TEXTILES.ORG
AI offers a fast turnaround on tasks that humans may find tedious, but they’re also necessary. The textile industry is discovering new and innovative ways to make AI and other advanced technologies do the time-consuming work and provide needed data, so manufacturers can use their employee talent to oversee processes and make the decisions.
AI tools to diagnose dye issues
Yuan Sun, Ph.D. candidate in the Color Science and Imaging Laboratories at NC State, has created an AI tool that can detect defects in a dye process and prevent stains on a fabric that can cost time and money.
But just one defect can have several possible causes – the material itself, an equipment malfunction, or a glitch in the process. Finding the problem and correcting it could take a coloration expert, but job availability for color expertise in U.S. textile finishing and coating has dropped 79 percent in the last 25 years.
In his Emerging Textiles (ET) Conference presentation, “Don’t let your color expertise retire: Building an AI tool that learns and diagnoses in the dye house,” Sun offers another expert called DEXPERT, a process for five coloration systems to diagnose and trouble shoot coloration issues:
- DEXPERT for cotton
- DEXPERT-P for polyester
- DEXPERT-PT for protein fibers
- DEXPERT-B for polyester/cellulosic blends
- INKJET EXPERT, cotton substrate
This system incorporates AI to help the users get answers in a way that’s most useful for them, he says, with a “large language model” (LLM) and using AI to draw on several knowledge bases to quickly resolve an issue or prevent one.
These “expert systems” are programs that apply expert knowledge and explicit rules in a specific domain. “Expert systems keep diagnostic expertise available; LLMs keep information updated and connected; agents speed up workflows; and production cases, data, and expert feedback continually help to improve the system,” his presentation explains.
“AI handles the repetitive tasks,” he says. “Real people make the decisions.”
AI in quality inspection: A case study
Arnold Kravitz, chief innovation officer at O-HIVE, an Austin, Texas, industrial automation startup, spoke at Advanced Textiles Expo about his company’s spatial vision intelligence — such as for fault detection in textiles or safety hazards in a plant.
He detailed a case study where O-HIVE’s robots and AI platform are helping identify
defects on items for a uniform rental company. He explained how the system was trained on images of pinholes, stains, popped seams, missing buttons and other issues and showed attendees the program’s reporting. Being able to kick out a segment of the products from inventory automatically allows the staff members to focus their efforts on pieces that take more time and judgement, rather than the obvious.
During the question-and-answer portion of the program, the discussion turned to quality inspection of large pool covers, such as for missed stitches, flaws or grommet issues. Kravitz noted that the cameras don’t have to be on robot arms on gantries but could be on any headset for employees to walk around the perimeter of the cover, doing those final quality checks. They can also be incorporated into companies’ existing machinery, such as knitting or embroidery machines.
“I’m a big believer in workforce multipliers,” he says. For areas where people are the best judge and decisionmakers, people should stay in those roles. “[Human judgement,] that’s your secret sauce,” he says. Areas where the work is backbreaking, tedious or has high turnover — that’s where automation can help a business.
“The stupidest things that [companies investing in automation] can do is lay off their best employees,” he says, noting that after volume of work goes up, they can’t get those people back. They should retain employees post-automation and better utilize their talents.