For people like me who are interested in automated vehicles (AVs), a moment we all dreaded finally happened. On Sunday March 18, 2018 an Uber AV, in self-driving mode with a human occupant acting as a safety monitoring, collided with and killed a pedestrian. According to some reports, the collision occurred at night, and involved a person crossing a road outside of a crosswalk and walking with a bike. Forget for a moment that this is the kind of language used to typically blame the pedestrian in auto-involved pedestrian fatalities, the most telling bit of information came from The Atlantic:
“Tempe police reported in their preliminary investigation that the vehicle was traveling at 40 miles per hour.”
This is an important fact, as multiple studies have shown that the chance of a pedestrian dying when hit by a car jumps from around 5% when a vehicle is traveling at 20 mph, to more than 83% when a vehicle is going 40 mph.
When I speak to cities, clients, and other interested parties about the topic of AVs, I often talk about how complex our roads really are. In vibrant cities — and I would consider Tempe, Arizona vibrant — there are all kinds of activities on, alongside, and near roadways. When I served as City Traffic Engineer in Santa Monica, California, I would often explain to any interested party that our City was the kind of place where you might encounter a person walking in the street or crossing the street outside of a crosswalk. In California this is considered illegal — just like running a red light or speeding is illegal — yet we must expect these things to occur. Maybe it’s because I’ve lived in cities all my life, or maybe I’m just being a harsh realist. In any case, we must recognize that there is the potential for pedestrians to be on or in or crossing the roadways of our cities, sometimes outside of crosswalks and sometimes at night. It is true in Tempe, and it is true in cities across America and the world.
This is important when we think about AVs and where we test them, and ultimately where we see them operate. We should be asking very critical questions about the place of AVs in our vibrant cities. My main question is: Why are we allowing AVs to be tested in places where they might travel at speeds of 40 mph or more and where they might encounter a pedestrian in the roadway? Clearly the vehicles aren’t capable of rapid-response to avoid the collision, and clearly a collision at 40 mph or more will most likely kill the person struck.
This is not to say we shouldn’t continue to experiment with AVs. Rather, we have to carefully think about where we want to test and use them. This is something I’ve written about before, and it warrants repeating.
When it comes to AVs on public roads, I recommend dual paths:
- Low-speed AV applications in urban/active areas: It is time to recognize that cities and other agencies that plan, build, operate, and maintain our roads need to take a stand and define exactly where (and maybe when) AVs can operate. In places where there is a likely possibility that pedestrians will be in the path of AVs, the speeds of AVs should be low enough to ensure a collision is avoided, or at least low enough to minimize the risk of fatality. This means testing them in low-speed urban environments, preferably in places where other human-driven vehicles are restricted, to minimize the chaotic effects of erratic human drivers and ensure safety for everyone on the roads.
- High-speed AV applications in limited access facilities: If we want to test AVs that go at speeds over 40 mph, we should do so in areas where the chance encounter with a pedestrian or other vulnerable road user is reduced to an absolute minimal level. Furthermore, we should be limiting the potential interaction with human-driven vehicles as well, as we know and have seen how human drivers can act erratically and collide with AVs. In these cases, we should be thinking about testing AVs in places like freeway toll lanes and other limited access, managed facilities.
It is very sad to learn of the death in Tempe over the weekend. I hope that we can harness this moment to align the priorities and objectives of the public sector, as well as with AV researchers and implementers, so that they can help ensure the safe deployment and integration of AVs. Let’s use this as a call to action, and most definitely let’s use this as a key discussion point in the pending federal-level debates about new AV legislation like AV START and the SELF DRIVE acts. This is something the National League of Cities and their respective State chapters should get behind immediately.