THE promise of self-driving cars can be alluring — imagine taking a nap or watching a movie in a comfortable armchair while being shuttled safely home after a long day at work. But, like many optimistic images of the future, it is also a bit of an illusion.
Automated cars may indeed make commuting more pleasurable while preventing accidents and saving tens of thousands of lives — someday. But, a recent fatal crash in Tempe, Arizona, involving a car operated by Uber that was tricked out with sensors and software meant to turn it into a latter-day version of KITT from the TV show Knight Rider suggests that at least some of these cars are not ready for the hustle and bustle of American roads. In fact, the technology that powers these vehicles could introduce new risks that few people appreciate or understand. For example, when a computer controlling the car does not hit the brakes to avoid a collision, the person in the driver’s seat — many automated cars on the road today still require someone to be there in case of an emergency — may also fail to intervene because the driver trusts the car too much to pay close attention to the road. According to a video released by Tempe police, that is what appears to have happened in the Uber crash.
“Technology does not eliminate error, but it changes the nature of errors that are made, and it introduces new kinds of errors,” said Chesley Sullenberger, the former US Airways pilot who landed a plane in the Hudson River in 2009 after its engines were struck by birds and who now sits on a Department of Transportation advisory committee on automation.
Sullenberger is hardly a technophobe. He has flown passenger jets crammed with advanced electronics and software and has a keen professional interest in technology. What concerns him and other safety experts is that industry executives and government officials are rushing headlong to put self-driving cars on the road without appropriate safeguards and under the unproven hypothesis that the technology will reduce crashes and fatalities.
Even as officials place a big bet that autonomous cars will solve many of our safety problems, American roads are becoming less safe. More than 37,000 people were killed on American roads in 2016, up 5.6 per cent from 2015, according to government data.
Experts who are sceptical about the unceasing forward march of technology say fatalities are rising because public officials have become so enamoured with the shiny new thing, self-driving cars, that they have taken their eyes off problems they could be solving today.
Consider automatic braking systems. The Insurance Institute for Highway Safety estimates that there is a 42 per cent reduction in rear-end crashes that cause injuries when this technology is installed on cars. Advocates for Highway and Auto Safety and other public interest groups asked the Transportation Department in 2015 to require that all new trucks, buses and other commercial vehicles have such systems, which have been around for years. The department accepted that petition but has yet to propose a rule.
Even as US regulators have dragged their feet, other industrialised countries have made great strides in reducing traffic crashes over the last two decades. Road fatality rates in Canada, France, Germany and Sweden, for example, are now less than half the rate in the United States. They have reduced accidents the old-fashioned way. Some of them have worked to slow down traffic — speed is a leading killer. They have added medians and made other changes to roads to better protect pedestrians. And, European regulators have encouraged the use of seat belts by putting visual reminders even in the back seat.
“The things that have been killing us for decades are still killing us: speed, impaired driving, not using seat belts,” said Deborah Hersman, the former chairman of the National Transportation Safety Board who now heads the National Safety Council. “The things that we know can save lives, some of them don’t cost any money, like seat belts.”
Silicon Valley technologists would argue that algorithms and machine learning will simply leapfrog what they might dismiss as the legacy problem of human fallibility. But Sullenberger, for one, is worried that the rush to develop automated cars will lead to many unforeseen problems.
“Even though there is a sense of urgency to prevent human-caused accidents, we need to do it in a responsible way, not the fastest way.”--NYT