|AV crash in Tempe this week (Photo found in NYT)|
Since then media is in overdrive criticizing the AV technology from that the AV is obviously not ready for prime-time to more outlandish declarations. Wired magazine went as far as saying that the AV now has officially a worse statistic than the human driver since normal mortals kill about 1.16 pedestrians per 100 million miles, a mileage that all AV's combined have not reached yet.
Nearly 40,000 people died on American roads last year. Almost 6,000 of them were pedestrians—that’s more than 16 per day.But human drivers kill just 1.16 people for every 100 million miles driven. Waymo and Uber and all the rest combined are nowhere near covering that kind of distance, and they’ve already killed one. (WIRED)It isn't known whether the back-up driver was as alert as an active driver would be, fact is, he didn't intervene and was as surprised by the pedestrian in the roadway as his vehicle. (A video shows him/her taking the eyes off the road for several seconds at a time). But, of course, the technology should have prevented such a crash, what with LIDAR (a type of radar), infrared and cameras, gizmos present on the AV that should have seen a person even if it would have been pitch dark. The analysis is still ongoing, even if all the devices all worked fine, a cause for the crash could also be programming. The issue usually is not what data are available but what one does with them. This applies to people and AVs.
Diagram prepared by NYT showing the crash area
The now commonly available half-autonomous vehicles with their dynamic cruise-control, automatic emergency breaking, self parking and lane control give the driver a good idea what can go wrong: The driver, set for a lane change, is accelerating behind a slowing vehicle to make a gap in the adjacent lane. This causes the vehicle to go into panic mode because it is sensing an imminent crash due to the large speed differential between the front vehicle and the approaching one. The planned lane change is in the driver's mind, but it isn't programmed or anticipated. An Interstate going around a curve while the exit ramp continues straight with stopped vehicles at a nearby signal will make a car in dynamic cruise control slow down drastically, even though the Interstate itself may be totally open and the driver has no intention of pulling up behind the cars on the exit ramp. The system sees what is straight ahead, not what is around the bend. Construction barrels, pedestrians on sidewalks and bus stops are not supposed to stop a passing car. A woman that walks a bicycle across the street was apparently not recognized as such a person or even as an obstacle. A sudden swerve was apparently not programmed into the system and an emergency stop wasn't even initiated, even though the system vendors say they can recognize pedestrians going perpendicular into the path of a car, as a child would which is running for a ball. It isn't known yet what the Volvo processed and what not.
|A crosswalk about 400' away from where the crash occurred |
In spite of all the public hoopla about AV technology, the lessons to learn from this incident are not so much technology or that it is necessarily more flawed than humans. Instead, the lessons are quite simple and have to do with the everyday flaws of our roadway designs which are made for cars only and not for people who walk, bike or simply stand around. The multi-lane road in Tempe has sidewalks but safe pedestrian crosswalks are far apart, even though there is development along the road which walking or biking people would want to got to (A park, a theater etc.). Many similar roads have no sidewalks at all. The road has four through lanes plus additional dedicated turn lanes and even a marked but unprotected bike-lane on each side. The speed limit is 45, the Uber car was going 40mph. Such a speed is too high to allow a sudden stop and, in the case of an impact, the speed is deadly for a pedestrian or bicyclist. AV or not, as long as those conditions are the norm in suburbs all across America, the death toll of pedestrians and bicyclists will remain high. (The pedestrian death toll rose from an all time "low" of 4,000 fatalities in 2009 to over 6,000 in 2016, a 50% increase in just 7 years! The media hoopla should be permanent, but it should not only address technology but the causes of the high number of pedestrian and bike crashes that happen all year long. (There are no convincing explanations for the increase except speculations about an aging, frailer population, more distracted walking and driving, faster cars, and more walking).
In that context it is a good sign that Baltimore's councilman Ryan Dorsey was finally allowed to introduce his Complete Streets bill which has been in the making for some time. Even better, he also received a national award for the way he and Bikemore conceived the bill as a measure of equity and social justice.
"We have communities in which 70 and 80 percent of households lack access to cars, but we use our public resources … primarily for cars. This bill is about building a Baltimore City that works for everybody who lives in it.” (Ryan Dorsey)It is telling that the first victim killed by an AV was a homeless person. There is no equity in who is most at risk of getting mowed down by vehicles, human driven or not. Dorsey's award notes that in his Complete Streets approach with a focus on equity Baltimore has become a leader after all. Congratulations!
Council during Dorsey's bill introduction (Bikemore)
Baltimore’s Complete Streets ordinance embodies the themes of equity and implementation in the National Complete Streets Coalition’s new and improved policy framework. Other jurisdictions can and should adapt and build upon the model of community engagement used to draft Baltimore’s ordinance, as well as their primary focus on reducing safety and accessibility disparities by embedding equity considerations into every step of the program. (Smart Growth America)
Klaus Philipsen, FAIA