Any discussion of driverless cars is likely punctuated by the question “When?” Chris Urmson is the director of driverless cars for the Google Self Driving Car Project, but the best he could do for an answer is between 3 and 30 years.
That may not be much help is you’re wondering whether you will be able to trade in your Toyota for a Google when the lease expires. But the point that Urmson was making during his presentation at SXSW Interactive is that this is a technology that will be rolled out incrementally over time. You might be able to catch a ride in a driverless car on a freeway in sunny, dry Arizona, before you can get one on a blustery winter day in Detroit.
What Urmson did offer up to his audience was a fascinating look at some of the technology behind the Google self-driving car. The cars do not use GPS. As all of us know who have been dropped off of a highway ramp into a stretch of desolateness while our GPS announces ‘you have arrived at your destination,’ GPS isn’t accurate enough if you don’t have a human driver. Instead the Googlemobiles use a combination of maps and sensors. The car’s OS captures 1-1/2 million laser measurements per second. It can zoom in and see up to 200 meters. And it anticipates the actions of other cars on the road 10 times per second. That information can be used, for example, to identify signs that a car with its left turn signal on is really going to try to shoot a U-turn. The vehicles are also equipped with something called anomaly detection, which, Urmson pointed out, could identify if some folks are playing frogger with your car.
Google driverless cars have already racked up 1.4 million miles on public roads. They currently do 10,000 miles of road testing very week in addition to 3 million miles of simulation testing daily.
Urmson made a compelling case for the driverless car. “The technology can’t get into the world fast enough for safety reasons.” He recited the statistics of 38,000 fatalities on U.S. roads every year. Globally the number is 1.2 million.
There are other potentially important benefits for the self-driving vehicle. If provides convenient transportation for people who can’t drive due to vision impairment or illnesses. If aging folks might seem to be losing some of the sensory sharpness that enables safe driving, why not transistion them to a driverless model? It also might alleviate the amount of time you spend sitting in traffic, or at minimum it gives you the ability to do things while your robotic chauffeur sits in traffic. One of Urmson’s slides calculates that the amount of time Americans sit in traffic every day is equivalent to 162 lifetimes.
But getting back to that question about when you’ll be able to trade in your Toyota for a Google the answer is probably never. Google has no interest in making cars, according to Urmson. Its interest is in the technology and they would look to partners for the manufacturing. Nor is Google interested in producing technology components, such as automatic braking, for driver-operated cars. Their goal is the fully driverless car.
Here’s some good news. Arunson believes the technology will become “relatively inexpensive” and will be accessible to everyone. Guess the car companies will have to get us on the next generation entertainment systems that we can enjoy while we are robotically escorted from place to place.
Urmson’s presentation can be viewed on the SXSW Interactive Channel on YouTube or by clicking here.