What Is a Digital Twin?

Digital twins are increasingly used for product design and engineering reviews. They accelerate virtual prototyping and design iterations, allowing designers and engineers to explore different design options without the need for expensive physical prototypes. These digital replicas of physical products are used to run complex simulations to test various scenarios, predict performance, and optimize designs.

Siemens Teamcenter X uses NVIDIA Omniverse APIs to enable designers and engineers to create immersive and photorealistic digital twins. Engineers can navigate, edit, and iterate on shared virtual models in real time, facilitating collaboration and reducing errors. With physically accurate models and real-time updates, Teamcenter X empowers users to validate designs, minimize workflow waste, and save time and costs in industrial-scale projects.

Real-time digital twins are the cutting edge of computer-aided engineering (CAE) simulation and are in high demand across the manufacturing industry, from aerospace to automotive and electronics design. They provide engineers with immediate feedback in the engineering design loop and enable them to innovate freely and rapidly explore new designs for cars, airplanes, ships, and many other products.

The NVIDIA Omniverse Blueprint for Interactive Fluid Simulation enables developers to create real-time digital twins by combining accelerated solvers, simulation AI, and virtual environments. Industry-leading software developers like Ansys, Cadence, and Siemens can use the blueprints to develop CAE software tools that enable real-time visualization and analysis of products as they develop. Luminary Cloud, a member of the NVIDIA Inception program for startups, leveraged the Omniverse Blueprint with its cloud-native, GPU-accelerated solver to realize a real-time virtual wind tunnel. Rescale is also incorporating the blueprint into its physics-AI platform, enabling real-time digital twins for industry software developers.

Auto companies, retailers, and CPG companies are developing 3D product configurators to deliver engaging experiences and content at scale using entirely digital products and environments instead of physical assets. These product digital twins can enable non-3D artists to create and customize photorealistic, personalized 3D content for marketing campaigns and reduce costs and content production times by repurposing datasets and automating repetitive tasks with generative AI.

Katana, a CGI studio, is enabling marketing teams at Nissan to create campaign assets on demand from 3D data via its user-friendly content creation application.

To take this experience to the next level, developers are taking advantage of the spatial streaming workflow guide to build solutions that stream interactive digital twins to the Apple Vision Pro, allowing consumers to enter immersive worlds and enter the vehicle in extended reality (XR). 

With NVIDIA NIM microservices USD Search and USD Code, marketing leader WPP is enabling The Coca-Cola Company to accelerate iteration on creative campaigns at a global scale.

Developers at independent software vendors (ISVs) and production services agencies—such as Accenture Song, Collective World, GRIP, Monks, and WPP—are building the next generation of content creation solutions, infused with controllable generative AI, built on OpenUSD and the 3D conditioning for precise visual generative AI Omniverse Blueprint.

Building design teams face a growing demand for efficient collaboration, faster iteration on renderings, and expectations for accurate simulation and photorealism. These demands can become even more challenging when teams are dispersed worldwide.

Digital twins enable real-time collaboration with building information modeling and non-BIM data sources at every design phase. OpenUSD allows building design teams to integrate their 3D architecture data in a digital twin, enabling users of different tools to collaborate in the same virtual environment.

Leading architecture firm Zaha Hadid Architects (ZHA) uses digital twins powered by OpenUSD to enable design teams to collaborate on complex project designs and accelerate iteration cycles.

The age of physical AI, with embodied AIs interacting with industrial facilities, is approaching. This will bring increased intelligence, automation, and autonomy to factories and warehouses worldwide. Physics-based industrial digital twins are crucial for this transformation, bridging the physical and digital worlds. These virtual facilities act as the birthplace and testing ground for intelligent facilities, robotics simulation, and multi-robot fleets.

Developers at Continental created ContiVerse, a factory planning and manufacturing operations application, on OpenUSD and NVIDIA Omniverse. The application helps Continental’s planning and operations teams optimize factory layouts and plan production processes collaboratively, leading to an expected 10% reduction in maintenance effort and downtime.

Wistron, a global electronics manufacturer, is creating factory digital twins by integrating CAD and process simulation data into a unified, physically accurate virtual environment, enhancing worker efficiency and reducing construction time by 50%. By connecting their digital twins to IoT devices at the edge, their teams also benefit from real-time monitoring of operations.

Foxconn is building digital twins of factories to optimize layouts, configurations, and equipment placement, significantly reducing the cost of physical changes and enhancing operational efficiency. This approach enables the company to train and test AI applications for robotic tasks in a virtual environment, ensuring accurate implementation and improved performance in real-world operations.

Industrial enterprises are increasingly integrating AI into their operations to deliver more automated and autonomous facilities. With these shifts, operations teams are becoming centralized in remote operations centers. These teams are increasingly embracing digital twins to monitor operations, gain deeper insights into systems and facilities, and accelerate problem identification and decision-making.

