Nvidia Corp. continues to push the boundaries of the industrial metaverse, a virtual space parallel to the real, that simulates detailed physical assets in the virtual world, with its hyper-realistic physics simulation and collaboration product Omniverse.
Nvidia announced today during the company’s GTC 2022 virtual developer conference the launch of Omniverse Cloud, a comprehensive cloud-based software-as-a-service solution for artists, developers, and enterprise teams to design, publish, and operate Omniverse used by Metaverse applications anywhere in the world.
Omniverse is a real-time collaboration and simulation platform that enables realistic replication of the world at scale. Using the platform, teams of designers and engineers can replicate and simulate cars, airplanes, buildings, factories and more. All parts of an engine or factory can be designed with fully simulated physics in real time just like in the real world to react as if they would react in real situations.
These are known as “digital twins,” the fully virtual twins of real-world spaces and objects that are part of the industrial metaverse. As a result, they can be repeated, modified, and experimented with to understand what would happen before expensive changes have to be made in the real world.
With the Omniverse Cloud, individuals and teams can collaboratively design workflows and collaborate without the need for on-premises computing power.
“The Metaverse, the 3D Internet, connects 3D virtual worlds described in Universal Scene Description and viewed through a simulation engine,” said Jensen Huang, Founder and CEO of Nvidia. “With Omniverse in the cloud, we can connect teams worldwide to design, build and operate virtual worlds and digital twins.”
Omniverse Cloud runs on a purpose-built cloud computing architecture in Nvidia’s data centers and on hardware running the Nvidia OVX architecture for graphics and simulation and Nvidia HGX servers for advanced artificial intelligence workloads. It uses the Nvidia Graphics Delivery Network, a globally distributed data center network to deliver low-latency Metaverse content, which the company has learned from its experience with GeForce Now, its cloud-based low-latency video game streaming service.
Deploy digital twins in the real world
The power of digital twins is that they allow the simulation of the real world in the metaverse to be as accurate as having the benefits of a virtual world for testing and visualization combined with the finality of the physical world.
Retail store Lowe’s has deployed Nvidia Omniverse Enterprise and augmented reality glasses to its employees, essentially giving them x-ray vision. For example, to read a small label on a box at the top, employees have to climb a ladder to see what is available. With a digital twin and AR glasses, they could simply look up and the glasses could reveal the data of the object that should be in that position.
Employees can also see through the glasses that shelves are correctly stocked by comparing the AR holograms of the digital twins with what is currently on the shelves – for example if the wrong products are on the shelves or not enough. Then they can correct it.
“With the Nvidia Omniverse, we’re bringing data together in unprecedented ways and giving our people superpowers,” said Seemantini Godbole, chief digital and information officer at Lowe’s.
For store management, it goes even further: With the help of digital twins and AI store planners, the customer experience can be optimized by examining which products are bought by shoppers in combination with each other. By then working to place those products closer together by examining 3D heatmaps of foot traffic through customers, associates can move products closer together to reduce the number of steps customers have to take to pick them up.
Nvidia also announced a collaboration with Digitale Schiene Deutschland, the digital division of German rail operator Deutsche Bahn, which began expanding the network’s capacity without building new tracks. The effort required a system to safely automate trains with less headway between each other and will involve building the first nationwide digital twin simulation of an entire track network.
Since it’s about creating a photorealistic and physically accurate simulation of the entire track system, it’s going to be a massive undertaking. It will also include routes passing through cities and countryside combined with data from many sources including platform measurements and vehicle sensors.
Using a digital twin of the entire network built into Omniverse, running concurrently alongside the actual rail network and fed the same data in real-time, it can use AI to monitor and simulate sensors and other data, and predict incident prevention.
“With Nvidia technologies, we are able to realize the vision of a fully automated train network,” said Ruben Schilling of the Lead Perception Group at DB Netz, part of Deutsche Bahn.