My kid’s school bus was almost an hour late this morning. It’s been about that late all week. He’s also been coming home very late as well, sometimes around 5:30 even though school let out about two hours earlier, and I know it’s not because of any after-school programs, because my kid’s whole modus operandi when it comes to school seems to be spending the minimum amount of time there. The reason is that there happens to be a pretty severe bus driver shortage where I live, though it’s not just in my area; there’s a shortage of school bus drivers nationwide. There’s a lot of reasons for this, like the absurdly low pay school bus drivers tend to get, and there really aren’t clear solutions just yet. Which is why I started thinking about this: Maybe school buses would be an ideal platform for automated driving?
[Editor’s Note: This is a reminder that Jason Torchinsky is an expert on the social elements of self-driving cars, having written the critically acclaimed book Robot Take The Wheel. -DT]
Now, of course I realize that we still haven’t fully developed self-driving cars; in fact, I think we’re a pretty long way from fully developing them, and I think that the semi-automated systems (SAE Level 2, where a human must be monitoring and ready to take over at any moment) that have been deployed onto public roads so far are deeply and inherently flawed. The more fully automated cars (Level 4, fully automated within a specific area) that are currently being tested have had plenty of their own issues, too. So why the hell am I even considering the use of autonomy for the vehicles that we entrust our precious offspring to?
The Bad Reason
Well, there are really two reasons, one admittedly terrible, and one that’s not so bad. Let’s get the terrible one out of the way first; it’s worth considering because, come on, it’s not like we’ve always had such great school bus solutions as it is, anyway. Look, most school buses don’t even have seatbelts. The drivers are woefully underpaid and overworked, and, get this, I grew up in an era when school buses were driven by high school students who could drive a bus instead of taking a couple classes!
That’s right, in North Carolina in the 1980s, until it was banned in 1988, 16- and 17-year old kids with maybe a number of months of driving experience under their belts were asked to drive us kids to and from school. I saw so many mailboxes decapitated by these barely-experienced burnout teens, and more than one flipped bus. They weren’t particularly safe, but, really, how much of a shock could that be if you stick a kid with six months’ of wheel time in their dad’s Chevy Citation and then plunk them down into a huge, top-heavy school bus? Of course it’s going to be a mess.
But this was the ’80s. People didn’t really start to love or give a shit about their kids’ well-being until well into the 1990s. So, I guess my point here is that trying out automated school buses shouldn’t be so alarming, because we have a rich and robust history of just kinda winging it when it comes to busing kids to and from school, and, while it wasn’t always perfect, it generally turned out fine. I’m alive, right? I mean if you call this living.
Okay, I told you the first one wasn’t really a great argument, but I had to get it out there, just to set that bar nice and low. I think whatever happens, it can’t be worse than letting 16-year olds drive the buses.
The Good Reasons: Regular, Predictable Routes
My real argument is that, among all of the possible contexts and use cases for automated driving, school bus routes may be one of the most ideal. I mean ideal for a very controlled Level 4-type of setup. Level 4 is full automation, no human driver needed, but only in a specific area. School buses are even one step better, because not only are they restricted to a given area, they follow the exact same route, every single day.
This is a huge deal for helping out with automation: you could pre-map and plan that route down to the inch if you wanted to, and the computers driving that bus would be programmed to follow that one route, and that’s it. Of course there would still be the need for sensors and cameras and all the associated AI that deals with other cars and obstacles and humans and all of that object avoidance, but knowing the exact situation, geography, and context for the drive is a massive help.
The Good Reasons: Low Speed
There’s also temporal regularity as well, in addition to geographic: School buses follow the same routes, at the same times of day, every day. All systems could be optimized for traffic and light conditions for each leg of the route, and adjust for changes in lighting throughout the year. Add to all of this the fact that school bus routes tend to be primarily low-speed driving, which adds an extra buffer of safety to the whole process. Sure, there are bus routes that require some highway stretches, but generally it’s pretty slow.
