Self-driving cars should be capable of cutting the annual number of traffic deaths in half by the time they’re ready for widespread use on American roads, a top federal official says.
In what may be a preview of the revised guidance on autonomous-vehicle development that government officials are scheduled to unveil next month, Mark Rosekind, administrator of the National Highway Traffic Safety Administration, said he wants concrete safety for the technology. “I’d actually like to throw the gauntlet down,” he said, speaking on a panel during the TU-Automotive conference in Detroit. “Start with two times better. We need to set a higher bar if we expect safety to actually be a benefit here as opposed to just an equivalency. While no one wants to say, ‘How good is good enough?,’ I’d actually say, ‘Start at two times and then let’s work from there.’ ”
Industry executives and government officials alike often talk about the promise of reduced traffic deaths in an autonomous era, but rarely put specifics behind those predictions. Rosekind’s comments, while as much rallying cry as projection, offer the most precision to date on how those expectations might take shape.
Previously, there have only been general answers from both public and private leaders on what sort of safety improvement could be achieved and should be expected. On one end, Google executives have stated their goal is to have autonomous vehicles perform better than the human drivers who are responsible for 94 percent of collisions. On the other, Rosekind’s predecessor, former NHTSA chief David Strickland, says self-driving cars will need to be “virtually perfect” to earn acceptance from consumers. Federal motor-vehicle safety standards are ambiguous. They say any new car, autonomous or not, cannot present an “unreasonable risk” to safety.
Feds Place Bets In Autonomous Tech
Whatever the role of self-driving cars, there’s renewed urgency in addressing road fatalities. Deaths on American roads reached 32,675 in 2014, the latest year for which data is available, and they’re increasing at a pace not seen since the 1940s. Earlier this year, the Department of Transportation said early estimates showed a 9.3-percent year-over-year increase in 2015 and the National Safety Council, which tracks both traffic and non-traffic motor fatalities, warned the death toll could rise as much as 14 percent. Either would mark the largest single-year percentage increase since 1947.
“
“At some point, it might be so large that there has to be a new organization or a spin-off or something.” – Mark Rosekind
”
Autonomous cars, of course, are not the only potential salvation; decidedly low-technology ways that would make drastic strides in saving lives already exist. Eighty-seven percent of vehicle occupants use their seatbelts, according to NHTSA, but the 13 percent who don’t accounted for 49 percent of all occupant fatalities in 2014. Drunk drivers cause approximately 31 percent of traffic deaths.
Without possessing the power to thwart poor human decisions, federal officials have placed their hope in autonomous technology. In January, as Transportation Secretary Anthony Foxx unveiled a budget proposal that would invest $3.9 billion into autonomous technology over 10 years, he said, “I’ll put this in plain English for you: In 2016, we are going to do everything we can to advance safe, smart, sustainable transportation innovations.” That push includes accelerated guidance on self-driving operations.
Manufacturers have clamored for that guidance, without which they’ve said it would be difficult to deploy amid an uncertain and potentially conflicting regulatory landscape. Updated policies are due in July, and for manufacturers, they’re expected to include operational guidance and interpretations of existing federal motor-vehicle safety standards. For state transportation officials and others bracing for autonomous cars, they will likely include model state policies and identify new authorities that may be needed for regulators.
Could that include a whole new federal agency that goes beyond the scope of NHTSA to regulate autonomous vehicles? Rosekind doesn’t think so, but he didn’t dismiss the idea either.
“Not yet,” he said. “What we’re going to see in July is the first part of what needs to be evolved, and I think we’ll be pointing to a lot of others that need to help us figure this out. At some point, it might be so large that there has to be a new organization or a spin-off or something. But right now at least, we have a vision that can get us to where we need to be safety-wise.”
Borrowing a Good Idea from Aviation
The need for a new agency is, at best, unclear. But what federal authorities and others agree on is that there’s a need for new ways to evaluate and measure the safety and effectiveness of autonomous cars.
On a recent trip to Mountain View, California, where several NHTSA officials and engineers had come to view Google’s latest self-driving developments, one test drove home that point. As they took a test drive near several parked vehicles, one NHTSA employee along for the ride opened a door, causing the Google vehicle to lurch to a halt. Had the incident occurred on public roads and not in a closed test environment, current rules would mandate that Google report the hard-braking stop to the California Department of Motor Vehicles.
“It would have gotten labeled as hard braking and used against them as ‘Oops, hard braking,’ as opposed to ‘crash avoided,’ ” said Rosekind, who was on the trip. “So we need new safety metrics.”
An ever-growing number of sources could provide data for those new metrics. Engineers now collect information from computer simulations, lab work, closed-course testing, and real-world test results. One additional data set under consideration is an anonymous data sharing network that could be set up in similar fashion to the federal Aviation Safety Reporting System, which permits pilots, air-traffic controllers, and others to provide confidential accounts of systemic problems and near-accidents, in the hopes deficiencies can be corrected and future accidents averted.
Rosekind is a proponent of such a network. He believes it could hasten the arrival of safer autonomous vehicles. In that sense, the reporting network could contribute to a system where his most important metric, traffic deaths, isn’t just a number, but also indicative of a successful end result.
from Car and Driver Blog http://ift.tt/28KbBqa
via IFTTT
0 comments:
Post a Comment