Toyota uses Nvidia simulator for development of self-driving cars

At the March GPU Technology Conference 2019 conference in San Jose Nvidia announced cooperation with Toyota within which the Japanese car maker uses technologies of the American company for creation of self-governed machines.

Toyota began to involve the open cloud platform of Nvidia Drive Constellation representing a virtual polygon for training and testing of the systems of pilotless management. Developers provided a broad spectrum of the road simulated situations and also weather conditions, road surfacings and a land relief in the simulator. Besides, the technology can imitate limited visibility at night and dazzle from the day sun.

Toyota uses the Nvidia simulator for development of pilotless cars

This system is separated into two principal server components: Constellation Simulator and Constellation Vehicle. The first of them is based on a software platform of Nvidia Drive Sim which imitates operation of sensors, cameras and radars of pilotless cars. 

Constellation Vehicle contains the Nvidia Drive AGX Pegasus chip (includes pair of Xavier processors and graphic solutions with a performance of 320 trillion operations per second) which is completely covering all stack of the software of the vehicle. This component processes data with Constellation Simulator as though they were registered from real automobile sensors. Commands from Drive Pegasus come back to the simulator, executing a cycle of digital trunking approximately each 30 seconds.

Thanks to Nvidia Drive Constellation testing of autonomous vehicles in the photorealistic modelled version of the real world is provided. Toyota became the first car maker who began to apply this technology. Besides, the Japanese auto giant uses the graphic processors Nvidia in the prototypes of robomobiles.^[1]


Robotics

Notes