Computers are slowly taking over one of the most potent symbols of human independence: Driving.
The government says that shift will make the roads safer and eventually free people to work, read or even watch a movie as they travel from place to place.
But in the wake of deadly manufacturer defects at Toyota and General Motors, analysts are raising questions about whether autonomous vehicles could hurtle into dangerous territory.
Last month, the National Highway Transportation Safety Administration took the first step toward requiring that new automobiles be equipped with sophisticated computers designed to communicate with other vehicles, with the aim of preventing accidents and, eventually, guiding cars through traffic.
Those systems are being field-tested in Michigan. Some other states permit fully autonomous cars to be driven on their roads — in California, Google’s fleet has logged more than a half-million miles without incident. Before the decade is out, Volvo and Nissan say, self-driving cars will be available to the average consumer in dealer showrooms.
“Decades from now, it’s likely we’ll look back at this time period as one in which the historical arc of transportation safety considerably changed for the better,” said David Friedman, acting NHTSA administrator.
But while autonomous cars herald great promise, they also pose difficult policy questions. Who is liable, the driver or the manufacturer, if autonomous vehicles wreck? Who owns the trove of data the cars generate? Should a computer steer a car off the road if a tree limb falls in front of it? What about a child on a bicycle?
And the prospect of manufacturer defects would become even more alarming in vehicles that rely heavily on increasingly complex computer technology.
“Cars are meant to be driven by people, not machines,” said Joan Claybrook, a consumer advocate and former NHTSA administrator.
“I have enough trouble trusting my computer, much less a computer to drive my car.”
Last week, the Justice Department fined Toyota $1.2 billion for covering up a sudden-acceleration problem that has spawned more than 400 lawsuits.
Meanwhile, GM is facing multiple federal investigations into an ignition-switch problem that has contributed to at least 31 accidents and 12 deaths since surfacing more than a decade ago.
In both cases, the problems turned out to be basic and mechanical.
But federal safety investigations into the defects were complicated by the millions of lines of computer code and advanced electronics that are already standard equipment.
“The reality is that the vast, vast majority of accidents are caused by human error and computers are going to dramatically improve on people’s driving,” said Joshua Schank, president and chief executive of the Eno Center for Transportation, a research organization that has studied the challenges posed by autonomous cars.
But “people are very nervous about the idea that computers could go haywire and cause us to die.”
“Cars are meant to be driven by people, not machines,” said Joan Claybrook, a consumer advocate and former NHTSA administrator.
“I have enough trouble trusting my computer, much less a computer to drive my car.”
Last week, the Justice Department fined Toyota $1.2 billion for covering up a sudden-acceleration problem that has spawned more than 400 lawsuits.
Meanwhile, GM is facing multiple federal investigations into an ignition-switch problem that has contributed to at least 31 accidents and 12 deaths since surfacing more than a decade ago.
In both cases, the problems turned out to be basic and mechanical.
But federal safety investigations into the defects were complicated by the millions of lines of computer code and advanced electronics that are already standard equipment.
“The reality is that the vast, vast majority of accidents are caused by human error and computers are going to dramatically improve on people’s driving,” said Joshua Schank, president and chief executive of the Eno Center for Transportation, a research organization that has studied the challenges posed by autonomous cars.
But “people are very nervous about the idea that computers could go haywire and cause us to die.”
Researchers such as Schank tend to focus on the positive. In the foreseeable future, they say, self-driving technology will save fuel, cut pollution and reduce highway costs. As more cars become autonomous, they could safely tailgate, packing more vehicles into existing lanes and making traffic jams a thing of the past. Autonomous cars even have the potential to offer the disabled the freedom of the road.
The pace of change has been rapid. Just 10 years ago, a self-driving-car competition held by the Defense Department’s Defense Advanced Research Projects Agency failed to produce a winner. Not a single entry was able to complete a 142-mile desert course between California and Nevada and claim the $1 million prize.
Fast-forward to 2007, when a team of researchers from Pittsburgh’s Carnegie Mellon University tricked out a Chevrolet Tahoe with conspicuous sensors outside and advanced electronics inside. The SUV was able to follow traffic laws, merge into moving traffic, make its way through traffic circles and avoid other obstacles at a decommissioned Air Force base, proving that autonomous cars were more than a pipe dream.
By 2009, Google was test-driving autonomous cars on busy highways. Last year, speaking at a ceremony after the signing of a bill allowing the testing of autonomous cars on California highways, Google co-founder Sergey Brin said, “You can count on one hand the number of years until ordinary people can experience this.”
Auto manufacturers have also stepped up the pace of research. Some have established outposts in Silicon Valley intended to foster collaboration with the region’s deep well of technology entrepreneurs and quicken the pace at which the automobile is re-imagined.
“The metabolic rate of what happens here is dramatically different than what happens everywhere else,” said Venkatesh Prasad, a senior technical leader and self-described “what’s-next guy” with the Ford Motor Co.’s Research and Innovation group in Palo Alto, Calif.
Other work is being done in Michigan, the ancestral home of automobile development. In Ann Arbor, University of Michigan researchers backed by a federal highway safety grant have outfitted 3,000 cars with special sensors and wireless devices that allow them to exchange information with one another and with nodes mounted on traffic lights, at intersections and along curves on more than 70 miles of city streets.
Ten times a second the cars and roadways “talk” to one another, relaying vehicles’ location, speed and direction, and alerting drivers if their cars are going too quickly around a curve or if another car is erratically changing lanes or braking.
Researchers are combing through billions of messages passed through the network with an eye toward creating a robo-road system that would guide driverless cars from the garage to the grocery store and beyond.
To complement that research, the University of Michigan is constructing a 30-acre facility that would serve as a kind of test track for self-driving cars. The test area would allow engineers to see how autonomous cars perform in a complicated urban environment that includes street signs, stoplights – even construction detours.
For consumers, researchers say, the automobile’s transformation is likely to unfold in stages. Within the next decade, cars will be able to drive themselves, but people will still have the option to take over.
After that? Some engineers envision vehicles that are little more than the private equivalent of railroad passenger cars, with no controls for humans at all.
