Please select a language

Please select the country/region where you would like to introduce your business.

Contact Us
Contact Us

Please select a language

Please select the country/region where you would like to introduce your business.

Knowledge Nov 30, 2023 The latest examples of Manufacturing Site Digital Transformation, "Smart Factory"

KDDI America


 This month, we will introduce practical examples of the initiative to enhance production processes through the aggregation and analysis of factory data, known as "Smart Factory." Although you often come across the term "Smart Factory" in the news, you might wonder if it's a solution for large-scale factories with cutting-edge facilities. Many people may not feel it is very familiar. However, with the rapid advancement of AI technology and the widespread adoption of IoT, the barriers to implementation have decreased. As a result, there is a growing number of cases where effective implementation is carried out even in manufacturing facilities with a small number of production lines to address specific on-site challenges.

 I will introduce three examples of Smart Factory Implementation.

  • Visual inspection of parts using image recognition AI
  • Inspection of components (connector parts) using acoustic AI diagnostics
  • Abnormal assembly detection using image recognition AI (wiring assembly)

I. Visual inspection of parts using image recognition AI

[Issues Faced]

  • Significant labor, time, and cost are being expended on the visual inspection of components (one line/one person for inspection tasks)
  • Inspection accuracy is not stable (variance in skill levels among workers and human errors)

[Improvements Made with AI]

 Introducing a machine learning model-based image recognition AI for the defective product determination process. By training the model with OK/NG image data through deep learning and iteratively tuning it, the accuracy of the judgment can be improved, maintaining close to a 95% accuracy rate. Leveraging the latest AI technology, image processing tasks such as angle correction and noise reduction are performed on images captured by the software, contributing to further accuracy enhancement. This not only results in cost reduction and addressing manpower shortages but also leads to the stabilization of inspection accuracy.

    II. Inspection of components (connector parts) using acoustic AI diagnostics

    [Issues Faced]

    • The detection of half-lock issues (incomplete engagement defects) in connector components is impossible through visual inspection, and it is challenging to discover them even through electrical testing
    • As a result of the above issue, the occurrence of malfunctions becomes a significant factor (cases where issues are identified after entering the market also occur).
    • Inspection accuracy is not stable (relying on individual workers' intuition and hearing, leading to variations in judgment)

    [Improvements Made with AI]

     To detect half-lock defects in connector components, acoustic data is recorded during the assembly process. A technology is introduced to analyze and visualize the acoustic features of the clicking sound during connection. By using AI models, separating and cleansing noise such as factory ambient noise further contributes to accuracy improvement. This not only streamlines the inspection process but also enables quantitative judgments independent of workers' intuition and hearing.

      III. Abnormal assembly detection using image recognition AI (wiring assembly)

      [Issues Faced]

      • Wiring assembly tasks are complex, and mechanization is challenging, with manual work being predominant (leading to the occurrence of human errors)
      • Due to the inspection point being at the end of the production line, early detection of defective products is challenging

      [Improvements Made with AI]

       An inline AI visual inspection is implemented by installing fixed-point cameras. By creating AI models tailored to specific use cases based on image data, it becomes possible to detect anomalies in spatial placement, wiring routes, and millimeter-level deviations in wiring. This automation replaces the traditional visual inspection performed by humans, contributing to cost reduction and quality improvement by early identification and sorting of defective products.

         Thank you for reading the KDDI America Blog for this month! Digital transformation in the manufacturing industry is rapidly advancing, and leveraging AI and IoT will be crucial for maintaining competitiveness in the future. KDDI America is committed to fully supporting your daily business improvement and contribution to your core operations through the technology of digital innovation.

           Please see more details about KDDI America’s DX Solutions here: DX Solution by KDDI America

            KDDI America, Inc.

             KDDI America is the US subsidiary of KDDI Corporation, a Fortune Global 500 company and is growing communications carrier with a proven track record in Japan and a longstanding reputation for quality and reliability. KDDI America provides a wide range of High Quality Services such as Communications, Data Centers and Solution Services throughout the world.

            Inquiry for this service

             If you have any questions or inquiries, please contact us by clicking on the following link : Inquiry Form

            Writer / Interviewer

            Daisuke Mitani

            KDDI America

            Kota Nagase
            Marketing Associate

            Joined KDDI America, Inc. in January 2023 right after graduating from the University of Houston with a Master's in Marketing. Loves working out, tennis, and fashion. Won a third place in Texas Tennis State Tournament back when he was in a highschool. Always on a look out for his favorite fashion pieces.