A report back in 2016 said that after two million miles, Google’s robot car drives better than a 16-year old. I wonder which 16-year old they are talking about, because mine drives really well, but I’ve seen some of the 16-year olds around Sentinel High School drive and wondered if I can report them to the Missoula City Police Department to have their license revoked.
By the end of 2017 the miles driven by Google without a driver was up to 4 million. Waymo, the Google company that runs the Google self-driving or autonomous vehicle program has said that “we practice rare and unusual situations, such as people jumping out of canvas bags or skateboarders lying on their boards. We can also prepare our cars to handle some of the most common scenarios that lead to collisions, such as aggressive drivers barreling out of driveways.”
I haven’t seen many canvas bags in Missoula, so I think this might not have been the best choice of material for the bag.
If Google has gotten all the way to testing its self-driving vehicles on people jumping out of bags, then it must be time to consider whether or not we should be waiting to buy a new car until these self-driving cars are available for purchase, right? The current best guess is 2019, which is just next year. But there is a big “if” and that “if” has nothing to do with the folks developing self-driving cars. It has everything to do with the regulators, and the insurers. And where you live.
If you’ve driven around Missoula lately you’ll have noticed that the roads are icy and the potholes are breeding with each other. Almost all of those 4 million miles self-driven by the Google car are in sunny northern California. Driving on snow and ice and around potholes is very difficult indeed for self-driving vehicles. Potholes is actually the easiest of these as the cars have sensors to detect these and can let other cars know about them too as they emerge. It will be some time before Google focuses on improving the snow and ice-driving technical ability of their cars since the markets in California and Texas, which are not known for their snow, are the easy first targets for selling self-driving vehicles.
Regulators and insurers are likely going to be the biggest hurdle this technology faces. At the best of times regulators are not quick, but when it is something completely new like this, they typically want to wait to write regulations until they can see how things need to be regulated. That puts insurers in the driving seat (pun intended!).
It’s one thing to insure those 16 year-olds driving around Sentinel high school because the insurers have had decades of experience knowing the odds of crashes and how expensive they are going to be. But who is at fault when a self-driving car makes “the wrong decision.” What if someone on a sidewalk jumps out of the wrong sort of bag and gets hit and injured? Surely the person owning the car is not at fault – after all, they weren’t driving? Does that mean the software programming company is at fault instead for not testing every type of bag that could hold a potential threat?
Insurance companies have plenty of experience insuring drivers, but not the programmers of self-driving vehicles, or really any programmer for that matter. This change in how insurance is going to have to work is going to have to be almost completely figured out before you can wander down to your local self-driving car dealership and be driven off the lot by your new vehicle.
David Firth is a professor of management information systems in the College of Business at the University of Montana and a faculty fellow with Advanced Technology Group in Missoula.