WASHINGTON (Bloomberg) -- The world has moved quickly from wonder at the idea of driverless cars to impatient expectation.
The Cadillac SRX zipping around a test track in suburban Detroit is flashing a sign: Not so fast.
The car can pilot itself at highway speed while the person in the driver’s seat eats a hamburger.
Yet the first versions of General Motors Co. autonomous vehicles, due out by 2020, will drive themselves only on controlled-access highways, such as an interstate. Don’t count on them to avoid accidents on their own; it will be up to a licensed driver behind the wheel to avoid the deer running out from the roadside. The reasons are parts technological, regulatory and psychological.
“The technology’s probably doable, but how do we implement it, how do we regulate it and how do we standardize it?” said Michelle Krebs, senior analyst with auto researcher Edmunds.com, based in Santa Monica, California, in an interview. What’s more, “there are certain people who want to be in control and they don’t want driving taken away from them.”
The cautiousness in developing fully autonomous technology, like that envisioned by Google Inc., reflects what GM officials say is a realistic view of what consumers will accept and the rules of the road will allow.
In the U.S., the federal government oversees vehicle safety and each state regulates insurance and licensing drivers.
Insurers and state authorities will have to decide how to assign liability and responsibility for an accident if an inanimate object rather than a person is driving. If an autonomous car is caught speeding or violating other traffic laws, who are laws enforced on?
“You’re going to have to have some kind of regulatory foundations in place at a reasonably nascent place,” said Bill Visnic, an analyst who follows autonomous vehicles for Edmunds.com. “Or else everybody’s going to get the feeling it’s the Wild West out there.”
The same obstacles likely exist in other countries, Visnic said.
U.S. auto-safety regulators in May released their first draft of an autonomous-vehicles policy. National Highway Traffic Safety Administrator David Strickland has said so-called active safety technologies that lead to self-driving cars are the next step in cutting U.S. highway deaths.
The policy encourages development of technologies envisioned as components of autonomous vehicles. Those include wireless vehicle-to-vehicle communications systems, brakes that apply themselves when a crash is sensed to be imminent, and sensors that are as accurate as or better than human vision.
So far, only three U.S. states -- California, Florida and Nevada -- plus the District of Columbia have laws or rules spelling out how autonomous vehicles may be used on their roads.
Proposed legislation in Michigan’s state Senate would allow and regulate automated vehicles of automakers or companies that have partnerships with them, and qualify for special license plates, said the bill’s sponsor, state Sen. Mike Kowall. A companion bill in the state House would establish rules for liability insurance for the cars.
Recognizing the policy and regulatory debates ahead, GM has designed its prototype so it wouldn’t require any changes to laws or regulations to operate, said John Capp, GM’s lead for active safety technology strategy.
“It makes sense that cars will be able to drive themselves in the future,” Capp said in an interview. “But getting there is a lot of work.”
Dubbed Super Cruise, the self-driving SUV's technology builds on an adaptive cruise control system on which development began about 15 years ago and now is available on many luxury vehicles, as well as some mainstream models.
To some extent, the GM system is an advance on old-fashioned cruise, the kind that drivers turn on to go a steady 65 miles per hour on the highway.
About five years ago, GM engineers began work to allow a car to “see” lane markings and move between two lines painted on a road, using radar to center itself. Cameras and radars around the vehicle supplement those systems to give the car situational awareness.
That’s why the system would work on a controlled-access highway. It’s not designed to brake for turns or direct itself, unlike the Google car, which uses mapping along with cameras and sensors to get around.
GM learned from testing on “hundreds of thousands of miles around the country” that small bits of road debris that a human driver would know to drive over without much thought could confuse a self-driving car, said Jeremy Salinger, head of Super Cruise research and development. Engineers have worked on radars to avoid such surprises.
Capp compares the Cadillac to a new driver to whom you wouldn’t entrust everything right away.
“As we move into automated systems, we need to make sure the vehicle is able to get itself out of a mess it gets into,” Capp said, calling Super Cruise “a realistic autonomous system that we can offer this decade.”
GM’s cautious approach isn’t unique. Daimler AG’s Mercedes- Benz in August announced a similar tempered approach, saying its autonomous vehicle will take over from the human in traffic jams. Google declined to comment.
As with many new safety technologies, GM plans to start selling its autonomous-vehicle package as an option on luxury models.
“To bring a $20,000 option package to market doesn’t make sense,” Capp said. “We’d sell two.”
Send us a letter
Have an opinion about this story? Click here to submit a Letter to the Editor, and we may publish it in print.