I dealt with an especially tragic fatal crash when I was the city traffic engineer in Maple Grove. A high school junior and her younger sister were driving to school one morning. The driver cut in front of a semi-truck at a rural, slightly skewed, side street stop sign controlled intersection. She survived, but saw her decapitated younger sister in the passenger seat. The situation still makes my skin crawl more than fifteen years later.
My condolences to every single family who experiences a loss due to an automobile crash, including the family of Joshua Brown who was killed due to a mistake by the Tesla Autopilot system. It is part of our mission as traffic engineers to improve the safety of transportation and I take that responsibility seriously.
So let’s dig into the first fatal crash with a “self-driving” car from a traffic engineering perspective:
The initial news coverage of the crash implied we should be quite concerned about autonomous vehicles and this crash is evidence that maybe we should halt our progress towards self-driving vehicles. Unfortunately, they got the headline wrong.
In 2014, about 33,000 people were killed in automobile crashes or 1.08 people per 100 million miles traveled (1 death per 92.6 million miles traveled) according to the U.S. FAR system. Tesla has reported its autopilot had driven 130 million miles before killing its first person.
Digging into the mortality statistics from the U.S. CDC, automobile crashes are the leading cause of 15-24 year olds and are one of the three leading causes of death for all people aged 1-64 in the United States.
Applying Tesla’s safety rate vs. human drivers would result in lowering fatalities in the United States by about 30%, many of whom are young adults whose lives are cut tragically short. If we could improve our overall vehicle safety to that of Tesla, about 10,000 people would not have lost their lives last year in car crashes.
We are at the dawn of autonomous vehicles and the Tesla autopilot system is already proving to provide a big jump in safety. And the remarkable thing is that Tesla will change its algorithms to improve its safety the more miles it travels based on these early crashes. Please don’t let the media scare you about autonomous vehicles.
The common industry estimate is that a fully mature autonomous vehicle fleet will reduce fatalities 90% in the United States. And I’m not going to dig into many parts of the world that have fatal crash rates much higher than the U.S. A rational response in the near future would be that humans are such bad drivers compared to computer driven vehicles that it becomes illegal for a human to drive a car.
I wish those sisters fifteen years ago were being driven to school by their autopilot car instead of an impatient teenager. That crash killed one sister and hugely damaged the survivor.
The headline from the Tesla Autopilot fatal crash should be that
“Self-Driving Vehicles Save Lives.”