SEOUL, South Korea — Nexen Tire Corp., has developed and implemented an artificial intelligence (AI)-based automated tire product inspection system at its manufacturing plants.
The system has been developed in a platform format, allowing application to new factories or equipment. With the introduction of this automated product inspection system, Nexen Tire, which has been expanding AI applications in the tire development process, said it has extended the scope of AI utilization to manufacturing processes.
Only tires that pass hundreds of tests during the post-production inspection process are sold, Nexen said, adding that manufacturers devote utmost efforts to the inspection processes in order to detect even minor defects that could prevent defective products from reaching the market.
Nexen Tire's AI-based automated product inspection system is applied to non-destructive inspection equipment using machine vision technology (Machine Vision, a technology that recognizes and analyzes visual information through cameras).
This includes X-ray inspection equipment for detecting structural defects and laser interferometry inspection equipment (shearography) for detecting air bubbles. The AI assists in interpreting inspection images, which previously relied on human visual assessment.
In particular, the system has achieved a defect detection reproducibility rate of up to 99.96%, Nexen claimed, noting it detects minute defects that human inspectors might overlook, thereby contributing to enhancing the quality of finished products.
Nexen said it has enhanced the system's practicality by automating the enTire process of AI training and application. To ensure the system's practicality, Nexen Tire collaborated with Neurocle Inc., known for its AutoML (machine learning automation) solutions, and PDS Solution Inc., which specializes in tire design, analysis, and data processing.
Beyond simple machine learning automation, Nexen Tire said it applied Machine Learning Operations (MLOps) technology, which optimizes and automates the enTire lifecycle of AI models — including selective data collection for AI training, AI model training, model validation, actual application and post-deployment monitoring — and successfully implemented a platform-based system, marking the first such application in the tire industry.
Nexen said this approach reduced the time it took to create a deep-learning model creation to about two days, from six to 12 months. The platform-based system also enabled immediate application to new factories or equipment. The AI trained with data from the factory where the automated inspection system was implemented aided in the early stabilization of systems introduced in other factories.
"By introducing AI technology, we have significantly improved the precision and efficiency of our tire inspection process," a representative from Nexen Tire said.
"We will continue to expand the application of AI technology to the enTire development and manufacturing processes, beyond non-destructive testing."