Chris Urmson: How a driverless car sees the road

Chris Urmson is the director of self-driving cars at Google, in this talk, he spoke about why Google chose a fully self-driving system, instead of driver assistance systems. He also demonstrated how Google's driverless car maneuver in real traffic.

In the beginning of the talk, Urmson talked about the death toll in traffic accidents. 1.2 million people are killed on the world's roads every year. 33,000 in the U.S. each year. Why we have this terrified number? Urmoson believed that is because humans are the least reliable part of the car. For example, in one test drive experiment conducted by Google, one driver who was supposed to be paying attention during the experiment. But he noticed his phone was going to out of battery. So he started to find his laptop and charger in the back seat while the car was going on a freeway with 65 miles per hour. Another great advantage of driveless car is that humans can save a lot of time at the wheel. Nowadays, the traffic is getting worse in the U.S., so people spend a lot of time on driving. Urmson said there are 6 billion minutes spent on commuting every day. That turns out to be 162 lifetimes spent per day on driving cars.

On the other hand, Google's driverless car learns to detect and navigate through the real traffic. Urmson demoed videos in the TED conference. We can see how Google's driverless car maneuvers the unpredictable traffic, such as a car making a U-turn, a cyclist running a red light. Even a woman in an electric wheelchair chasing a duck on the road can be detected by Google's self driving system. Urmson believed driverless car can outperform human beings for many reasons, one reason he mentioned in the talk is that driverless car can use laser to detect objects from a blind spot.

My opinion:
If we just talked about the self-driving system itself without other obstacles such as privacy, security issue. I believed the trend of self-driving system will be inevitable. Self-driving software could gain driving experience as it is tested as a human driver does. The major difference is that everything it learns can be shared to every other car with the same software. Self-driving system also don't get tired or distracted by others. And self-driving system could be easily applied to a new business model, so called "Shared economy".