They might be a hard problem, but if you can't figure them out you can't have robotaxis. Sure you can have niche applications of automated vehicles which are probably more neato than commercially viable (maybe the latter, too), but not useful robotaxis.
Anyway my bigger point is whatever happens with automated vehicle technology generally, the focus on "robotaxis" was always absurd, both as a technological issue and as the idea that it was the path to riches for companies like Uber. Making them kinda sorta work in a geofenced bit of a slice of suburbanish Phoenix or newer California after a years-long dedicated effort is very different from creating something which can adapt to the mean streets of San Francisco or Philadelphia.
And more realistic applications, such as fixed or semi-fixed route shuttle systems in lower density environments, while more likely to be technologically feasible, aren't much more likely to be something communities are going to invest in. Sure drivers are expensive, but so are roboshuttles, and demand for these things is likely to be too low in most places where people they imagine they should be for the same reasons systems with drivers have low demand in these places. What if a bus was more like a taxi is always the fantasy, and the idea is that if you make them driverless this is for some reason more possible, but it still doesn't make any sense.
