There are two schools of thought out there as regards autonomous vehicles:

1) Make the car smart enough that it can handle any situation

2) Make the map good enough that it can handle any car

Google is going with (2). Tesla is going with (1). Google obviously wants to sell maps. Tesla obviously wants to sell cars. Personally, I think (1) is going to be hella harder to implement in an "every day" sort of way because there's no good verification. There's no audit trail. There's no real way to say "no, this won't run over a pedestrian." I mean, the first autonomous vehicle fatality was a Tesla that didn't "see" a semi crossing the freeway.

This article is about Volvo, which is kinda-sorta doing (1) but not in a serious way; they're probably looking for a viable way to nickel'n'dime Google (or the other big European mapping concern which veen can elucidate at the drop of a hat) into getting a better rate. I, personally think we'll see a lot of (1) on proscribed routes that eventually spread out from population centers, but that's an educated guess.