Since 2009, Google has been working on cars that will take you where you want to go with the push of a button, no driver required. For a while now, you’ve been able to see them petering around Google’s Mountain View campus. There is one problem, though: Google’s cars might be just too perfect.
A Stickler for the Rules
A Google car is designed to follow the exact letter of the law. In a perfect world, this would be a good thing, but there is a rather large caveat: human drivers are far from perfect. Most of us don’t come to a complete stop at stop signs unless there is a police officer in view. For the Google car, this was a large problem: in one scenario, the driverless car came to a complete stop at a four way juncture. Human drivers continued to inch forward, and since the Google car is designed to wait for others to come to a complete stop, the program was paralyzed. Drivers zoomed away from all directions.
Humans and Computers Don’t Mix
Google has reported 16 fender benders since the inception of the program in 2009. All of these, according to Google, are the result of human error. In the latest accident reported on August 20, the Google car (rightfully) slowed to a stop in front a crosswalk to let a pedestrian pass, and was rear ended by a sedan.
A New York Times report said that the automated car had other issues: in one instance, the car swerved abruptly to avoid a poorly parked car; in another, it swerved harshly to the right to avoid a car that it sensed was speeding.
So how do you teach a car to drive defensively? It’s something that Google researchers will have to perfect before automated cars go mainstream. In any case, it seems like it will be a while before we commute with robots.