cloud network hunting note: intelligent has been unstoppable trend in our life, smart cars are no longer out of reach. Tedious road driving will by advanced technology services: Nvidia powerful Drive PX technology made it all possible. Is this also means that there is no driver’s license can drive a car, think about is excited.
with the intelligent development of car, car drivers also gradually to the development of automation, so information systematic and automatic navigation devices, such as software and hardware facilities become equally important engine. This is a challenge for carmakers, for high-tech companies such as Nvidia it is indispensable to a chance.
Nvidia video game is one of the largest semiconductor company which mainly provide display chip and the mainboard chip. Now for car manufacturers to provide digital dashboard, and in-car entertainment systems, cooperation for nearly ten years. Its product prototypes mobile processors (think the car 17 inch touch screen) is Tesla Motor S, 8 million, BMW and Volkswagen, Honda car widely used.
with the development of its processor technology matures, Nvidia has entered the stage of another: the development of more and more attention to on the calculation of the processor, also includes a new type of artificial intelligence called deep learning products. Deep learning usually from huge offer speech recognition, computer data, and natural language translation services on the Internet. But Nvidia to the powerful features into the on-board computer. The product is to Drive the PX “a powerful Drive on-board computer”, as long as developers, researchers, manufacturers are willing to, only need little money it can be completed, human do not need to be in the back of the car steering wheel operation idea implementation.
shortly in San Jose (San Jose) developers conference, CEO jen-hsun huang used from YouTube about a talented baby is a very popular video, to show the need how many machines can “learn” to self-driving car.
he asked, “how do you teach a baby to play table tennis?” . So, if we apply the same way to a car with automatic navigation system and automatic braking system advanced auxiliary driving system, it could be like this: first of all, you will have to teach children how to recognize a table tennis; Then, step by step to help him understand the matter; Finally, to see how they would react.
“we will teach baby to Newtonian physics,” he said.
talk easy! Huang said the teaching method of “first, and then, the second” for children or the computer is too unreliable. Especially for automatic car need to transport a large number of digital information more of this kind of circumstance is not applicable. Because you can not plan those uncertain factors.
to teach children to play table tennis, a better, more direct way is to let him move, and then teach them sum up experience from practice. Also the most suitable for self-driving cars in the same way. “Train the baby to play table tennis and teach a car drive can have in common.” He said.
the idea of this Project is Project DAVE by DARPA deep learning on the basis of autonomous development system, its working principle similar to the human brain. Dave through video collection situation around, and then like a man control all the obstacles, learn to predict the best lines. Because to see to understand the situation around, so it can better determine the need to go beyond, the need to avoid.
“Drive PX 3000 times faster than its predecessor DARPA” Nvidia said. Two prototypes X1 processor, make it has 2.3 trillion floating point arithmetic processing capacity, more than the world’s best super computer data provided by 15 years ago. It can be to 12 cameras at the same time (or camera, the combination of radar and laser radar) to process the input. “In this small motherboard have a supercomputer, which is currently the world’s most advanced computing platform [the car]”
starting in May, only need $10000, developers, researchers, and automobile manufacturers can get Drive PX. It comes with the software, including computer vision algorithm optimized the prototypes of the processor. Its purpose is to help users to develop their own self-drive platform. Depth study on this perfect computer may not be the most effective, when a lot of vehicles into the streets, then need to deal with a lot of data, (whether a car or all car has the function of the deep learning) deep study on the sui generis computers may not be the most effective), but it is still an improvement.
auto marketing, senior director of Danny Shapiro said: “for carmakers, Drive the PX for establishing the automated driving may be a better development platform. This is Nvidia next generation of technology is the audi called the autopilot function integration in a zentrale Fahrerassistenzsteuerger? T (abbreviated as zFAS) “the central brain” in a computer, store the autopilot function of car all here. Audi electronics leader Ricky Hudi said: “the car makers think DRIVE PX as further research and technology development is an important tool.”
“for those who are thrilled to have a own self-driving vehicles do not need to worry about the $10000 price tag. If all the auto makers decided to provide consumers with Drive PX, or its derivatives of auto technology, the price of each car will be very affordable.” Shapiro said.
to learn more knowledge, entrepreneurial innovation quickly add hunting cloud network micro letter public account: