I got to race around in one of the Google robocars, and it was thrilling. Photos: http://www.flickr.com/photos/jurvetson/5499949739/
It felt like a racetrack demo lap, where the instructors show off what they can do, with wheels squealing at the edge of performance... a newfound performance, in this case, eked out of an intelligence-enhanced Prius with no human driver.
Imagine being driven entirely by the car for your work commute... in traffic, through tollbooths, and across bridges to the front door. Some Google engineers are already doing this every day.
No need to look for a parking space; just let the car find one for you as you enter the building.
Two computers tucked into the trunk are doing all the steering, braking and acceleration by issuing commands to the drive-by-wire bus already designed into modern hybrid and electric cars.
The spinning LIDAR on the room maps the environment in 3D, including pedestrians and, in this case, traffic cones. A camera to the side of the rear view mirror tracks the road. Three radar in the front bumper and one in back also detect proximal objects. GPS, inertial sensors and wheel-speed monitors give feedback to the computer on the car's performance.
There are times when I want to drive, and times when being a driving machine feels more like the drudgery that is better left to the machines. Or when you are tired, or want a designated driver for an evening out. Commute time is a staggering collective time waster, and Ford estimates it will escalate to global gridlock in the near future. Google's Thrun estimates that autonomous vehicles could pack more closely together and achieve 2-3x throughput improvements from existing roadways, with fuel and pollution savings from the efficiency gains. And in urban environments, the largest amount of drive time is spent looking for a parking space.