One of the key features of the platform is its hardware, which is powered by two Tegra X1 chips and comes with the ability to process video simultaneously from 12 different HD cameras placed around a vehicle that feed information about the environment back to the system, thereby allowing Drive PX to facilitate surround view, collision avoidance, pedestrian detection, mirror-less operation, cross-traffic monitoring and driver-state monitoring.
Nvidia is billing deep learning as the integral differentiator for the Drive PX platform, with CEO Jen-Hsun Huang stating that the system is far more advanced than conventional Advanced Driver Assistance Systems available today, most of which center around lane guidance and collision prevention. Drive PX will be able to train itself to react to different situations, and there’s the added feature that any actions learned in one vehicle are shared with the platform as a whole.
Enabling the learning experience is a deep neural network software development kit called Digits that allows the Drive PX system to understand objects in the real world and react to ever-changing scenarios. The system will be available in May to select automakers, automotive suppliers and research institutions, with the developer kit priced at $10,000.