Furthermore, all of those research prototypes are geofenced, since they rely on extremely detailed mapping and lots of training data from humans driving the same routes. There is not a single AV in the world close to capable of driving on any road, at any time, in any condition, that a skilled human could drive on without ever having seen it before.
In my mind, that means we are indeed nowhere close to a true AV.
But when I think "autonomous vehicle" I definitely don't think "arbitrarily capable of driving on roads" - I think something capable of navigating well-understood national roads. I don't see any inconsistency with the idea that there'd be "no automation" zones or roads, or that pre-approved travelways for AVs is a failure scenario.
My (and Doctorow’s) definition of AV is so stringent because AVs need to deliver the same value add of regular cars relative to other forms of transportation: i) relatively fast, ii) unscheduled, and most importantly, iii) point-to-point transportation between any two points connected by roads. Trains are bad at iii), decent at ii) if run often enough, and far superior at i). If iii) is off the table, then the relative value add of cars is greatly diminished.
Obviously we don't know to what extent they have remote drivers but you wouldn't be able to run such a service if every ride required them.
[1] https://www.youtube.com/channel/UCP1rvCYiruh4SDHyPqcxlJw