Recent reports show that Waymo robotaxis have, on at least 19 occasions just in Texas, passed school buses which were stopped in the opposite lane. That violates the vehicle code which requires vehicles in both directions to stop. NHTSA, the federal safety agency asked questions, and Waymo reports they have fixed the problem and issued a voluntary “recall,” which is to say did one of their frequent software updates. Waymo claims the problem won’t happen again.
But should they have needed to fix it? And why are school buses letting off children where they must walk along or cross a busy street?
This event got a fair bit of attention, because of the natural desire to be extra demanding of high safety standards in anything involving children. Laws require cars to stop in both directions when a school bus stops for children to assure an extra degree of safety, particularly because children may cross the road getting on or off the bus, or otherwise be near the road. We don’t trust drivers in this situation (or children.)
While Waymo admits they erred, and the law is clear and they have fixed the problem, there are some interesting new issues which occur when we consider robocars and how they change the world of transportation and roads. To be direct, the law is clear and Waymo must follow it, but what should the law be?
Was It Unsafe?
Waymo says the vehicles were deciding to proceed after initially slowing or stopping. This suggests they identified the presence of a school bus, but may have not identified that it was stopped, or thought it was finished its stop. (It seems unlikely they were not aware of the rule requiring stopping, but that would also explain this behavior.) Robocars will make mistakes, and continue to make new ones for decades to come. What matters is whether the mistakes cause harm, and the important fact that they generally get fixed, and for all vehicles, including competing vehicles from other companies.
Nobody was hurt, and it seems likely nobody could have been hurt, as long as the vehicles kept to low speeds. Unlike humans, Waymo’s robotaxis have shown themselves to be very reliable. They would have seen any child crossing the road, or making like they were going to cross the road, and stopped for them. In a mature robocar, that’s something that can be depended on in a way you can’t depend on it for humans. (Whether the Waymo can be classed as mature is a different question.) The vehicle code was written for humans, based on our not being dependable. Fine one human for a mistake and it doesn’t mean all other humans will never repeat that mistake–it doesn’t even mean that particular human won’t repeat it. Many intersections are marked “no left turn” or “No U-turn” even though people could pull it of 99.9% of the time, because every so often that one in a thousand chance would turn bad. Robots, on the other hand, can always judge the timing, and know when it’s safe to go. It improves traffic if they do.
This principle applies even to safety rules about children, though the bar is of course much higher. But the bar is not insurmountable.
Robocars as School Buses
Today, in the USA, a little over half of children go to school by bus, in particular the lower income ones. 30-35% are driven in a private car, and the rest walk or bike, particularly if they live within a short distance of the school. In the coming robotaxi world, we might expect children over a certain age to take a robotaxi, and many below that age would still take a robotaxi accompanied by an adult or older sibling. Waymo already allows 14 year-olds to ride alone in Phoenix.
Young children need supervision, which comes from a bus driver. If not, an interesting solution would present itself, namely using smaller, van-sized vehicles instead of buses. Such vans would not only provide more direct service with fewer stops, they could usually avoid stopping on busy streets, instead using driveways and side streets. (If you have 40 children in a bus, you can’t pull into a driveway for every stop, but you can if you have a much smaller number.) For children over 12-14, this van approach could work, but at younger ages a supervisor is needed.
That supervisor could be a hired person like a bus driver, but it could also be a rotating parent, or a trusted teen also on the way to school near the elementary school. Indeed, that person might get paid for their 20 minutes of work and still be affordable. Pooled robotaxis could take smaller groups of 4-6 children, and be fairly affordable to a wider swath of families if desired. Parents commuting by robotaxi could accompany their child (and perhaps other children) and continue in another robotaxi to work. Families with a private robocar could have it take the child to school and return home, though if the robotaxis are priced to match the cost of private car ownership, the one-way robotaxi trip might compete with the round trip in the private vehicle if the trip is not short.
Children and the Shape of Cities
Robotaxis are going to vastly change the lives of the children allowed to ride in them alone. Suddenly they will have the mobility of adults, if their parents will pay. They will be able to have schools, activities and friends they can reach that are much further than their current bike and transit range. This, in turn, will give parents much more liberty on where they have their home–today the needs of the children figure strongly in that. That will have big affects on the distribution of population in our cities and suburbs, and even rural layouts.
