Driving from Macon to Atlanta on I-75, you may think that it would be fairly easy to engineer a driverless car to deal with such a route. And while I-75 is not some lightly travelled route, after all it is six lanes most of the way, the wide lanes, gentle curves and controlled access intersections limit the number of unpredictable events you may encounter.
Sure, driving the road still comes with any number of potential causes of a car accident, from deer crossing in twilight to a recapped semi-trailer tire blowing out as you pass by, but as compared to a busy city intersection, the interstate makes for a easy drive, especially for a computer.
As driverless car prototypes are already operating on U.S. highways, you may expect that they could appear in your favorite dealer's showroom any day. That may be overly optimistic. The interstates, for all their challenges, are the easy part. Developing a car that can safely and cost-effectively deal with roads that contain more erratic traffic and other idiosyncrasies is the greater problem.
An experienced driver, paying close attention to literally thousands of stimuli as they await the change of a turn signal, can reliably determine when it is safe to turn and how to remain in their lane as they turn, accounting for traffic flow and unexpected pedestrians.
As one engineer notes in a story about driverless cars, "Just looking at a traffic light and deciding if that traffic light is for you is a very, very complex problem." And the technology that can process and correctly execute all of this information must be affordable. Much of it can be done today, but the car would cost hundreds of thousands of dollars.
Source: MIT Technology Review, "Driverless Cars Are Further Away Than You Think," Will Knight, Oct. 22, 2013