Robots Roam the Cathode Plant: EcoPro BM's AI Experiment

Inside EcoPro BM's Pohang Plant
AI-Driven Quality Prediction and AMR-Based Equipment Inspection
"Outpacing China Through Manufacturing Innovation"

On June 11, during a visit to the EcoPro BM production plant in Pohang, Gyeongbuk, the scene was dominated by massive pipes and equipment stretching up to the ceiling as soon as one entered the production building in cleanroom attire. Although the towering production facilities—reaching several dozen meters in height—were operating nonstop, there were not many workers present. Instead, small Autonomous Mobile Robots (AMRs) were moving between the production lines.

The application of AI observed in the battery plant was somewhat different from what is typically seen in steel mills or shipyards. In the cathode material production site, data played a more important role than robots. While AMRs inspected the equipment, AI was used to predict product quality.

On the 11th, an autonomous mobile robot (AMR) is performing production facility inspection tasks at EcoProBM Pohang Campus in Pohang, Gyeongbuk. Photo by Joint Press Corps

The most expensive aspect in a battery plant is neither the equipment nor the raw materials. It is defective products. Even a minor error in the cathode material production process can cause hundreds of kilograms of products to fall out of specification at once. This is because numerous variables—such as temperature, pressure, input amount of raw materials, and firing time—affect quality. Reducing the defect rate by just 1% can have a significant impact on profitability. This is why EcoPro BM has accelerated efforts to build an "AI factory" that analyzes the vast amounts of data generated in the production process to predict quality and detect anomalies in advance.

Song Hojun, CEO of EcoPro BM, told reporters that day, "The rechargeable battery industry is not one led by a handful of genius scientists or engineers. It is an industry in which countless material and process developers spend years finding optimal conditions through trial and error." He continued, "While we once enjoyed world-class competitiveness, in recent years we have faced aggressive pursuit from China. In the end, we must enhance our competitiveness through manufacturing and production innovation."

In fact, EcoPro BM considers the data generated at the plant to be a core asset for AI competitiveness. Each day, an average of 250,000 cases of MES (Manufacturing Execution System) data and as many as 470 million pieces of equipment sensor data are generated in the factory. In the past, this data was scattered across individual systems and not fully utilized, but recently, it has been integrated and leveraged for AI learning. The accumulated data so far exceeds 20TB.

CEO Song explained, "Data generated in the manufacturing process is essentially the company's core asset. Training AI to learn what conditions lead to high quality and which situations cause problems, using this data, is the starting point for an AI factory."

The AMR that drew attention at the scene is also part of this broader trend. EcoPro BM is working to automate equipment anomaly inspections using AMRs. Rather than having workers walk the shop floor to manually check equipment status, robots equipped with autonomous navigation are now performing inspection tasks. The company aims to enhance safety and improve equipment management efficiency in the production environment through this approach.

Song Hojun, CEO of EcoPro BM, is explaining the current status of AI-based autonomous manufacturing promotion and plans to strengthen competitiveness in the battery materials industry at the Manufacturing AI Transformation (M.AX) on-site meeting held on the 11th at EcoPro BM Pohang Campus in Pohang, Gyeongbuk. Photo by Joint Press Corps

The role of AI is not limited to equipment management. A key initiative at EcoPro BM is "quality prediction." Instead of inspecting products after production, the company analyzes data generated during the production process to predict quality. Currently, EcoPro BM is developing AI models with a target quality prediction accuracy of over 95%.

Traditional manufacturing inspected product quality after production. If a problem was found, the process was modified to address the cause. In contrast, AI identifies abnormal signals during production. Rather than merely identifying defective products, it anticipates the likelihood of defects before they occur.

Based on these efforts, EcoPro BM is also advancing a "Dark Factory" initiative. The company aims to implement AI across all areas—production, quality, equipment, and safety/environment—to achieve a fully autonomous factory. The goal is to reduce manufacturing and processing costs by 30% and automate office work by 50%.

The battery industry has faced aggressive competition from China in recent years. With production capacity alone no longer sufficient for competitiveness, ensuring quality and productivity has become the top priority. This is the rationale behind EcoPro BM's focus on AI—not simply as a means to automate human tasks, but as a tool for manufacturing innovation that raises both quality and yield.

CEO Song stated, "Our goal is to create an autonomous operating system by applying AI across all aspects of production, quality, equipment, and safety/environment. We will establish a data-driven system that manages everything from raw material and component management to quality control and equipment inspection."

This content was produced with the assistance of AI translation services.