| 1:22 PM EST

Accidents Will Happen

 In the development of self-driving vehicles, it is going to be necessary to have them operate in the real-world, not just simulated on screens or driven on test facilities where there is almost-absolute environmental control. 


Facebook Share Icon LinkedIn Share Icon Twitter Share Icon Share by EMail icon Print Icon
At the risk of seeming insensitive or flippant—neither of which I am trying to be—the Uber self-driving vehicle-caused pedestrian fatality in Arizona in March should have been completely expected.
No, I am not talking about that particular incident. But in the development of self-driving vehicles, it is going to be necessary to have them operate in the real-world, not just simulated on screens or driven on test facilities where there is almost-absolute environmental control. And accidents will happen.
The word accident goes back to the Latin accidere, which is a combination of ad, which means toward or to, and cadere, to fall. No one really expects to fall. And if we go forward in time to now, the word can be defined as something that occurs unexpectedly and unintentionally. 
The National Highway Traffic Safety Administration (NHTSA) collects data, including the statistics related to traffic-related deaths. Deaths by accident. Even though the moniker for the compilation is a breezy “Quick Facts,” the latest data is absolutely sobering.
The number of fatal crashes in the U.S. in 2016 was 34,439. That’s up from 32,539 in 2015 and 30,056 in 2014. And if those numbers aren’t large enough to make your eyes widen and your jaw drop, the number of people who died in those crashes were, for those three years respectively, 37,461, 35,485 and 32,744. Accidents all. Another way “Quick Facts” breaks down those numbers: there were 102 fatalities per day in 2016, 97 in 2015 and 90 in 2014.
Drilling down, there were 16 pedestrian fatalities per day in 2016, 15 in 2015 and 13 in 2014. And I’ll cite just another set of numbers for additional calibration. The number of “Alcohol-Impaired Driving Fatalities per Day”: 2016: 29; 2015: 28; 2014: 27.
All of which is to say that there is a massive number of people who are dying on America’s roads every day. People die who have been struck by a motor vehicle as they are walking along. People die because some driver has had too much to drink. People die because, well, accidents happen, whether it is operator error or equipment failure.
There is a lot of rhetoric bandied about regarding how there is a seemingly inherent right to drive and need to adventure on the open roads, but somehow (1) the responsibility related to piloting something that weighs two tons or more is not talked all that much about unless one is taking a test at the DMV and (2) the fact that more and more areas are highly urbanized and therefore congested with both other vehicles and people on foot seems to be brushed over. Driving is hard even for those who are really good at it, and there aren’t very many people who are.
While we take automobility for granted, about 100 years ago the streets of America were entirely different places than they are now. According to Peter D. Norton in Fighting Traffic: The Dawn of the Motor Age in the American City (MIT Press; 2008), “before 1920 American pedestrians crossed streets wherever they wished, walked in them, and let their children play in them.”  Norton points out that in the early days of the automobile there were massive protests in cities across the U.S. against the traffic fatalities that occurred; in November 1923 the St. Louis Safety Council erected a massive monument dedicated to the 32 children who had been killed to that point during that year.
Today? Well, going back to “Quick Facts”: “Motor vehicle crashes were the leading cause of death for age 10 and every age 16 through 23 in 2015.”  
There’s no monument to them so far as I’m aware.
Let's circle back to the development of autonomous vehicles. To be sure, there is a profit motive involved across the board, whether it is at the level of someone supplying software, sensors or systems. There are the companies that are presently in the business of providing ride-hailing services that would like to take the driver out of the equation because that can be more profitable. There are the OEMs who figure that car ownership may give way to something that might be described as “car usership,” which is predicated, in part, by self-driving technology.
But I’ve had the opportunity to talk with a number of key individuals who are involved in developing the technologies—the sensors, the software, the systems—and the number of people who die on the roads every year is something that concerns them, something that they would like to minimize if not eliminate.
Drivers become drowsy. Drivers drive drunk. Drivers text. Drivers apply makeup. Drivers think they’re James Corden or Jerry Seinfeld (regardless of gender). Drivers eat breakfast sandwiches. Drivers get angry with other drivers. 
Autonomous systems do none of those things.
Yes, accidents will happen. But in the not-too-distant future, thousands of lives will be saved.