Microsoft Azure collaborated with NVIDIA to develop a reference architecture, Azure Arc Jumpstart, and a public GitHub repo to help developers build operations digital twins. Developers can leverage these resources to connect 3D models of industrial systems and production environments to real-time data from Azure IoT Operations and Power BI reports.

Autonomous machines, such as self-driving cars and warehouse robots, require vast amounts of sensor data to be adequately trained and prepared for the environments in which they operate. 

Digital twins are the birthplace of these physical AI systems. They provide a solution to the data gap that artificial intelligence developers often experience, as they can be used as a safe sandbox to generate synthetic data and train, test, and validate AI models. Amazon Robotics, for example, uses digital twins of its warehouses to simulate and optimize warehouse design and flow. The company uses these environments to generate large photorealistic synthetic datasets to accelerate training, improve the accuracy of computer vision models, and improve overall productivity. When models are deployed to the real world, warehouse robots can more effectively detect objects and navigate around the facility.

In the automotive industry, creating digital twins for simulation is crucial for training, testing, and deploying autonomous vehicles, but achieving real-world fidelity is challenging. The NVIDIA Omniverse Blueprint for AV Simulation helps address this by enabling large-scale, high-fidelity sensor simulation. With this API-based reference workflow, developers like CARLA, MathWorks, and Foretellix can deliver digital twins and render physically based sensor data for cameras, lidars, and radars, enhancing AV development.

KION Group is leveraging Mega, an NVIDIA Omniverse Blueprint for testing multi-robot fleets, to train and test its robotic agents—including intelligent cameras, forklifts, and robotic equipment—in a virtual environment before real-world deployment. By simulating warehouse operations, KION ensures seamless integration, reducing deployment risks and improving operational efficiency.

Delta Electronics, a global leader in power and thermal management technologies, uses digital twins to train computer vision and AI-assisted Automated Optical Inspection (AOI) models to quickly detect defects like missing components or misaligned screws, reducing the need for manual inspection.

Pegatron uses NVIDIA Metropolis for Factories to enhance its printed circuit board (PCB) factories with simulation, robotics, and automated production inspection, achieving 99.8% accuracy in defect detection with small datasets.

Digital twins are revolutionizing the design and operation of next-generation data centers and AI factories. With OpenUSD, engineers can integrate and visualize CAD datasets with physical accuracy and precision, allowing for simulations of aspects such as airflow and cooling systems. The use of digital twins also enables faster deployment and more efficient, accurate optimization of data center designs, significantly enhancing the planning and execution stages of data center development.

Healthcare institutions are finding compelling use cases for digital twins in areas such as surgical preparation. Surgeons can rehearse procedures using multimedia tools and then transition to highly realistic simulations with the assistance of digital twins. In neurosurgery, these digital models are customized to match the patient’s brain anatomy, enabling surgeons to practice on virtual brains that accurately replicate the patient’s specific size, shape, and lesion position.

The simulations also employ AI algorithms to suggest safe surgical pathways and predict how brain tissue would respond during the operation. Furthermore, operating room digital twins allow surgeons to immerse themselves in lifelike environments and receive feedback on their performance.

Smart cities are transforming how we live by using technology to solve complex urban challenges. By harnessing the power of video cameras, edge computing, and AI, these smart city systems gain deep insights into various aspects of urban life, including parking, traffic flow, and crime patterns. This valuable data empowers urban planners to make informed decisions, optimize city designs, and enhance the residents’ overall quality of life.

Digital twins enable the simulation of system-level behavior without abstraction, catering to the unique demands of advanced 5G and 6G networks. Their detailed 3D models accurately replicate electromagnetic propagation, facilitating the stress-testing of numerous cells with large user volumes. 

The NVIDIA Aerial Omniverse Digital Twin enables accurate simulations of 5G and 6G systems, from single towers to entire cities, incorporating software-defined RAN, user-equipment simulators, and realistic terrain properties. This allows researchers to simulate and build base-station algorithms using site-specific data and train models in real time to enhance spectral efficiency.

Digital twins are even being applied to climate modeling and energy efficiency initiatives.

NVIDIA’s Earth-2 is a climate digital twin cloud platform designed to enhance the simulation and visualization of weather and climate on a global scale. This platform is part of NVIDIA’s broader initiative to address the economic and safety impacts of extreme weather conditions, which are exacerbated by climate change.

By utilizing AI surrogates, Earth-2 enables the creation of interactive, high-resolution simulations that range from global atmospheric conditions to local weather events like typhoons and turbulence. Earth-2 enables faster and more accurate weather forecasting, which is critical for timely disaster response and planning.

Digital twins are significantly enhancing energy efficiency across various industries by enabling more precise and faster simulations and operations.

For instance, Wistron has utilized NVIDIA PhysicsNeMo and Omniverse to create digital twins that simulate airflow and temperature in testing facilities. This has reduced simulation times from hours to seconds, improving energy efficiency by up to 10% and reducing carbon emissions. Similarly, Siemens Energy is accelerating simulations of heat-recovery steam generators, reducing potential downtime and fostering greater sustainable computing practices.