I was discussing the benefits of fully automated driving with a friend of mine the other week. What I think he couldn't quite grasp was how it would look if all the cars were automated. If we have driven, we have all likely witnessed poor driving decisions made by people that have resulted in accidents or near accidents, both of which have an end result of causing back-ups (and the possibility of related accidents). Also, people fundamentally cannot make instantaneous decisions based on knowledge of the entire network (or even a regional portion of the road network). Things are getting better, with smartphone navigation having information about traffic congestion, and even providing alternatives that circumvent it, thus wending our way toward network-smart driving.
However, even with this, it would be difficult to maximize the potential efficiency of a four-way intersection. Especially if each direction had five lanes of traffic. We currently rely on traffic lights and patterns to try and move people through the intersection. However, even with these rather crude measures, people still manage to balls it up. But what if it were automated?
Well, it would look something like this:
Pretty cool, right? Kinda scary, but pretty cool. "What about pedestrians?" you might ask. Well, there are cars that already take into account various "obstacles" that exist on the road, as well as navigating within lanes and keeping in relative motion with respect to other vehicles. In fact, Google uses one for its street mapping:
And its operation is pretty cool, too!
UPDATE (April 2, 2012): Steve Mahan - a 95% blind man - is out testing one of Google's self-driving vehicles!