What is a digital twin?

A digital twin is a technology that builds a virtual space that resembles the real space on a computer based on various data collected from the real space. The name was given because it is exactly like a twin.

Analyzes and simulations are performed using AI in the constructed virtual space, predicting changes that may occur in the future, and reflecting the obtained results (data) in the real world to improve products and services. You can use it for

Such simulations have been performed for a long time, but digital twins have the unique advantage of being real-time. In short, users can now get faster, more accurate and precise data than ever before.

Also, although the concept of digital twins has existed for a long time, it is rapidly being put into practical use due to the remarkable development of technologies such as wide-area data collection by IoT, detailed analysis by AI, and ultra-high-speed communication by 5G. In addition, there is a close relationship with AR, which fuses reality with real space, and VR, which creates virtual space, and the development of these will also be the key to the spread of digital twins.

Digital twins can be used in a variety of fields, including manufacturing, construction, medical care, urban development, autonomous driving, and robot control.

In addition, some governments and local governments are also working on it, such as the "Digital Twin Project" being promoted by the Tokyo Metropolitan Government. The Tokyo metropolitan government intends to improve the quality of life of Tokyo residents by grasping the situation of residents, implementing disaster countermeasures, and predicting traffic congestion.

Advantages of utilization

Leveraging a digital twin offers a number of benefits. Here we summarize the main ones.

Freedom from Physical Constraints

In the real world, we always need physical space to work and place things. In some cases, it may be necessary to handle something huge, and tests may not go as expected. But with virtual space, you don't have to worry about that. You can freely design the space and move things easily. By the way, with the opposite idea, it is also possible to set up obstacles in the virtual space and conduct experiments when they collide.

Cost reduction

For example, when developing a product, it is necessary to make many prototypes until completion, which requires a lot of costs such as time, labor costs, and material costs. On the other hand, in the virtual space, it is possible to make prototypes as many times as you want without consuming man-hours or materials, so you can reduce costs such as time and costs while reducing risks.

In addition, by acquiring data from remote factories in real time and reflecting the machine status in virtual space, it is possible to reduce transportation and maintenance costs. You can also think of how to use it. Of course, we can also expect cost reductions in various aspects other than the manufacturing industry.

Quality improvement of products and services

Prototyping and hypothesis testing in the virtual space can be repeated any number of times, making it easy to implement trial and error with the aim of improving the quality of products and services. In addition, unlike the real world, data can be visualized and analyzed, so it is useful for identifying and improving products and services while quantitatively analyzing shortcomings and defects.

Providing high-quality after-sales service

In the past, it was difficult to grasp the situation after product shipment, but it is said that this can be achieved by using digital twins. One example would be to attach sensors to the product to monitor user usage and product life.

This makes it possible to propose usage, maintenance, and support that are suitable for each user, and as a result, we can expect to improve customer satisfaction through high-quality follow-up. Furthermore, by analyzing the collected data, it is possible to develop products that better meet the needs of users and formulate marketing strategies.

Examples of Digital Twin Utilization Proposed by NVIDIA

From here, we will introduce NVIDIA 's efforts as a more concrete example of digital twins.

In addition, the following content is the lecture "Omniverse" at the seminar "Thorough explanation of NVIDIA's latest software platform-Digital twin / MLOps / AI simulation more familiar" held in Shinagawa on June​ ​7, 2023 (Wednesday) Digital twin to be realized (Speaker: Mr. Sou Takahashi, Nvidia G.K.)”.

What is NVIDIA Omniverse™?

NVIDIA Omniverse™ is a software platform provided by NVIDIA that supports the development of metaverse applications such as digital twins. By importing various data (IoT, AI, 3D, AR/VR, Robotics) owned by the user, the actual physical space can be constructed in the virtual space. It is also a platform, not just a tool, so users can customize it according to their company's challenges.

Omniverse has attracted a lot of attention from industries that require on-site verification, such as autonomous driving and automobiles. It can be used in a variety of ways, such as large-scale weather forecasting.

Source: NVIDIA Omniverse Computing Platform for Building and Operating Metaverse Applications (NVIDIA) P.4 (used on July 27, 2023)

Source: NVIDIA Omniverse Computing Platform for Building and Operating Metaverse Applications (NVIDIA) P.6 (used on July 27, 2023)

▲ How Omniverse works. By importing data from applications that handle 3D data such as CAD, BIM, and CIM, and using a function called USD (universal syndescription), you can handle it as a unified format. In addition, "Core Omniverse Technology" can physically reproduce gravity, light reflection, texture, etc. correctly.

various use cases

The images below are some of the use cases leveraging Omniverse. In this lecture, utilization in factories and distribution warehouses was mainly introduced. Mr. Takahashi said that these are unique strengths of Omniverse that can create a physically correct space. Hereafter, we will briefly describe the outline and examples by content.

