An article in the print edition of Autocar magazine suggests there are problems at Google with it's AV program. I'm not sure how valid the report is but in summary
1) Their AV Chief Engineer is said to have resigned.
2) They have been dealt a big blow by the California highways dept. which has insisted in its AV on the public highway legislation that a steering wheel and controls remain on AV’s. This is something Google has fought as it sees an AV as completely free of human control in production.
3) Their testing mileage is still low, quoted as 1.7M miles and never above 25 mph.
4) The Google maps software is now realised as inadequate for safe AV control so they need better mapping and LIDAR etc.
I'm not sure if any of those issue will stop a company with the huge cash reserves of Google but it does suggest that the more incremental approach of the traditional mfrs. may be closing the gap to Google’s AV project. Certainly on-highway testing is essential and Google still seems to drive its prototypes only around Silicon Valley.
The last point about the limitations of Google maps for AV control is maybe the most interesting. Apparently Google maps are a fairly simple 2D database. My son who works with digital mapping surprised me about 12 months ago by explaining that Google maps may be very useful and are the de facto source for route planning and looking at potential holiday spots etc. but are not considered very good by in depth digital mapping experts. So what one might think of as Google’s obvious competitive advantage may actually be weakness.
Maybe that’s why Mercedes bought up one of the very detailed digital mapping companies recently and might now have an edge on Google.
I'm sure Google can fix these things but the question is how much will they spend if they can’t secure new revenue sources out of it?