[Edaily Reporter Shin Yeong-bin ] SamsungElectronics(005930)is fleshing out its vision for an “autonomous fab,” in which robots handle even non-routine tasks such as equipment inspection and maintenance at semiconductor factories. This strategy aims to increase the level of factory automation by connecting multiple robots into a single system.

According to the Korea Intellectual Property Rights Information Service (KIPRIS) of the Korean Intellectual Property Office on the 20th, SamsungElectronics disclosed a patent titled “Method for Controlling Heterogeneous Robots and System Therefor” on the 4th. The patent application was filed in February of last year.

Notably, the patent is specifically targeted at semiconductor fabs (factories). SamsungElectronics pointed out that while routine production tasks, such as wafer transfer, have already been automated, non-routine tasks—such as equipment maintenance—still require human intervention.

The company explained that as artificial intelligence (AI) and robotics technology advance, it will be necessary to integrate the control of transportation and tasks performed by autonomous mobile robots (AMRs), collaborative robots, and humanoids, and connect them to the fab’s internal systems.

In its patent, SamsungElectronics specified that a dedicated robot control system (RCS) can connect various types of robots, including collaborative robots and humanoids.

The vision is to use robots to fill the gaps in semiconductor factories where automation is already well advanced. For example, if equipment inspection is needed, the system locates the required parts and dispatches a robot, automatically handling everything from part retrieval to maintenance work.

SamsungElectronics has proposed a structure that manages different types of robots under a single overarching system. The system coordinates the movements of multiple AMRs while connecting to the Warehouse Management System (WMS), Elevator Management System (EMS), Manufacturing Execution System (MES), and chargers.

It also verifies that no “deadlocks”—situations where robots block each other’s paths—occur when they move within the same space, and adjusts work sequences and movement paths accordingly. Once a task is completed, the results are fed back into the system. The robot’s movement speed, the force applied during operations, movement patterns, and travel routes can all be adjusted.

The movements of the factory and robots are visualized using a digital twin, allowing managers to monitor them at a glance. The patent explains that this approach can enhance the level of automation and ultimately facilitate “unmanned operations of equipment.”

This patent also reveals some details regarding the specific operational methods of the autonomous fab unveiled at NVIDIA GTC 2026 last March. At that time, SamsungElectronics demonstrated a scenario in which an AMR transported the tools needed for equipment inspection, and a humanoid robot then received them to inspect the equipment.

The moves by its information technology (IT) service subsidiaries are also drawing attention. Samsung SDS (SAMSUNG SDS CO., LTD.(018260)) is pursuing a “robot orchestration” business that connects multiple robots to production equipment for integrated operation. The company also announced plans to launch its own platform next year.

SamsungElectronics’ patents also describe a structure that manages various types of robots within a single system and integrates them with factory systems such as MES and WMS, suggesting that the direction of the two technologies is quite similar.

Given that SamsungElectronics is expanding the use of robots in its semiconductor fabs while Samsung SDS is growing its platform business to integrate and manage these operations, attention is also focusing on the potential for future collaboration between the two companies.

Song Yong-ho, Head of the AI Center (Senior Vice President) at SamsungElectronics’ Device Solutions (DS) Division, emphasized the need to implement the entire fab as a digital twin during GTC last March, stating, “Agentic AI must coordinate the entire process from start to finish.”