Source: NVIDIA Omniverse Computing Platform for Building and Operating Metaverse Applications (NVIDIA) P.11 (used on July 27, 2023)

■ Construction design & layout
For example, when an automobile manufacturer builds a new EV factory, it was introduced that it would be possible to check whether the layout is optimal and what kind of design would maximize work efficiency. 

■ Logistics/factory/warehouse flow verification
A use case for checking the flow when machines and robots are introduced and optimizing processes such as production processes. Various other efforts are being made with digital twins.

 ■ Ergonomics work simulation
You can verify whether the space in which people work is ergonomically optimal. For example, if you don't need to bend down or if you need to raise your hips, adjust the height of the shelf.

 ■ AMRS training & deployment / Manipulator robot training
You can train and teach AMRS (automated transport robots), and create AI learning data with digital twins to improve the image recognition accuracy of manipulator (arm type) robots. This is also a recent use case that has increased significantly.

Next, I would like to introduce two specific initiatives of specific companies.

BMW builds new plant to be operational with digital twin

​BMW, an automobile manufacturer (hereafter BMW), is promoting digital twins at multiple existing factories. Two years later, Omniverse is being used to prepare a new EV manufacturing plant in Hungary (related article).

Source: NVIDIA Omniverse Computing Platform for Building and Operating Metaverse Applications (NVIDIA) P.12 (used on July 27, 2023)

At the company, employees use VR to enter a virtual space factory and check the wiring of workers and whether the layout is suitable for optimal work efficiency.

​NVIDIA also announced plans to link Omniverse with Teams (Office365) in the future so that users can converse while watching the screen of the factory running on the cloud together.

For this function, NVIDIA uses the screen that everyone can see (rather than one person's screen that is shared) as the planning data for the factory, allowing participants to arbitrarily switch to the viewpoint they want to see, and the position of the robot. We are aiming for something that can perform operations such as changing and increasing in real time.

Mr. Takahashi said that such a mechanism would facilitate consensus building among stakeholders and accelerate planning, which would be a huge advantage in utilizing digital twins.

Amazon Robotics Makes Automated Guided Robots More Efficient

​Amazon, which is famous for retail etc., owns a distribution base using robots called Amazon Robotics (hereinafter referred to as Amazon Robotics), and uses Omniverse in its fulfillment center (warehouse where packing is done for shipping). (Related article　* Displayed in English).

Source: NVIDIA Omniverse Computing Platform for Building and Operating Metaverse Applications (NVIDIA) P.14 (used July 27, 2023)

At Amazon Robotics' fulfillment centers, automated transport robots lift shelves from underneath and transport them to their proper locations. At this time, the robot reads markers such as QR codes drawn on the floor and moves.

In the past, in order to check how the marker would look from the robot, workers would take pictures directly at the site, but it seems that the lack of efficiency due to the size of the warehouse has become an issue.

Therefore, Amazon Robotics uses Omniverse to generate a large amount of learning data to improve the accuracy of image recognition, and to check whether the position of the camera attached to the robot is appropriate. As a result, development time was reduced from several months to just a few days.

In this example, the robot was used for automatic transportation, but Mr. Takahashi concluded that it can be used in the same way with surveillance cameras.

Summary

This time, starting with the basics of digital twins, we introduced case studies of NVIDIA 's advanced technological capabilities.

As you can see from the examples of BMW and Amazon Robotics, digital twins can be used in a wide variety of situations, and can be said to be a technology that contributes to significant efficiency improvements in operations and opens up new possibilities. . In addition, with the development of technology, the fields in which it can be used are also expanding.

​NVIDIA also provides NVIDIA Isaac™, a platform that supports robotics development. In cooperation with Omniverse, AI learning data acquisition, generation and simulation in robot development, AI model training and model generation, etc. can be performed, and there are software and development environments that meet a wide variety of needs.

If you have some kind of problem in your business, why don't you take this opportunity to consider whether it can be solved with a digital twin?

■NVIDIA Omniverse™
https://www.macnica.co.jp/business/semiconductor/manufacturers/nvidia/products/143448/

■NVIDIA Isaac™
https://www.macnica.co.jp/business/semiconductor/manufacturers/nvidia/products/143970/

■ NVIDIA Omniverse™ turns a distribution warehouse into a digital twin (video)
https://www.macnica.co.jp/business/semiconductor/articles/nvidia/143449/