Nvidia’s new driverless car AI is palm-sized and energy efficient
Nvidia’s new driverless car AI is palm-sized and energy efficient
Today, Nvidia unveiled their new Nvidia Drive PX 2 computer. It’s an AI computing platform built to accelerate driverless car technology, and it will be the brain behind Baidu’s self-driving cloud-to-car system.
The new palm-sized, single-processor configuration found on the Drive PX 2 consumes only 10W of power. It boasts highway automated driving and HD mapping, and it’s capable of using deep neural networks to process data from multiple cameras and sensors. In totality, it combines deep learning, sensor fusion, and surrounding vision for a different driving experience.
Drive PX 2 is also scalable, and it ranges from a single passively cooled mobile processor operating at 10W, up to a multi-chip configuration with two mobile processors and two discrete GPUs that can put out 24 trillion deep-learning operations per second. Multiple Drive PX 2 platforms can be used in tandem to enable fully-autonomous driving.
The new AI for driverless cars will succeed the old Drive PX, which was already in use by 80 automakers, “tier 1 suppliers”, startups, and research institutions that develop autonomous vehicle tech. The Drive PX 2’s new configuration and architecture will enable Nvidia’s partners to quicken the production of such driverless vehicles.
You can find the technical specifications of the Drive PX 2 here. It will be available to production partners in Q4 2016.
Source: Nvidia