Elementary rolls out AI-powered vision systems for quality inspections
GP's existing Cognex vision system was unable to execute the type of inspection that they needed. Elementary proposed a solution that would inspect individual tissue rolls before they have been wrapped, and then inspect the complete package as it comes off the line. For tissue rolls, a multi-camera setup, with one camera and built-in light per roll, to inspect each unit individually as it passes through on the conveyor system. If any rolls in the wrong orientation were detected, Elementary’s inspection system would signal to the production control to stop the conveyor. Operators would then remove any defective products, inspect for issues that might be causing irregularities, and restart/reset the conveyor to resume production when problems are fixed.
For wrapped packages, Elementary used a three-camera system with side and top-down lights and cameras to detect anomalies and communicate them to the reject pusher, which automatically pushed defective packages from the line. If five consecutive non-conforming packages were detected, the system sounded an audible alarm to alert operators.
Going beyond immediate problem solving
The power of AI lies not just in its ability to identify existing problems, but in solving broader issues that can help minimize defect occurrences and improve quality and product yield down the line. Using Elementary's Root Cause Analysis feature, GP was now able to find defects in individual parts and components and prevent them from impacting the finished products' quality. This proactive approach allowed GP to address potential issues before they escalated into larger problems.
Elementary's Quality as a Service (QaaS) model ensures that the vendor handles all tasks associated with the creation, maintenance, and management of AI models, adjusting to any specification changes, process modifications, or new product introductions, while Elementary continuously monitors for and fixes issues remotely.
GP also implemented Elementary’s Daily Reports functionality to review all defects reported by the automated visual inspection system, evaluate the day’s operations, and compare yield and quality across lines. The daily report proved particularly useful when a production manager noticed that a new SKU line was lagging in performance compared to other products. Assessing defect images stored on the Elementary cloud revealed that a decreased yield was caused by a particular recurring defect. A technician was able to quickly find a cause of the defect and remedy it, increasing yield by 27% overnight.
.svg){kind=icon}
