’s added electronics to fall to $10,000. By comparison, just the LIDAR system on the roof of one of Google’s autonomous vehicles costs $70,000, and the additional cost of other sensors, software and engineering now can bring the total cost of an autonomous vehicle to more than $100,000, the Eno Center paper notes.

“It’s going to be a phased approach,” says Andy Gryc, automotive product marketing manager at QNX Software Systems Ltd., in Ottawa, Ontario, which has supplied technology to power autonomous systems since the 1990s—including technology for vehicles competing in the DARPA Grand Challenge, the agency’s first autonomous vehicle contest, in 2004. “How people define those phases is variable,” he adds, “but generically it’s from the vehicles providing some sort of assistance like adaptive cruise control, so the car can drive itself under limited circumstances (such as automated steering on highways), to autonomous driving under normal circumstances with the driver required to be on call, and then fully autonomous.”

Gryc says the lines between phase changes will be blurry. Two different automated driving systems could appear to be part of one phase or another. “I see it being more of a gradual introduction of technologies. Certainly, by 2020, I think we’re going to see close to fully autonomous vehicles,” he says.