For years, auto manufacturers and software makers have touted self-driving cars as the next big trend to hit the American road.
Then, in quick succession last month, those bright predictions were shaken by crashes involving partly self-driving cars that killed a pedestrian in Arizona and a driver in California.
On March 18, an Uber automated vehicle with a human backup behind the wheel struck a 49-year-old mother in Tempe, Arizona, killing her.
Five days later, a Tesla Model X SUV with a semi-autonomous driving system struck a concrete highway lane divider and burst into flames, killing the 38-year-old software engineer behind the wheel. Tesla confirmed that its autopilot system was turned on at the time of the crash, according to reports.
News of the fatalities reverberated through the world of carmakers and regulators, resulting in autonomous car testing being shut down temporarily in several cities, including Boston.
But despite such setbacks, experts say they see no pause in the march toward automating cars and trucks, and the eventual appearance of totally self-driving vehicles on the highway.
In fact, with the development of such features as automatic braking and self-parking, progress toward autonomous vehicles is already well down the road.
“Autonomous vehicles aren’t going to be here tomorrow,” said Josh Ostroff of the think tank Transportation for Massachusetts. “They’re here today.”
Self-driving cars are being tested widely across the nation, including the Seaport District of Boston. Many industry observers predict such cars will be in the hands of consumers by 2025.
In the meantime, carmakers such as General Motors, with its Cadillac brand, and Mercedes-Benz are already marketing vehicles with limited hands-off driving features.
As early as 2014, Mercedes’ S class, using its combination of cameras, sensors, and radar, was able to accelerate and decelerate along with traffic and detect traffic at an intersection and slow down in response. It could also apply brakes automatically if it sensed an object in its path.
But some industry experts warn against over-promising on driverless technology and caution that truly driverless cars — requiring zero human interaction and capable of operating on any kind road and in any traffic condition — may not be perfected for several years.
Those cautionary notes seem particularly prescient in light of the recent fatalities.
Video of the Tempe accident is particularly disturbing. The car is shown zipping along the highway and not slowing down before striking the woman as she walked her bicycle across two lanes.
The video does not show the aftermath of the accident. An interior view of the car, caught by a video camera, shows the human occupant behind the wheel apparently not looking at the road prior to impact.
Uber reportedly has reached a settlement with the victim’s family.
Negative publicity surrounding such incidents may make headlines but are unlikely to halt the development of self-driving vehicles, say experts like Ostroff.
“The incident in Arizona was rightly seen as tragic,” he said. “But at the same time, tens of thousands of deaths are caused each year by vehicles that had human drivers at the wheel.”
Many futurists expect self-driving cars will actually reduce the number of crashes and possibly even do away with that other driving bugaboo, traffic tickets. After all, robots don’t text or drive drunk.
“Vehicle safety and self-driving technology are really two sides of the same coin,” said Bharat Balasubramanian, an engineering professor at the University of Alabama and former Daimler official who played a major role in developing driver-assisted technology for that company’s vehicles.
Not only should self-driving cars make the roads safer, Balasubramanian said, they could eventually reduce crashes by up to 90 percent — saving an estimated $900 billion a year in the bargain.
Just the same, many consumers remain skeptical. A recent Reuters poll found that 55 percent of respondents said they wouldn’t be comfortable riding in a fully autonomous car.
Many industry experts expect such objections to melt away as vehicles are perfected, just as turn-of-the-century horse-and-carriage owners eventually came to accept the hated automobile.
“If you look at the history of automobiles in this country, human error is still the cause for a vast majority of accidents,” said Marc Lamber at lambergoodnow.com, an Arizona attorney who concentrates on technology issues. “Innovations in safety such as seat belts, anti-lock brakes and airbags were all implemented by manufacturers to help keep people safer, and we’re now witnessing the next steps in this evolution.”
But while technological advances may eventually make cars and roads safer, it’s unclear how they will impact traffic congestion and commuting habits.
Even with automated vehicles, many of us will still need buses, subways and commuter trains to get to work, Ostroff said.
“Some people think that autonomous cars will make trains and buses obsolete,” he said. “That’s fantasy. Unless, of course, you think of self-driving cars as trains and buses.”
Autonomous cars also raise a host of legal and ethical questions, many of which have yet to be tackled by lawmakers.
For instance, where and under what conditions is a driverless car allowed on the road? If a self-driving car gets in an accident, who’s liable? How does an insurance company rate a robot that has no driving record?
Massachusetts legislators have already filed a passel of bills to deal with autonomous vehicles. At least 13 other states are regulating the technology. In Florida, for instance, self-driving cars are allowed to operate without a driver. And Michigan allows such cars to be used for ride-hailing services.
