The scenario that’s most plausible is that there is a set of trunk highways (such as the Interstates) that have intensive infrastructure for automated trucks. The automated truck will park in a staging area (the manual ones already often do, because of the fact that tandems are usually infeasible on the surface streets) to be joined by a human driver for the ‘last mile.’
That will eliminate probably 75% of the long-hauler’s hours. Moreover, the short-hauler may run on a slave-labor business model like Uber.
In any case, that’s the right starting point for the economic analysis - it doesn’t presume an immediate advance to technology capable of navigating a semi through city streets, but only in a controlled environment - possibly even a dedicated lane.
Paradoxically, it’s also a model that starts looking like container freight aboard a railroad, without the efficiency that comes of running steel wheels on steel rails. But whatever.
