Every Day Matters with Driverless Cars

Imagine there’s a horrific virus that kills over 35,000 people a year in the United States—over 90 people a day. Miraculously, a new cure is developed that can save nearly everyone who has this disease. We would want to make sure that the cure works properly, yes, but we would also want to balance that concern with the cost of waiting, with the tremendous loss of human life happening daily. If this miracle cure existed, we’d want to get it out into the hands of the public as soon as possible. Every single day would matter and save lives.

That’s the situation we face today with driverless cars.

Let’s face it: Humans are really, really bad at driving. Of the 35,000 lives lost every year from car fatalities, 94 percent are attributable to human error. Thousands more are injured each day, and car crashes have many other significant costs for the public.

Even more horrific, we appear to be getting worse at driving during an age of evermore distracting digital devices. The National Highway Traffic Safety Administration (NHTSA), the federal agency that oversees auto safety, recently reported “an alarming uptick in fatalities” with a 10.4 percent increase in the number of vehicle deaths in the first half of 2016 over the previous year.

This is why driverless car technology is so exciting. Autonomous systems will take the wheel and make us much safer in the process. Put simply, robots don’t get distracted, drowsy, or drunk. They don’t have blind spots, and they won’t text. These facts alone will result in a precipitous drop in the number of auto crashes and corresponding injuries or deaths.

While they’ll never be perfect, autonomous systems can be programmed to make quicker and better decisions than most humans in life-and-death situations. The algorithmic programs that power these systems will be constantly refined to provide real-time improvements in how to deal with those situations. While fully autonomous cars may be many years away, even semi-autonomous technologies can augment human driving and help us radically improve road safety.

Every policy consideration must be judged upon how rapidly it can help us move us away from the current baseline of crashes and fatalities and closer to NHTSA’s stated goal of “zero fatalities on our roads.” That’s what makes the agency’s recently announced Federal Automated Vehicles Policy so important.

On one hand, NHTSA should be applauded for embracing the revolutionary potential of driverless cars. They have articulated well the dangers of human drivers, along with the way in which clear guidelines can encourage and spread the use of these technologies. In a positive step, the guidance aims to overcome potentially conflicting state regulations of driverless cars and clarifies how existing federal auto safety rules apply.

Unfortunately, other portions of NHTSA’s proposed guidance could undermine its ambitious safety goals. The most problematic is the agency’s suggestion that some sort of “pre-market approval authority” might be needed for these technologies. This would require any new driverless car design to be approved by NHTSA before reaching market and could even be extended to pre-approval for new software updates on existing models. NHTSA admits that this “would be a wholesale structural change in the way NHTSA regulates motor vehicle safety.”

The main problem with such a pre-market approval regime is that it significantly increases the time and cost of deploying driverless cars onto the road.

NHTSA uses the Federal Aviation Administration (FAA) pre-market regulatory powers as an example of the type of process they would like to emulate. Yet NHTSA admits that FAA certification often lasts three to five years and that the most recent FAA certification process for the Boeing 787 Dreamliner consumed an estimated 200,000 hours of FAA staff time and lasted eight years.

That fact alone should foreclose further discussion about the wisdom of NHTSA employing an FAA-like pre-market approval regime for driverless cars. Over a quarter of a million people will have died in accidents during that eight-year approval process. Even the average duration of FAA pre-market certifications (three to five years) is far too long if we hope to get life-saving driverless car technologies deployed on a timely basis.

Moreover, even if NHTSA sticks to applying its existing regulations, it could restrain beneficial change. The agency’s regulatory standards, which originated in the 1960s, are centered mostly on human drivers and technologies, like the steering wheels and pedals used to operate vehicles. But robot cars don’t need those things and mandating compliance with Analog Era regulations will only delay the rollout of better technologies.

For these reasons, NHTSA should abandon the idea of pre-market approval as a dangerous restraint on driverless car innovation; a restraint that would put lives at risk. Instead, the agency should focus on encouraging industry best practices and post-market oversight.

It’s also important to remember that driverless car manufacturers have powerful reputational and legal incentives to get things right. Companies like Google, Ford, Uber, and Tesla have been exceedingly careful in the development and deployment of these technologies and rightfully so. Every mistake in this process has been, and will continue to be, loudly broadcast by the media. To sell skeptical consumers on letting the robots drive, these companies will have to go above and beyond in building a reputation as a reliable and safe form of transportation.

Recognizing that these incentives are aligned with safety should give us ease about letting private companies take the lead within a light-touch regulatory framework.

We share NHTSA’s goal of a future without traffic fatalities, but what is just as important is the speed at which we reach it. With the pre-market regulatory authority requested by NHTSA, driverless cars may hit a road full of speed bumps, and the consequences are deadly. Every day matters.