Artificial intelligence (AI) is evolving from merely assisting engineers to making decisions about the movement of production equipment, a development referred to as 'Physical AI.' Siemens is broadening the application of AI in manufacturing by integrating digital twins, industrial AI, and automation technologies. The company is expanding its collaborations with domestic firms in key industries such as semiconductors, shipbuilding, and biotechnology.
In an interview with Aju Economy on September 16, Jeong Ha-jung, CEO of Siemens Korea, noted, "Even global experts did not anticipate such a rapid transition from agentic AI to Physical AI," indicating that the shift in South Korea's manufacturing sector towards AI is occurring much faster than expected.
Changes are already evident in the engineering field. A prime example is the Siemens Industrial Copilot, developed in collaboration with Microsoft, which utilizes generative AI. Previously, engineers had to write programmable logic controller (PLC) programs and design human-machine interface (HMI) screens manually. Now, they can input requirements in natural language, allowing AI to suggest program code and generate drafts for screen layouts.
Jeong explained, "In the past, changing the factory control screen required rewriting the program, but now you can simply say, 'Put temperature and humidity on the left and create a change graph on the right,' and the AI will adjust the screen accordingly." He added that while natural language-based technology is still new and has not yet been widely applied in domestic manufacturing, its adoption is expected to accelerate significantly.
The role of AI is expanding from assisting in program writing to determining the actual actions of machines. Jeong cited a robotic arm showcased at the Hannover Messe as an example. When instructed to place items of varying shapes and materials, such as shoes, hats, and scarves, into a bag, the AI-guided robotic arm recognizes the outcome and adjusts its next actions accordingly.
He described a scenario where the robotic arm placed a scarf into the bag, but half of it spilled out. The AI then determined there was an issue with the task outcome. "While traditional manufacturing focused on repeating standardized tasks, AI can now assess situations and results to modify its approach," he said.
Automated guided vehicles (AGVs) are also moving beyond fixed routes by integrating AI. Jeong explained, "If you instruct an AGV to transport a 10 kg item to a specific location, it can recognize obstacles along the way and navigate around them to reach the destination," highlighting how AI enables these vehicles to make decisions based on their surroundings.
Supporting this Physical AI is another key component: the 'digital twin.' This technology creates a virtual representation of actual products and production processes, allowing for pre-validation and optimization of design and production conditions. Siemens is developing an industrial foundation model (IFM) that understands the 'engineering language' of industrial sites, linking digital twins with the Industrial Copilot.
The company is broadening its industrial collaboration scope in South Korea, recently signing a memorandum of understanding (MOU) with HD Hyundai to explore cooperative opportunities. Jeong emphasized the significant value of digital twins in shipbuilding, where vessels can contain up to a million components, making their structures complex. He noted that traditionally, verifying whether designs functioned correctly required physical testing, which incurred substantial costs. "If we can simulate in a digital environment, we can validate various conditions before actual production," he said, explaining that this approach can significantly reduce cost burdens.
Siemens is also pursuing collaboration with Naver Cloud to combine AI and cloud technologies with Siemens' industrial expertise. Jeong stressed that the transition to AI in manufacturing cannot be achieved by a single company alone, highlighting the need for collaboration across various sectors.
He stated, "Digital transformation and AI cannot be accomplished by one company alone. Siemens also requires collaboration with various partners, making alliances crucial." He noted that Naver Cloud has strengths in AI and cloud technology, which can synergize with Siemens' industrial capabilities.
Jeong also emphasized that manufacturing AI should enhance product quality and price competitiveness. "The competitiveness of manufacturing ultimately lies in supplying good products at low prices," he said, adding that products lacking quality, even if inexpensive, are meaningless, and conversely, high-quality products that are excessively priced struggle to compete. He concluded, "Ultimately, we must consider how to produce good products at lower costs."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.