FREMONT, Calif. — The data storage company Seagate is being recognized for its manufacturing operations.
Seagate won a 2022 Manufacturing Leadership Award for its “outstanding innovation” in smart manufacturing from the Manufacturing Leadership Council (MLC) at the National Association of Manufacturers (NAM), according to the company last month.
The winning project provides a foundation for high-volume artificial intelligence (AI) analytics in a factory. The technology strategy — called optical inspection with centralized analysis (OPICA) — supports analytics at the factory edge. It is no-code software with deep learning (DL) that identifies defects that can occur during production and prevents them from escaping downstream.
Inspection accuracy improved by at least 20%, and seven automated machines are needed to do the work of 150 operators, using manual microscopes to inspect millions of parts daily, according to Seagate. The company says the implementation generated a return on investment (ROI) of up to 300%.
Seagate believes expertise in development of mathematical computer vision algorithms is “no longer a requirement.” Instead, specialists who work in preventing defects can take images of their processes and devices, mark them up, and use the no-code software to “train the model to identify contamination.”
The software uses AI as an “augmentation of human work, not its replacement.”
The OPICA solution handles three million multi-class deep learning inferences per day and was built using modular and scalable data engineering technologies: such as as Docker containers, Kubernetes container orchestration, RabbitMQ data messaging, edge line servers, human-in-the-loop model management dashboards, and high-speed Seagate storage.
“The judges who evaluated this entry noted that it pushes the envelope on inspection and demonstrated an advanced use of technology with strong, proven results,” said Penelope Brown, senior content director, MLC.
Brown said Seagate’s solution is an example of “what’s possible in manufacturing’s emerging digital era.”
Gary Kunkel, an engineering director who helped develop Seagate’s solution, said, “Computer vision challenges are common in manufacturing.”
“Historically, solutions would be developed by specialized resources such as computer vision engineers,” Kunkel said. “Early on, it became apparent that with deep learning techniques, we could develop no-code tools to train neural networks to identify defects and features in images.”