Google revealed this week that it has been testing a fleet of advanced driverless robotic cars, running them in regular traffic on extended trips all over California, including the 350 mile trip from Google's Mountain View campus to their Santa Monica office.
So far the cars (six Toyota Priuses and an Audi TT) have covered more than 140,000 miles, which makes this project far more ambitious than the DARPA Urban Challenge from a few years ago. In fact, Google recruited several of the top developers from the DARPA Challenge to head up its own project.
Each car is equipped with a phalanx of video cameras, along with laser and radar sensors to keep track of everything around them, and uses Google's vast mapping database for route planning. Up to this point, all of the cars have had a trained driver behind the wheel who can take over at a moment's notice, and a software engineer in the passenger seat to keep an eye on all of the systems.
Google sees this research as a way to develop a safer car, and believes by not relying on fallible humans, a driverless car can be much less accident prone than one driven by a person. They also envisage "highway trains of tomorrow", where groups of cars will run together over long distances to save fuel, while still providing the autonomy of a personal car. Productivity is also increased, as the passengers will now be free to do other work while commuting.
This is all very cool, but I wonder if computer systems will ever be reliable enough to depend on them 100% for your safety. When your laptop crashes it's just annoying, but if your car's steering computer crashes it could be a disaster. Already, all kinds of vehicles including cars and planes rely of computers to operate the various systems, but there's always a human in there who can take over if something goes wrong. I'm not sure if I'll ever reach the point where I would be comfortable rolling back the seat and take a nap, while driving down I-95 at 65-mph.
Google Blog, via Electronista
