The Perfect Parking Garage: No Drivers Required
With security, reliability, and legal issues yet to be resolved, the first self-driving vehicles will perform only specific tasks.
More than 90 percent of traffic accidents are the result of human error. Congestion also costs more than $121 billion in the U.S. each year.
Drivers who use a parking garage in Ingolstadt, Germany, could be forgiven for thinking they’ve died and gone to commuter heaven. They can pull up outside, step out of their car, and let it drive into the garage to find a parking spot for itself. Later, they simply press a button on a smartphone app and their car will obediently return to the garage entrance.
The garage is an experimental project run by the German carmaker Audi; outfitted with numerous laser systems that map the environment in 3-D, it allows specially outfitted cars equipped with radar and wireless receivers to snake their way up through the garage while navigating past other cars, sense an open spot, pull in, and shut off the motor—potentially shaving minutes off a person’s daily commute.
It will likely take a decade to perfect and implement the technology. But the project demonstrates how fully autonomous driving may one day be possible. Vehicle autonomy is advancing at remarkable speed, and it promises to make driving safer, easier, and considerably less annoying, but will appear first in very controlled contexts such as the closed environment of a parking garage.
In many ways, developing the technology is the easy part; automakers now face many practical road bumps as they try to bring that technology to the road.
“It’s actually going to take a while before you get a really, fully autonomous car,” says Annie Lien, a senior engineer at the Electronics Research Lab, a shared facility for Audi, Volkswagen, and other Volkswagen Group brands in Belmont, California, near Silicon Valley. “People are surprised when I tell them that you’re not going to get a car that drives you from A to B, or door to door, in the next 10 years.”
The problem, according to Lien and others within the industry, is that the kind of technology seen in many demonstrations given by automakers and in Google’s self-driving cars (see “Self-Driving Cars” and “Is This Why Google Doesn’t Want you to Drive?”) is nowhere near ready for the showroom. The hardware needs to become cheaper and more compact; systems need to be easy and intuitive to use; security and reliability has to be assured; and legal issues need to be ironed out. “You can make a car drive fairly autonomously, but it’s not a business product, and that’s the difference,” Lien says.
Both Audi and Toyota attended this year’s Consumer Electronics Show (CES) in Las Vegas to preview advanced autonomous technology. Toyota showed a Lexus fitted with similar equipment to Google’s self-driving cars and showed videos of the car driving autonomously around a racetrack. Audi’s demonstrations, meanwhile, were more focused on practical considerations. It showed off a compact laser sensor designed to fit inside a car’s body and sweep the road ahead to build a 3-D picture of the vehicle’s surroundings (see “Audi Shrinks the Autonomous Car”).
That sensor is far smaller, and far cheaper, than the towering device that performs a similar job from on top of one of Google’s cars, which use a lidar instrument, made by a California company called Velodyne that costs around $70,000. The drawback with Audi’s more compact device is that it offers less resolution and accuracy.
Audi also showed off its self-parking system in Las Vegas by modifying the parking garage of the Mandarin Oriental hotel. The system uses laser ranging equipment to relay 3-D imaging information to the vehicle, which mean fewer bulky and expensive components need to be packed under the hood. But Lien says it will be a while before this is available; it would need to be standardized and then built into high-end office buildings or apartments. Audi also demonstrated two semi-autonomous parking systems at CES that could appear within the next five years. The first performs a perpendicular parking maneuver while the driver sits behind the wheel; the second lets a driver park a car into a home using a smartphone or tablet.
All major carmakers are developing self-driving technologies. The man responsible for the General Motors effort, research program manager Jeremy Salinger, says the company is already experimenting with vehicles capable of steering themselves in highway traffic. But he stresses that it will be important not to rush the implementation of such technology, and to make sure the interface is right. “A driver’s relationship with a car changes suddenly,” Salinger says, “from one where they’re constantly pushing the car where they want it to go physically, to a supervisory role in a sense. We have to be careful to do that in a way that people understand what the limits of the system are.”
The National Highway Traffic Safety Administration is already preparing for the inevitable. It has created a $2 million research project focused on autonomous driving. This effort will explore driver-vehicle interaction issues (when and how autonomous driving technology takes over), security and reliability, and ways of measuring the performance of autonomous systems.
So far, most testing has involved skilled drivers or engineers interacting with the technology—not ordinary drivers. “We don’t yet have self-driving cars in a realistic sense,” says Bryant Walker-Smith, a lecturer at Stanford Law School who specializes in the issues surrounding self-driving vehicles. “We have vehicles that are determining their own path under the constant monitoring of trained professionals.”
Walker-Smith adds that the development of autonomous vehicle technology will be significantly affected by legal issues. Although Nevada, California, and Florida have passed laws to make autonomous driving legal for testing purposes, so long as human drivers are in the car, the legal liability issues have not yet been worked out. This issue will be complicated by the fact that for a long while there will be mixed levels of autonomy on the roads.
“We don’t know how the public is going to respond when bad things happen,” Walker-Smith adds. “We don’t know how judges and juries are going to apply existing laws when we see the first crashes.”
Become an MIT Technology Review Insider for in-depth analysis and unparalleled perspective.Subscribe today