IBM applies AI to factory QA
- 26 April, 2017 02:41
IBM's Watson artificial intelligence technology has found plenty of white-collar work in places like hospitals and banks, but soon it will be off to get its hands dirty on the factory floor.
Working with ABB, a maker of industrial plant, IBM has developed a new AI assistant to help factory workers spot manufacturing defects on the production line.
Connected to an existing industrial monitoring system, ABB Ability, it will help manufacturers improve speed, yield, and uptime, according to ABB.
The Cognitive Visual Inspection system, as IBM calls it, pipes images from a UHD (ultra-high-definition) camera to an instance of IBM's Watson software that has been trained to detect and classify production faults in real time.
Watson can inspect parts up to five times faster than production line workers and even detect faults not visible to the human eye, according to IBM. Learning fromÂ human classification of defects in images, Watson is able to spot scratches and pinholes.
By using computers to support human quality assurance inspectors, manufacturers can potentially inspect every product coming off the line, rather than random samples, allowing them to reject or rework only faulty products rather than complete batches. In that way, they can increase not only production speed but also production yield.
ABB also hopes to put Watson to work in another of its markets, electricity generation, to predict demand for electricity based on weather patterns.
The ability of software like Watson to learn to identify patterns that it previously took an experienced human to recognize is allowing computers to support staff in a variety of fields.
IBM hopes to have Watson supporting or second-guessing many of our decisions for us over the next five years.