Nvidia reveals AI for Baidu’s self-driving cars

Nvidia revealed a new configuration of its DRIVE PX 2 in-car autonomous driving computing platform on Wednesday, September 14 which vastly improves power efficiency and drops physical size compared to the original unveiled at CES earlier this year. It makes it possible to process visual and sensor data from self-driving car-mounted monitoring systems in real-time on board a vehicle, which is crucial for autonomy because it has to be able to act without a connection to a powerful remote server. This capability, allows Nvidia’s AI solution to process inputs from multiple cameras, plus lidar, radar and ultrasonic sensors.

Drive PX 2 can also receive over-the-air updates, similar to what is in Tesla’s self-driving cars. That system uses Nvidia’s Tegra “Parker” processor, which includes two of the vendor’s “Denver 2.0” CPUs and two Pascal-based GPUs created to power deep-learning applications that will make cars smart enough to recognize and respond to obstacles.

Nvidia says more than 80 carmakers, suppliers, startups and research organizations working on autonomous vehicles are using its Drive PX architecture.

It uses 10 watts of power and is mostly looking at providing AutoCruise functions including highway automated driving and HD mapping.

AvailabilityThe new single-processor DRIVE PX 2 will be available to production partners in the fourth quarter of 2016.

The new DRIVE PX2 AI computer is chiefly aimed at automakers that are interested in developing fully or partially self-driving vehicles, which would typically require human intervention. The new version of the platform will make it easier for auto makers to incorporate the Nvidia technology into their designs, according to Rob Csonger, vice president and general manager of Nvidia’s automotive business unit. It is now available from a single mobile processor configuration (as above), to a combination of two mobile processors and two discrete GPUs, to multiple PX 2s. It supports automotive inputs/outputs, including ethernet, CAN and Flexray. It targets the world’s most demanding users – gamers, designers and scientists – with products, services and software that power fantastic experiences in virtual reality, artificial intelligence, professional visualization and autonomous cars.