What also may have gone unnoticed was a January news item that Waymo had ordered "thousands" of Chrysler Pacifica minivans from Fiat Chrysler (NYSE:FCAU), which is the car Waymo used in the 100-vehicle early rider program. The order signals that Waymo is ready for widespread deployment, with Waymo CEO John Krafcik announcing, "with the world's first fleet of fully self-driving vehicles on the road, we've moved from research and development to operations and deployment."
...
Not only did Waymo score more miles, but its disengagement rate fell to 0.18 per thousand miles, down from 0.20 in 2016. For reference, a disengagement occurs when a human monitor has to assume control of the autonomous vehicle. Waymo's figure was also well below GM-Cruise, which scored a 0.8 disengagement rate, though GM claimed this was a huge 1400% improvement from 2016, which, admittedly, is also impressive.
These figures are kind of meaningless unless you know how much they're gaming the results by driving lots of "easy" miles. Also, I suspect (but don't know) they're obscuring first/last 100 feet problems. But even if you take it at face value that you can run an automated taxi service that only needs a driver once every 5 thousand miles, you still need a driver (I mean "human monitor") all the time if you need one once every 5 thousand. Neat if true! Still not good enough, and no way they work that well in my urban hellhole, which is pretty easy to drive in compared to, say, Boston.