[Edaily Reporter Shin Yeong-bin ] Samsung SDS (SAMSUNG SDS CO., LTD.(018260)) is launching its “Robot Transformation (RX)” initiative in earnest, which involves the stable deployment of hundreds of robots in manufacturing facilities to enable 24-hour operations.

Bae Soon-min, Executive Vice President and Head of the RX Business Team at Samsung SDS, emphasized at “Real Summit 2026,” held at COEX in Seoul on the 8th, “While operating a single robot is relatively easy, keeping 100 robots running 24 hours a day without issues is a challenging task.” He added, “Nowadays, it is more important to establish a system that intelligently controls the entire production process rather than simply purchasing robots.”

Samsung SDS is leveraging the Manufacturing Execution System (MES) expertise and experience in equipment and logistics automation accumulated over the past 25 years as the core foundation of its RX business. Based on manufacturing site experience gained from over 1,000 production lines at Samsung affiliates and approximately 430 external client sites, the company plans to expand its existing automation capabilities to include robots and physical AI.

The company identified six key elements necessary for manufacturing site RX: △autonomy △orchestration △interoperability △resilience △safety △security. The underlying principle is that robots from various manufacturers must be able to operate in conjunction with existing equipment, and that even if a site failure occurs, the entire process must not halt but quickly return to normal.

These capabilities will be integrated into the “Robot Orchestration Platform” currently under development by Samsung SDS. This platform interfaces with the Manufacturing Execution System (MES) to assign tasks to various types of robots and consolidate results; in the event of a task failure, it analyzes error logs to identify solutions. The core of the platform lies in managing robots and production equipment from different brands within a single, unified operating system.

It also features a “Data Flywheel” system that reuses on-site data to enhance the performance of Physical AI. It is a cyclical structure in which evolved AI is reapplied to the actual production floor after passing through the “Data Factory,” which selects and processes process data, and the “Model Factory,” which manages AI models. Executive Vice President Bae explained, “Physical AI is not something that’s finished once it’s developed,” adding, “Continuous re-learning occurs through the various tasks the robots perform after being introduced into the factory.”

The process of introducing robots has also been systematized. Samsung SDS has proposed the “D-ART” methodology, which progresses from on-site diagnosis to evaluation, redesign, and transition. The strategy involves assessing the feasibility of robot application through on-site visits and, if necessary, redesigning not only the workspace and processes but also the robot’s form and end-effector (the gripper that picks up parts), thereby providing a comprehensive service encompassing implementation, evaluation, stabilization, and maintenance.

Executive Vice President Bae emphasized, “RX is not just about building a single robot; it’s about designing the manufacturing process for an entire factory,” adding, “Although robots and physical AI are not yet 100% perfected, we must start preparing now for the changes that will come within the next one to two years.”

Walden Robotics, a U.S.-based physical AI company collaborating with Samsung SDS, also participated in the event and unveiled the results of its on-site applications. Walden introduced technology that teaches robots tasks using natural language or human demonstration and supports operations via remote control when necessary.

Moira Doherty, Head of Product Management at Walden Robotics, stated, “Together with Samsung SDS, we are deploying robots that can be immediately integrated into actual production environments,” adding, “We designed the system so that even in new facilities, robot setup can be completed in about 10 minutes and actual operations can begin within an hour.”