A self-driving Uber car killed a pedestrian. Human drivers will kill 16 people today.

Authored by vox.com and submitted by mvea

A woman in Tempe, Arizona, has died after being struck by a self-driving car on Sunday in what’s believed to be the first pedestrian fatality caused by an autonomous vehicle on a public road.

The vehicle, operated by Uber, was in self-driving mode, though the car had a safety driver — who in theory could take control of the car in the event of an accident — behind the wheel, according to the Tempe Police Department. The woman, 49-year-old Elaine Herzberg, was crossing the street outside of a crosswalk around 10 pm when she was hit.

Self-driving cars are widely expected to be safer than cars driven by humans, especially as crash avoidance features become more advanced. But it’ll take time to know this for sure, and to develop the regulations that minimize potential harms.

In the meantime, this tragedy will no doubt stoke fears about putting our lives in the hands of car computers. So let’s not forget that getting into, or walking or biking near, a car piloted by a human is one of the most dangerous activities we can do every day.

According to a report released by the Governors Highway Safety Association, there were 5,997 pedestrian fatalities in the US in 2016 — or 16 each day. Globally, road traffic is the fifth leading cause of death in the world.

What’s more, driving a car and being a pedestrian are particularly dangerous undertakings in America, relative to other high-income countries. The US ranks 41st out of 52 high-income countries on road traffic deaths per population, with only a handful of countries — including Qatar, Kuwait, Oman, and Saudi Arabia — trailing behind.

In a recent blog post at JAMA, health policy professor Lawrence Gostin pointed out that the US has been lagging its peers when it comes to implementing evidence-based policies that reduce the risk of traffic deaths. Here’s Gostin:

If the US had kept in step with other countries, David Leonhardt calculated at the New York Times, “about 10,000 fewer Americans each year — or almost 30 every day — would be killed.”

The government could do many things to prevent these accidents — long before self-driving cars become the norm. The US might consider simply lowering speed limits more often and making speed cameras more ubiquitous, Leonhardt suggested. That’s something officials in many other high-income countries have already done.

But it doesn’t have to stop there. According to Laura Bliss at CityLab, “Roughly half of all vehicle occupants killed in crashes over the past five years weren’t wearing seat belts. Yet 16 states still lack primary enforcement seat belt laws for everyone on board.” Some 40 percent of motorcyclists killed on the road were helmet-free, but “more states are repealing helmet requirements than tightening them,” Bliss writes.

So before we freak out about this tragic self-driving car death, we need to keep sight of the bigger picture: At least right now, the non-autonomous vehicles we don’t give a second thought to are the much, much greater public health threat. Regulators should focus on them. And as we emerge from the “Wild West phase of autonomous vehicles,” on understanding whether these computer-driven cars harm fewer people than their human-driven counterparts.

Our hearts go out to the victim’s family. We’re fully cooperating with @TempePolice and local authorities as they investigate this incident. — Uber Comms (@Uber_Comms) March 19, 2018

For now, Uber is investigating the incident and has already suspended its autonomous vehicle operations in Pittsburgh, Tempe, San Francisco, and Toronto. “Our hearts go out to the victim’s family,” the company said in a statement.

Kost_Gefernon on March 20th, 2018 at 05:54 UTC »

Now that it’s had its taste of human blood, there’s no going back.

Edit : removed the extra possessive.

NathanaelGreene1786 on March 20th, 2018 at 03:55 UTC »

Yes but what is the per capita killing rate of self driving cars vs. Human drivers? It matters how many self driving cars are in circulation compared to how many human drivers there are.

jeremyc74 on March 20th, 2018 at 01:31 UTC »

The latest story I read reported the woman was walking a bike across the street when she was hit, and it didn't appear the car tried to stop at all. If that's the case (and it's still early so it may not be) that would suggest that either all the sensors missed her, or that the software failed to react. I'm an industrial controls engineer, and I do a lot of work with control systems that have the potential to seriously injure or kill people (think big robots near operators without physical barriers in between), and there's a ton of redundancy involved, and everything has to agree that conditions are right before movement is allowed. If there's a sensor, it has to be redundant. If there's a processor running code, there has to be two of them and they have to match. Basically there can't be a single point of failure that could put people in danger. From what I've seen so far the self driving cars aren't following this same philosophy, and I've always said it would cause problems. We don't need to hold them to the same standards as aircraft (because they'd never be cost effective) but it's not unreasonable to hold them to the same standards we hold industrial equipment.