Issues discussed at a Massachusetts legislative hearing last year ranged from how much regulation of driverless cars is desirable (auto companies want as little as possible) to how will autonomous vehicles will impact traffic volume.
Even more intriguing issues have been raised by those outside government and industry circles. For example, how does one regulate potential hazards that didn’t exist in an era when vehicles had to be driven and parked by a human?
“For instance,” Ostroff said, “I don’t think you want a self-driving car circling around and around while its owner sits in their office. That’s chaos.”
Such so-called “zombie vehicles” could present giant traffic as well as safety headaches, some suggest.
“We have a serious concern that a large number of zombie vehicles could create more gridlock,” said Chris Dempsey, director of T4Mass, an organization cautions about a world where empty, self-driving vehicles clog the roads as they rush to pick up and drop off passengers.
As if matters of liability and traffic regulations aren’t enough, there’s also the existential question hanging over the self-driving car issue: How, exactly, do you define an autonomous vehicle?
The question is more subtle than one might imagine. While many of us think of a self-driving car in George Jetson terms with little for the driver to do but nap, we’re not nearly there yet.
The Society of Automotive Engineers, the umbrella group for auto designers, actually divides “driving automation” into five progressive levels, from zero for no automated features to five for fully automated driving.
Most cars currently being advertised with features like automatic lane centering, braking or parking could be classified as level 2, with a human still doing much of the driving.
Automated driving with the vehicle doing most of the work would be level 3 and would have a human driver ready to take over if something unexpected happens.
We’re still working on that one.
What’s the timeline?
So when should we expect to see fully self-driving cars on the roads? It depends on who you ask.
Some companies have promised that their self-driving cars will be available as early as 2021. Others say the introduction will be a gradual process, beginning with partially autonomous vehicles and progressing year by year.
One person with reason to know is Lior Ron, cofounder of Otto, the driverless truck company that was purchased by Uber.
Speaking at an MIT Technology Review conference last year, Ron said it will take a while before truly self-driving vehicles hit the road.
“It’s not going to happen overnight,” he was quoted as saying. “It’s going to happen in very discrete baby steps.”
Long before the first consumer-owned vehicle begins tooling down the road on its own, there are still plenty of technical and engineering questions to be answered, Balasubramanian said.
Autonomous cars use computer technology and a variety of cameras and sensors to guide them and help them avoid obstacles as they drive along routes pre-mapped by Google and others.
But what happens when road conditions unexpectedly change, such as when an accident shuts down a section of road or when a vehicle happens upon an unanticipated detour?
“Connectivity with the outside world is critical,” he said. “We need to think about how 3D mapping information will be updated in real time and how will the vehicle be able to alert the driver with sufficient lead time to take over the controls.”
Despite such obstacles, driverless cars and trucks in forms that essentially require little or no human assistance could be practical relatively soon in special circumstances, Balasubramanian said.
An example would be a dedicated lane where the vehicles would not be subjected to the vagaries of reckless drivers or other unpredictable hazards.
While perfecting autonomous vehicles poses many challenges, it also offers opportunities.
Auto and technology companies are already investing billions hoping to cash in on the next technological wave in driving. And the impact stretches far beyond Detroit, Tokyo and Beijing.
In Attleboro, technology company Sensata has placed a big bet on self-driving technology in cooperation with Quanergy, a California driving automation pioneer. Sensata was one of several companies including Daimler, Delphi and Samsung that invested a total of $150 million in the company’s technology, according to an article in Forbes magazine.
Quanergy is working to perfect and mass produce its LiDAR technology — a combination of lasers and radar that provides critical sensing data needed by self-driving systems.
A major challenge in commercializing autonomous cars has been bringing down the cost of LiDAR, which by itself can approach the price of a luxury sedan. Quanergy has the potential to become a major player with its solid state LiDAR, which Forbes said is expected to sell for around a more affordable $250.
Technical challenges haven’t sapped the auto industry’s enthusiasm for self-driving cars. A 2017 report by the Brookings Institution estimated total investment by companies in the technology in the U.S. topped $80 billion over a three-year period.
While automated vehicles could be a marketing and sales bonanza, self-driving cars might also have social costs by eliminating jobs once done exclusively by humans. By one estimate, robot taxis and other conveyances could put as many as 3 million humans out of work.
But long before cab, Uber and Lyft drivers start worrying about their jobs, self-driving cars are going to have to negotiate winding turnpikes of regulations and roads riddled with technological and legislative potholes.
And that just might take a while.