Nvidia’s Omniverse is accelerating AV simulation tests

The metaverse, defined by Nvidia as the 3D internet, is making waves across all industries—particularly automotive. Metaverse applications are diverse: from developers training artificial intelligence (AI) systems in virtual worlds to enterprises building digital twin simulations of their industrial processes.

However, Nvidia is proposing the Omniverse as a space for industry operations. The Omniverse is a metaverse populated by industrial digital twins and virtual autonomous robots, allowing a company to create 3D models and simulations in the virtual space with all the constraints and details of the real world. Specifically, the Omniverse highlights the company’s approach to integrating industrial digital twins at scale using the universal scene description (USD) file format. USD, originally invented by Pixar Animation Studios, is an open and extensible ecosystem for describing, composing, simulating, and collaborating within 3D worlds.

Thanks to its design and features, USD is poised to be the open standard that enables the 3D evolution of the internet. “The Omniverse is a platform that we’ve created as a method for bringing people from all over the world together to work on a specific project,” says Danny Shapiro, Vice President of Automotive at Nvidia. “It allows people who use particular 3D software applications to keep using them within a larger, virtual space.”

This platform is facilitating the growth of autonomous vehicles (AVs) through its ability to produce lifelike simulations. However, doing so poses two challenges: generating a world with enough detail and realism to facilitate a realistic response from an AI driver, as well as creating simulations complex enough to cover all the scenarios in which the AI needs to be fully trained and tested.