Google on Friday announced it will begin testing prototypes of its fully autonomous self-driving cars on public roads. The vehicles won’t roam too far from their Mountain View, California, home — but this latest phase of testing could be critical both for cultivating positive consumer perceptions, and influencing future legislation affecting the nascent sector.
Google has “a few” driverless car prototypes that will be testing this summer, said Chris Urmson, director of the Google Self-Driving Car Project. The prototypes are based on the familiar Smart Fortwo-like bubble designs that have become almost synonymous with self-driving cars.
Google previously had leaned heavily on modified Lexus RX450h SUVs to conduct its research into autonomous vehicle technologies. The SUVs have logged a close to a million miles of autonomous driving on test tracks and public roads since the start of the project, and they have been traveling roughly 10,000 miles each week, Urmson said.
Mountain View residents needn’t worry about driverless cars flying around their neighborhoods at high speeds. The Google cars will operate at a “neighborhood-friendly” top speed of 25mph, and they’ll have driver on board — as well as a removable steering wheel, accelerator pedal, and brake pedal — should manual operation be warranted, the company said.
In the coming years, Google wants to launch pilot programs so that it can learn more about how people would use driverless cars. For now, though, the company wants to gauge public perception of the vehicles and discover challenges unique to driverless cars.
Roughly 94 percent of automobile accidents are caused by human error, Urmson noted, citing a National Highway Traffic Safety Administration study.
People will have to believe that self-driving cars are safe before they embrace the technology, said former NHTSA administrator David Strickland, now a partner at Venable.
“The Google vehicles are certain to perform better than people in executing the driving task safely, but there is a truism that drivers overestimate their own skills, and the first time a self-driving vehicle is involved in a crash, the media will drill in on whether the machine made an error,” he told TechNewsWorld.
The self-driving cars will have to be “virtually perfect to overcome these acceptance issues,” Strickland said. It’s why Google has been so focused on exhaustively researching every possible scenario involving the vehicles.
While Google’s cars will putter around Mountain View at Sunday-driver speeds, having the vehicles on public roads will go a long way toward helping the public warm to the idea of them, Strickland pointed out.
People are ready to embrace them, and they are starting to view the vehicles as a tangible reality more than as a “Buck Rogers sci-fi experiment,” he said.
“First adopters and technologists are clearly going to be the first in line, but the change that needs to happen is that everyday people and families need to look at this technology as normal as a smartphone, in terms of how it is integrated into everyday life and the advantages it provides,” Strickland continued.
Much like regional trains — which are embraced in spite of their frequent stops and, often, slow going — driverless cars will take vehicles off the highways and afford people time for productivity as they commute, Strickland suggested.
In the next phase of testing and research, Google will investigate more of the scenarios the vehicles will face — likely and unlikely — once the public starts to adopt driverless cars.
As promising as the technology is, there are still some defining challenges that could prove or disprove its viability for the near term, according to Kelley Blue Book analyst Akshay Anand.
While humans are prone to making errors associated with handling a vehicle, the artificial intelligence guiding autonomous cars could struggle with identifying critical elements of the roadways, according to Anand.
For example, how well will a robot be able to discern a rock from a grocery bag in the roadway?
“Google also has to constantly update each and every map, a tall task considering its autonomous cars haven’t come close to scouring a majority of the roads out there,” Anand told TechNewsWorld. “If a new red light isn’t installed onto the map system, the autonomous car might simply run the red light. This is an issue that needs addressing.”
Another challenge for Google will be securing and leveraging a supplier network, something OEMs have had years to establish, Anand pointed out.
Lacking a storied history of vehicle development, Google also will have to convince consumers of its automotive know-how. That could be less of a challenge for autonomous vehicles released in Detroit or Germany, or any of the automotive industry’s other powerhouse regions.
Despite the obstacles its faces, Anand observed, Google does have one major element going for it: resources.
“Deep coffers help with ventures such as this, and that’s something Google has on its side,” he said. “Google also has the drive, as a company, to see something like this through, as we’ve seen with other projects, from phones to Internet to maps to Gmail. If any non-OEM company is going to succeed at this, it’s likely Google.”