When Google is out and about testing robot cars, something big has to be behind it. Probably plans to change society on a level and in ways as complex as their engineers can figure out, because Google certainly wouldn’t be looking for a ‘car solution’ without a ‘social car problem’.

Tuns out they’re envisioning “highway trains” made of cars, that would cut the number of road fatalities in half. Cars that can react to real conditions on the spot, but also be aware of pre-assembled road maps for the path they’re on.

It’s under these conditions that the automatic Google cars have been tested – the same heavily overpopulated road and sometimes brutal highways they’ll have to navigate in the future. Sure, there was prior police approval and a human passenger present, ready to override the computer and take control if necessary. And someone had driven one of the cars all the way through once before, by hand, pre-mapping the route ahead.

Good conditions for a test, but the cars still had to drive every mile of it and react to traffic conditions. The Toyota Priuses and Audi TTs used made it from San Francisco to Los Angeles, covering about 140,000 miles without incident – much more than any previous robot car system. Behind this accomplishment was an imaging array of video cameras, laser sensors and radars, which gave the cars a live image of the road and other vehicles on it.

So what’s Google’s idea for the future?

Cars driving better and more reliably than humans, with always-on attention, while passengers use the commute time for entertainment or socializing. Maybe even – though it probably still wouldn’t be wise – legally sleeping at the wheel.

And is there a step after that? There is, and Google probably already has it in mind. What will make things even safer is enabling the cars to be aware of each other, plugged into a sort of car social network, where their always-on presence further reduces the possibility of accidents happening.