We don't have self-driving trains yet, except in very tightly controlled environments. So I think it's going to take more than a few test runs to show that self-driving trucks will displace drivers anytime soon.
Mining companies in Australia are already using self-driving trucks to transport ore.
http://www.businessinsider.com/rio-tinto-using-self-driving-...
And self-driving cars are just around the corner in the USA. Trucks will follow maybe half a decade after that, at most.
It also seems to me that such things won't be a big deal. Some Google demo shows their system recognizing a bicyclist making a hand signal. It's the dynamic traffic stuff that will take longer to deal with.
I was thinking about one problem the autonomous systems will have to handle on my run yesterday; pedestrians can look at drivers and use eye contact to make sure the driver has noticed them. An autonomous system doesn't have eyes to look at. The first idea I came up with was a highly directional light source that the system could point at the system could point at the heads of pedestrians it had recognized. Not a laser, but a moderately bright light in a recess that limits the direct visibility of the source.
(Yeah, yeah, I know about the kilometre long Australian semi-trailer trucks, but those aren't really practical in most situations...)
(They are still really imposing and impressive, especially at night. On the extremely flat, straight desert roads in the Outback, you can see them miles before they arrive, like a star on the horizon. When they finally get to where you are, they fly by with a hefty gust of wind. They have steel "roo bars" mounted on the front for efficiently generating roadkill. They are loud, brightly lit, and more than half a football field long.)