Nvidia has announced the successor to its existing computer system for self-driving cars: the Drive PX 2. This new system for autonomous cars uses two Tegra processors and two Pascal GPUs.
Nvidia revealed at CES 2016 that the Drive PX 2 uses a neural network based on natural neural networks, such as the central nervous system. The Drive PX 2 is thus able to mimic the behavior of biological systems when the functions depend on a large number of variable factors. With this, Nvidia hopes that the PX 2 will eventually be able to teach itself new situations. This is partly made possible by the fact that the device has an 8-teraflop processor.
According to Nvidia, the neural network makes it possible to make computer systems learn faster, because all data is processed in the cloud. As a result, all linked systems learn from all situations that individual autonomous cars experience. Nvidia therefore claims that this system increases the safety of autonomous cars. The network is made possible by the software that Nvidia places on the systems: Nvidia Digits. Digits is the program with which the neural network is trained and developed. The program is suitable for all systems that are based on an Nvidia GPU.
The Drive PX 2 is about the size of a lunch box and uses water cooling. Nvidia indicates that several manufacturers, including Audi and BMW, are already working on the implementation of the PX 2. Only Volvo says Nvidia that it will come next year with a hundred XC90s that use the new autonomous system.