Source: Maeil Business Newspaper Date: December 15, 2024 Original article: [IR52 Jang Young-sil Award] Solar Cell Inspection System Detects Defective Panels at a Rate of One per Second — Maeil Business Newspaper Updated: December 15, 2024, 19:28 Print Edition: Page A18 Connected Insight's “Solar Cell Printer & Sorter Automatic Vision Inspection System” From left: Donghwa Na, Head of the R&D Center at Connected Insight; Jaejun Lee, Head of Development; and Yunkyoo Lee, Principal Researcher. Connected Insight's Solar Cell Printer & Sorter Automatic Vision Inspection System has been selected as the winner of the 51st Week 2024 IR52 Jang Young-sil Award. The system can analyze an image of a solar cell in just one second and automatically identify defective products. Equipped with artificial intelligence (AI), it can detect 18 types of defects, including foreign substances, cracks, and scratches on the surface of solar cells. A solar cell is a device that generates electricity through the photovoltaic effect when exposed to light. When light energy enters a semiconductor, it causes electrons to move, generating an electric current and producing electricity. Because solar cells generate energy without releasing pollutants during operation, they are considered a form of clean energy. Solar cells consist of multiple layers, including metal electrodes. A printer produces the electrode layer, while a sorter classifies manufactured solar cells as either good or defective. The Connected Insight research team developed vision inspection systems for solar cell manufacturing lines, including the printer and sorter. To improve the system's performance compared with competing products, the team incorporated AI technology. The research team explained: “Conventional inspection processes run on a CPU, while AI-based inspection runs on a GPU. We designed the vision inspection system so that the CPU and GPU can perform computations simultaneously.” This reduced the inspection time to approximately one second per solar cell. According to the company, reducing the workload on the CPU also improved the overall stability of the inspection system. The improved system stability made it possible to add four additional cameras to the original single-camera configuration. This helped improve both processing speed and image quality. The research team also explained: “With conventional systems, contamination on the lens can cause continuous false detections. Our system includes a function that automatically removes contaminants from the lens. We also developed a mechanism that effectively removes damaged products from the production line.” From Development to Commercialization The system completed a one-year mass-production validation process in 2023. Following its first mass-production order in 2024, the product generated approximately KRW 8.5 billion in revenue. The company expected to receive new orders from the United States worth approximately KRW 10 billion in 2025. It also planned to establish a subsidiary in Vietnam in 2025 to accelerate its expansion into overseas markets. Jaejun Lee, CEO of Connected Insight, said: “We successfully localized vision inspection equipment that had previously relied on German-made products. We will continue researching and advancing our AI algorithms.” Hosted by: Ministry of Science and ICT Organized by: Maeil Business Newspaper and Korea Industrial Technology Association Reporter: Jae-won Ko