It’s been a long day. As you ride home from the office, you start to nod off. You close your eyes as the self-driving car merges onto the highway. When you’re zonked out 15 minutes later, the car changes your route because of traffic, and eventually you wake up at your destination.
That’s the dream of autonomous cars — and some very smart people, including Google’s Sergey Brin, thought they’d already be driving people around public streets by now. That hasn’t happened. In fact, in 2018, progress stalled, sometimes with deadly consequences.
In March, a woman was struck and killed by one of Uber’s driverless cars in Tempe, Arizona. The same month, a Tesla SUV in Autopilot mode erupted into flames as it crashed into a barrier on Highway 101 in California, killing the driver. And earlier this month, on the same freeway, police arrested a suspected drunk driver who fell asleep at the wheel of his Tesla, which was also in Autopilot. It took seven minutes for the cops to get the man off the road.
There have been self-driving success stories, too. Autonomous shuttles took people to work in Detroit. Driverless delivery services brought fresh produce to the front doors of Silicon Valley techies. And Jaguar Land Rover developed shuttles that encourage robot-human communication with “virtual eyes.”
Still, the tech picture for self-driving cars at the end of 2018 doesn’t look much different than it did at the beginning. It’s mostly the same equipment, software, regulations, and companies that were here a year ago.
So is the self-driving dream any closer, or have the accidents and fatalities pushed it farther away?
Hitting the brakes
Just this month, Waymo launched its self-driving taxi service. It wasn’t the splashy, the-future-has-arrived moment promised. Instead, it was more of a modified pilot program for a select group of about 400 people. Waymo One, as the service is now called, isn’t the only example of the slow progress of autonomous vehicles.
After the fatality in March, Uber pulled its test vehicles off the road. Only recently did Uber broach the possibility of returning its self-driving cars to the streets in a scaled down test in Pittsburgh, which would involve slow speeds on a 1-mile loop in good weather.
“While we are working to get back on public roads, we would never compromise on safety in order to get there,” an Uber spokesperson said.
That’s probably a good philosophy, seeing as the public doesn’t exactly trust the technology yet. An Intel study from earlier this year asked 1,000 Americans if they trusted autonomous vehicles. Only one-fifth said they’d own a self-driving car, and 35 percent of respondents said they feel unsafe around the vehicles. In a soon-to-publish Deloitte survey of more than 1,600 Americans, 50 percent of respondents said they don’t believe autonomous vehicles will be safe — up from 47 percent last year.
Other companies are also scaling back their lofty goals. GM’s autonomous vehicle department, Cruise, was supposed to launch a self-driving car service in 2019. But in recent comments, GM CEO Mary Barra hinted that we’ll see something akin to the Waymo “launch”: a limited service contained to small areas.
Moving forward… carefully
While “previous years have been about hypothesizing, building, and testing,” said Tactile Mobility CEO Amit Nisenbaum, the next year will be about “industrialization of the offering,” meaning putting the technology in more cars and roads.
That means incremental changes in products that already exist. Vincent Roche, CEO of Analog Devices, a company that supplies autonomous vehicle equipment, anticipates better Level 3 autonomy (that’s a mix of human and machine driving) on systems like Tesla’s Autopilot, ongoing improvements to the RADAR and LiDAR equipment that lets cars “see” and make better decisions, and more testing and expanded commercialization from companies such as Waymo and Baidu.
“You have to ensure the whole safety system is there and you have to pace yourself,” said Hu Wen, COO of Chinese self-driving company Pony.ai, last month. He’s in no rush, limiting testing to a 1.7-mile fixed route that opened to the public earlier this year.
Even if the technology advances, Scott McLaren, CMO at auto insurance company Fortegra, says there’s another roadblock that could bring it all to a halt: government legislation.
The U.S. Senate has yet to pass a bill, the AV Start Act, that would loosen some federal restrictions, while requiring more crash data from automakers.
Just because highways full of self-driving cars aren’t coming in 2019 doesn’t mean they won’t be here eventually. As a panel of technologists and futurists predicted in a recent campaign, no one will own a car in 25 years and the road system will all be part of an autonomous and controlled traffic system. A report out this week forecasts nearly 200,000 autonomous vehicles on U.S. roads by 2021. The robot cars are coming.
(With that will come new problems. David Barzilai, co-founder of Karamba Security, predicts more state-backed attacks on cars, similar to hacks already seen on connected Tesla vehicles.)
Perhaps the best prediction for 2019 comes from Kyle Lui, principal at venture capital firm DCM, which backs companies like Lime and Grin Scooters. He says e-scooters will beat out autonomous cars with auto-assist features, making the two-wheeled electric vehicles safer and easier to use.
So, it looks like napping in the backseat of your self-driving car isn’t coming soon. Instead, we get … scooters.