omniture

Using Artificial Intelligence to Upgrade Manufacturing in the Optics Industry

TAIPEI, Nov. 7, 2022 /PRNewswire/ -- A key feature of current and future technologies, from smartphones, self-driving cars, medical testing equipment, to security monitoring, is the optical camera. It is vital that optical components are developed correctly and efficiently, which could be achieved with the use of artificial intelligence (AI). Much of these components are manufactured in Taiwan. In Taichung city, where most manufacturers are based, the annual revenue has exceeded 100 billion Taiwan dollars. Despite the established industry, there was a question of whether the optics industry in Taiwan had challenges as it grows.

Through its AI Program, Taiwan's Industrial Development Bureau (IDB) set out to determine these difficulties and to see how AI could assist and enhance the competitiveness of Taiwan's manufacturers. The project entailed visiting and interviewing 20 companies in the Taiwan Optics Association and the participation of experts and consultants to determine issues and provide AI solutions.

The most urgent problems that led to increased production costs and difficulties in customization for manufacturers occurred during production quality control (PQC). Issues stemmed from error results derived from miniscule defects and inability for easy precise focusing. To overcome these challenges, AI machine vision was proposed as the best solution if sufficient data on defects can be collected. Using the human eye for defect inspection, which some companies did, increased labor costs and a continuous need for turnover given the natural decline of one's eyesight over time. Recently, it has become more difficult to recruit for positions like these, impacting the company's ability to keep up with demand. A successful case from the AI Program involves an optics company using an AI defect detection technology and improving its defect detection rate by more than 60%. This resulted in an increase of more than 18 million Taiwan dollars in production capacity.

Another case from the AI Program improved the inspection process of another company by switching its photography method from geometrical optics to diffractive optics. With diffractive optics, AI reconstructs the images with super-high resolution and high precision to create a clearer image where defects can be more easily spotted.

After helping 10 companies implement AI technologies in their production line, the AI Program suggests the following to other companies interested in doing the same:

  1. Upgrade automatic technologies first, then collect big data on experiences.
  2. Construct the infrastructure for machine vision with sufficient optical sensors combined with the regulation and processing of mechanical control parameters.
  3. Improve digital environments to connect the entire production line.
  4. Extend smart manufacturing from AI inspection for PQC to all production lines for the overall optics industry.

The AI Program believes that more comprehensive sharing on the blueprints of how AI is being efficiently adopted in manufacturing could serve as a useful and productive tool for technology diffusion.

Source: Chung-Hua Institution for Economic Research
collection