Macnica Builds Physical AI Verification Environment to Support Robotics AI Implementation - Realizing Integrated Verification of Virtual and Real Spaces Using Collaborative Robots -

Macnica (Headquarters: Yokohama, Kanagawa Prefecture; Representative Director and President: Kazumasa Hara; hereinafter referred to as Macnica), CLAVIS Company, announced today that it has built a physical AI verification environment that uses actual collaborative robots to instantly verify the results of simulations in virtual space with robotics in real space. Based on the best practices gained here, the company will provide strong support for the implementation of robotics AI in Japan.

Physical AI opens up a new era of solving social issues
Advances in "Physical AI," which allows autonomous machines such as robots and autonomous driving cars to recognize and understand the real world and perform complex actions, are ushering in a new era of "Human with CPS," in which humans and Cyber-Physical Systems (CPS) collaborate to achieve a sophisticated fusion of real (physical) and digital (cyber) spaces, thereby resolving various social and industrial challenges. The use of digital twin and Sim2Real technologies will dramatically improve the safety and productivity of manufacturing operations, while autonomous mobile robots will significantly reduce the risk to human life in dangerous work and disaster areas. Furthermore, the realization of smart factories and smart cities will contribute to the construction of sustainable social infrastructure and improved energy efficiency.

■Barriers to implementing robotics and AI
While the fusion of AI and robotics is expected to grow, many companies are stuck at the proof-of-concept (PoC) stage and fail to move forward with full-scale implementation. Behind this lies a complex web of issues, including the difficulty of developing AI learning and simulation environments, delays in human resource development in the fields of AI and data science, and a slow pace of learning about the latest rapidly evolving technologies.
To overcome these obstacles, it is essential to combine cutting-edge AI and robotics technologies and create an environment where they can be rapidly verified. In addition, strengthening technical training that can be used in the field and forming a partner ecosystem that connects companies, research institutions, and startups will be key to promoting the transition to the implementation phase.

■ Macnica 's Initiatives
Macnica has adopted NVIDIA's "3-computer solution" (AI learning platform, virtual space simulation infrastructure, and edge AI device) in-house to develop and build a physical AI verification environment using an actual collaborative robot. This demonstration generates and expands robot learning data in a virtual space, enabling a consistent process from AI model development to actual verification, such as pick-and-place of objects in the real world. Macnica will support the entire industrial ecosystem, responsible for everything from robotics and AI development to implementation and operation, in accelerating the implementation of AI in Japan and in society.

Demo #1: Generating a training dataset in a digital twin environment
Utilizing the NVIDIA Omniverse library and NVIDIA Isaac Sim, an open-source robotics simulation framework, the system recreates robots and work environments in a virtual space. While simulating real-world work scenarios, it efficiently generates the diverse datasets required for training AI models. This makes it possible to safely acquire large amounts of data that would be difficult to collect on-site.

Demo #2: Expanding the training dataset
NVIDIA Cosmos uses the world-based model to expand existing datasets. By adding various conditions and variations, the versatility and adaptability of AI models can be improved. This enables the development of robust robotics AI that can withstand unexpected situations in real-world environments.

Demo #3: Verifying the optimal route between two points
By utilizing NVIDIA Isaac ROS and NVIDIA cuMotion, the robot can generate an optimal path between two given poses in just a few seconds using robot and obstacle information obtained from camera footage, while avoiding unknown obstacles. Simulations are performed in both virtual and real spaces to evaluate movement, accuracy, reaction speed, and other aspects. This is a technical verification that is directly linked to actual robot operation.

Demo #4: Verification of a series of work processes, such as picking and placing an object
Using the NVIDIA Isaac GR00T N1.5 Vision Language Action (VLA) model, we control and verify a series of work processes in which a robot arm picks up an object and accurately places it in a specified location. The results of the simulation in the virtual space are fed back to the real world to verify the accuracy and stability of the movement on the actual machine.

Future Prospects
In the future, we will develop joint verification opportunities with customers in Physical AI and aim to form a community network with partner companies. In addition, as NVIDIA's domestic distributor, we will support training and building of NVIDIA's cutting-edge technologies, including the Cosmos global infrastructure model. Macnica By entering the field of Physical AI, NVIDIA, Inc. will accelerate cross-industry value creation through the partner ecosystem and promote social implementation of fundamental technologies that will support the future of Japanese society.

A demonstration will be exhibited at the 2025 International Robot Exhibition, which will be held at Tokyo Big Sight from December 3, 2025.
・Venue: Tokyo Big Sight
・Date: December 3rd (Wednesday) to December 6th (Saturday), 2025
・Time: 10:00-17:00
Booth No.: W4-65
・URL: https://www.macnica.co.jp/public-relations/news/2025/148749/

[Physical AI Verification Environment Demonstration at the International Robot Exhibition]
https://www.macnica.co.jp/business/semiconductor/articles/nvidia/148716/

[For product inquiries, please click here]
Macnica CLAVIS Company Inc. Product Manager
TEL: 045-470-9821

*Company names and product names mentioned in this text are trademarks or registered trademarks of Macnica and each company.
*The information published in the news release (including product price, specifications, etc.) is current as of the date of announcement. Please note that the information may be subject to change without prior notice.