For example, let's say they can only run during daylight hours, in conditions other than rain and snow, only on long-haul routes, and humans will do the "last-mile" piloting on local roads.
Now let's suppose that for routes that fit those parameters driverless trucks end up taking over 40% of those kinds of routes. That's huge. It would absolutely transform the economies of all the states that are looking for truck drivers here. http://media2.policymic.com/bf05a1c9e6b8a55095f8b4726e30b52f...
Furthermore, a lot of these sort of technologies have already made it into lorries and cars, just commonly called 'driver aids'. Such as radar controlled cruise control, lane control, etc.
Most of lorry hauling takes place on motorways, where the process can be easily automated. Only when trips from the origin to the motorway and the motorway to the destination could remain in control of the driver.
And moreover, this sort of technology could be fitted to lorries very soon. Fully automated self-driving lorries are still decades off. Partly of all the minor quirks and issues, but mostly because once more than 50% of the actual lorry driving is automated, there will be less urgency for it.