Easymile self-driving shuttle
Recently an EasyMile self-driving shuttle operating in Columbus, Ohio came to a sudden stop from 7mph, apparently for no visible reason. A passenger seated in the shuttle fell from her seat, sustaining minor injuries. As a result, NHTSA has ordered EasyMile to not carry passengers in their testing until further notice.
EasyMile has been a popular choice for “last mile” shuttle projects using self-driving. Several companies have attacked the shuttle problem, because it’s easier. Such shuttles tend to go slow on limited routes. More than a dozen such piliot projects were underway in the USA, and many more around the world. At present, these prototypes still have a human inside who doesn’t steer, but can cause an emergency stop for safety.
The vehicles themselves also will stop suddenly if they detect something in their way. In the prototype phase, you actually want them to be conservative and stop if they detect any problem. This will include what are called “false positives” or ghosts, where you brake for something that isn’t there. Apparently this was the case here.
While you don’t want false positives, they are much better than false negatives, where you don’t see something on the road and hit it. So in the broad sense, EasyMile was doing the right thing, though we don’t have more specifics about what took place. Does it make sense to shut down all their operations?
While not reported, typically passengers on shuttle buses do not wear seatbelts. It’s not entirely clear why they don’t, since even those driven by humans will come to sudden stops some of the time, though not usually for ghosts. We allow passengers to stand on these vehicles, but those who stand usually do it in a way that they can brace or grab something if there is a surprise stop. It is a person in a forward seat with no seat in front who is at risk of being thrown forward by sudden braking. (People in rear facing seats should normally do fine.)
Almost no transit uses seat belts. That is partly because transit vehicles do have lower accident rates than private cars, but the rates are not zero. The main reason is tradition and convenience. People are constantly getting on and off transit vehicles, and if you made it so that the vehicle could not leave until all are belted in, and people could not unbelt until it comes to a stop, transit operations would be much slower, especially on a busy line. And you might even feel all passengers would need to be seated, which could greatly reduce capacity and passenger flow.
The desire to mirror the belt-less style of many transit lines has led several of the self-driving shuttle companies to not use seatbelts. Once they are in fully safe production, it may make sense to have them be beltless, but during the prototype phase it may not be the best choice. Of course, once you have belts, it’s very hard to remove them, which then leaves you less competitive with the existing shuttles.
One idea might be to have a vehicle that knows if passengers are belted, and allows them to be unbelted when it starts accelerating, but never goes above a certain speed until it gets the signal that all passengers are belted or hand a hand on a pole or loop or seat. Likewise, when slowing for a stop, once it is below a safe speed, it could let passengers unbelt before it comes to a stop, to give them the chance to get to the door as they would if never belted. It is unclear though, what you do if people undo a belt while at high speed — no vehicle or driver has ever reacted to that before.
Even with this, having to put on belts is more work, and wearing them is more constricted and less comfortable, so people will resist it while older transit vehicles require no restraints. This is not a problem for more private vehicles (for example Waymo vans use belts) because people are used to belts, and on private door to door service they do not waste much time.
Even after the cause of the sudden braking on this vehicle is found, the general concept will not go away. In fact, the supervising operator of these vehicles doesn’t get a traditional wheel and brakes — their main task is to stop the vehicle if there is a problem, and passengers may get thrown. If there is time, braking can start more slowly to give people a chance to grab a rail.
This problem is also hard because people like the option for face-to-face seating on these shuttles. That means people will be facing forward with no seat in front of them, the way it is on something like a school bus. There will be stops, even if they are not for false reasons. If they run like transit shuttles, people might fall out of forward seats. It was overkill to shut down all passenger operations because this took place, and doesn’t suggest a good future where people can test these vehicles under the same circumstances as transit shuttles exist.
I will point out that I’ve never been a fan of shuttles. Several companies build them because it’s a much easier problem than a door to door car. But acting as a “last mile” shuttle is not what these vehicles are for, and it’s not how they will change the world.
