Will vehicles ever be able to handle the complex and unpredictable challenges of driving without ever needing humans? One leader in the field is skeptical.
Add Gilbert Gagnaire, co-founder of autonomous shuttle company EasyMile, to the list of skeptics about the ability of cars to be driverless.
Fully driverless, that is. Speaking as part of a panel at the second annual Redefining Mobility Summit in Concord, Calif., on April 21, Gagnaire said in no uncertain terms that he doubts whether cars will ever reach the Society of Automotive Engineers’ (SAE) fifth level of automated driving.
“If you ask me whether one day some will be level five, I think it’s not going to happen. Never,” Gagnaire said. “Driving in New Delhi at peak hours — serious, guys? Never. And level five means anywhere, anytime. So New Delhi, Mumbai, Paris at peak hours, Roma — no way. Never, unless all cars are autonomous.”
The quote came as Gagnaire was answering an audience question about what level of automation his company’s shuttle falls into. Under SAE’s definitions, he said the company’s low-speed shuttles are level four — that is, they shouldn’t ever require the presence of a human driver, but they’re also limited to driving on fixed routes. The company has been testing the shuttles around the world, and plans to do so in Northern California as well, but it’s always on a pre-set route the company defines. Its California demonstration, for example, will involve shuttling people around a business park.
That would, of course, have big implications for the usefulness of autonomous vehicles. Futurists and industry representatives have speculated on concepts ranging from shared-ownership AV fleets to sans-human package delivery services. Those all rely, ostensibly, on the ability of the vehicles to operate by themselves, even in circumstances engineers can’t plan for.
The idea would also appear to contradict Google X’s vision for the future. Chris Urmson, leader of the company’s self-driving car project, has said repeatedly in public appearances that he wants to build a car that isn’t even capable of handing control over to a human.
That’s because he doesn’t think such a paradigm would be safe. If a human gets into a car with the expectation that they won’t have to drive, they might not be ready to take over if something goes wrong.
Yet questions remain about the ability of cars to handle lots of different scenarios. A big one is inclement weather — most of the testing of AVs has happened in places like Texas and California, where there is very little snow and relatively little rain. Google has expanded its self-driving car testing to Washington so that it can start operating in more intense weather conditions.
Regardless, leaders in the AV field still envision a lot of utility coming from cars that don’t quite reach level five. Several cities participating in the U.S. Department of Transportation’s Smart City Challenge have proposed running driverless shuttles on fixed routes at places such as airports, universities and downtown corridors. Others have spoken about the ability of self-parking cars to either shrink the number of parking lots or fit more cars into existing lots.
And there are plenty of safety applications that don’t meet the level five definition either. That includes technology already available in cars on the road today such as automatic braking and collision warnings.