Graphics cards eye self-driving cars, deep learning apps(2)

Graphics cards eye self-driving cars, deep learning apps

The Drive PX cards aim high, supporting 12 camera inputs. Car makers may simplify and lower the cost of the technology in production vehicles. "We can do a lot more than the simple chips in the market today, so we will see augmented ADAS and a leapfrog [technology] in the future," Shapiro said.

Tesla aims to drive self-driving cars

According to Elon Musk, founder of Tesla Motors, "Autonomous driving is much safer than people think. We used to have elevator operators, then we developed simple circuitry for elevators to go to the right platform automatically.

Nvidia showed its GPUs learning to recognise vehicle types.

In an on stage interview with Nvidia's Huang, Elon Musk, founder of Tesla Motors, said he wants his company to lead the charge to self-driving cars.

"Someday cars will be like that. You will be able to tell your car 'take me home.' In the distant future, people may outlaw driving cars because it's too dangerous. You can't have a person driving a two-ton death machine.

"With two billion cars on the road and 100 million more made a year, it could take 20 years to replace the fleet. So the transition will take quite a while, and it's the same for the electrification of cars.

"Radar and camera fusion tracks the car ahead now...even with that sensor suite we can make great progress. We can make the car steer and do lane changes autonomously...but if there are no lane markings, or children playing, you need a bigger sensor suite and more computing power.

"Right now it's fairly easy to do sub 5-10 miles per hour with ultrasound, but at 10-50mph...in complex environments you get unexpected things happening, manhole covers may be open, there are bikes...but in the freeway environment the set of possibilities is much reduced. So the intermediate speeds are hardest. Still, I almost see it as a solved problem, we know what to do and how to get there.