The average speed of a school bus on a route is 23 mph, with a minimum average of 13 mph, and the maximum average of 54 mph. School buses keep it slow, and for AVs at this moment in time, that’s ideal and gives a significant margin of safety.
The Good Reasons: Promo for AV Companies, Free For Schools
Companies like Cruise are already operating Level 4 rideshare cars in cities like San Francisco and Phoenix; what is demanded of an automated school bus should be an order of magnitude simpler than what their cars are already doing in San Francisco, where they operate at night and anywhere in a large, crowded, chaotic urban area. This is do-able, now. And, also importantly, I think companies like Cruise should be the ones doing it, and I think they should do it for free. At least at first.
There’s a lot of good testing and R&D that an AV company can accomplish by operating a fleet of AV buses for a school system. Companies like Cruise or Waymo could learn a lot, and it would offer them some fantastic PR and marketing opportunities. Think of the visibility of a Cruise or Waymo-branded school bus! Think of all the good press that would come from everyone knowing that, say, Waymo has solved North Carolina’s bus driver shortage with their miraculous technology! That’s gold right there.
Now, there are some specific challenges very specific to the job of a school bus. The biggest is that the moment any kid gets even remotely hurt by one of these, it’s all over. I know it probably shouldn’t need saying, but the safety of the kids is the most important part of this. And kids walk in front of and all around school buses; that’s just how it works. So there would need to be pretty stringent safeguards; if the bus door is open or the illuminated STOP sign is deployed, for example, there should be a mechanical switch that cuts power to the motors.
That way, if the bus stops and opens the door to let kids off and on, it physically cannot move, even if the sensors or computer systems make a mistake. The door-open-entry-exit time is the most dangerous phase of a school bus route, so that needs a very strong safeguard.
Dealing With The Kids, Maybe With Wasps
There would also need to be live monitoring of the buses on their routes, with feeds from external cameras to act as a failsafe for the bus’ own systems (there would be the means for an emergency shutdown, too) and internal cameras, because part of what a bus driver does is maintain order on the bus.
I think it’s possible to have a bus with no adult on it if the inside of the bus is monitored well. It may not be enough, of course, as it’s possible shit can really go sideways in there, but even in that case, a human bus driver is often also unable to do anything because they’re, you know, driving the bus, and can’t just get up and break up fights or whatever.
Loudspeakers could be on the bus so monitoring agents can inform students of things or let the jerky kids know that they’ve been caught. I’m also not opposed to having a large box of wasps on the bus with a remotely-actuated lock, and using that as a deterrent from kids acting like fools. Just make sure the kids know that if there’s any bullying or other bullshit, a code is sent and then the bus is full of angry wasps, and then nobody is happy.
Of course, that’s kind of absurd. All you really need is an opaque box labeled ANGRY WASPS and some buzzing sound effects and a very convincing and well-spread rumor that some kid’s friend’s cousins’ brother was on a bus when they remotely released the wasps, and it was absolute hell. The kids just need to believe in the threat of the wasps, you see.
Okay, I’m maybe being a little silly about the wasps, but I do think there are likely some options for school buses to operate without direct adult supervision; they would of course be GPS tracked and have remote adult supervision. I’mm not sure how many parents may be comfortable with that, but the whole point of this is that there simply aren’t enough people to take these jobs, so an automated bus that still needs an adult supervisor would still have that issue. Of course, it could open the job up to people unable or unwilling to take a bus driving job, so perhaps it could be helpful even with the requirement of an adult on board.
So, here’s my elevator pitch: make school buses automated, even with existing Level 4-ish technology because the low-speed, consistent-route school bus use case is ideal for really optimizing what AVs are good at; this can help alleviate the nationwide school bus driver shortage, adequate monitoring can be implemented to keep the kids safe, and besides, it’s hardly the most reckless thing we’ve tried when it comes to busing our kids to and from school.
I should probably leave out the bit about the wasps.
Support our mission of championing car culture by becoming an Official Autopian Member.