Tech companies adopt digital twin technology for robot training

Naver Labs’ mapping robot ‘M2’. [Courtesy of Naver Labs]

The robotics industry is increasingly adopting digital twin technology, which involves creating a virtual replica of a physical object in a digital space and thus allows for various scenarios to be simulated as well as pre-validated.

While digital twin technology has primarily been applied in manufacturing and aviation industries until now, its application is now expanding into the robotics sector.

According to information technology (IT) industry sources on Sunday, Naver Corp. and Nvidia Corp. are accelerating the development of digital twin platforms for robotic learning. The duo’s digital twin platforms are characterized by their integration of cloud computing, artificial intelligence (AI), and robotics technologies.

A key point is the rising demand for 3D simulations needed to teach and operate robots like humans, and it is anticipated that digital twin technology will be integrated with robotic operating systems. Big tech companies such as Naver, Nvidia, Amazon, and Microsoft are particularly expected to see growth in the OS market as they sell robotic hardware.

Naver has developed ARC mind, a robot-specific OS for which it has begun pilot testing, and is experimenting with the integration of digital twin and robotics technologies at its new headquarters, 1784. The company has also filed various patents related to map generation methods and remote-control systems for robots and path guidance methods using augmented reality views in 2024 to date.

“What sets Naver Labs apart from other AI and robotics companies is our commitment to proper digital twins,” Naver Labs Chief Executive Officer Seok Sang-ok said. “We are researching robots in all aspects by merging virtual and real spaces.”

Nvidia is also deeply invested in robotics with its platform, Omniverse, a key component of its strategy. Omniverse is a platform that provides essential technologies for businesses to build virtual replicas of real factories.

Nvidia Chief Executive Officer Jensen Huang noted that a simulation engine that digitally represents the world is necessary for robots to learn how to become robots. The idea is to conduct various tests on robots in virtual worlds that closely mimic reality, thus training real robot brains and potentially eliminating the need to send actual robots to distant overseas locations for training. Taiwanese global collaborative robot company Techman Robot is currently working with Nvidia on developing digital twins to improve the efficiency of robot production line inspections.

General Electric of the United States, Siemens of Germany, and Dassault Systemes of France are also diving into digital twin technology development.

According to market research firm Fortune Business Insights, the global digital twin market is projected to grow by an annual average of 39.8 percent from $12.91 billion in 2023 to $259.3 billion in 2032. Global Market Insights forecasts that the robot software market will grow from $13.5 billion in 2023 to $80 billion in 2032.

Currently, the focus is on learning from the digital twin to uncover various use cases for robot brains. Naver has released or is preparing to launch services in various fields, including augmented reality based on digital twin technology, as well as maps, real estate, and entertainment.

HL Mando also unveiled Parkie, an autonomous parking robot which lifts the vehicle, transports it, and parks it accurately at the designated spot, in December 2023. HL Mando is planning a collaboration using Kakao Mobility’s digital twin technology.

The adoption of robot automation is picking up speed in the manufacturing sector. POSCO DX is continuously developing technologies that allow for simulation and digital twin construction in a digital environment before introducing robots to its steel plant.