Digital Twins are becoming increasingly popular with companies looking to simulate their environments before rolling out large-scale configuration changes. These changes include factory and warehouse designs and extend to the training of robots and autonomous vehicles via machine learning and AI. It is crucial to create 3D art assets for these simulations that represent their real-world counterparts as closely as possible to achieve high accuracy in the data collected through a simulation.
However, it is not enough just to have 3D art assets that are visually accurate. Simulation environments, such as Omniverse, can leverage additional model information and metadata to make that content far more useful for research and training of a particular product or ecosystem. In order to achieve this level of fidelity from both a visual and simulation perspective, NVIDIA is working to create a new 3D standard specification called SimReady.
At a fundamental level, you can think of SimReady assets as building blocks for industrial virtual worlds. SimReady assets are supposed to be more than just visual 3D objects. They are built on top of the Universal Scene Description (USD) platform, are designed to be modular to take advantage of the flexibility that the USD framework provides while including embedded physical properties, semantic labeling, behaviors, and connected data streams so that they are immediately useful within simulation environments.
The following guides will provide you with best practices for 3D asset creation so you understand how to create SimReady 3D models consistently. The preliminary SimReady specification will help explain how an individual asset is constructed and it will also give you an understanding of the kinds of machine learning benefits these assets can currently provide within your simulations.