Editorials
Protecting Your Quality with AI Visual Inspection
POSTED 03/03/2022 | By: Ed Goffin, Manager, Marketing
Visual inspection is the oldest method for quality control. Humans excel at detecting cracks, deformities, subtle flaws, and missing parts. Depending on the product, we can rely on taste and smell to spot differences. We adjust for unpredictability, are easily trained, and can quickly learn by example.
However, we make mistakes when we get tired, bored, and distracted. Human manual inspection tasks can exhibit error rates of up to 30%. Often, these “errors” are actually false positives where an inspector had started to question their decision-making. Our error rate is even higher for assembly tasks.
These errors result in quality concerns, and higher costs due to waste, additional screenings and manufacturing downtime. Considering these costs, and the considerable financial, brand, and even potential consumer health risks associated with poor product quality, there is increasing demand for AI-based technologies to provide decision-support for manual tasks. In particular, these technologies are well-suited for lower volume, higher value, and customized products where fully automated inspection isn’t cost-effective or practical.
AI Brand Management for Packaging & Labelling
Dairy Distillery is a Canadian spirits manufacturer that has pioneered a unique process to produce vodka from a dairy byproduct. Brand appearance plays a significant role in consumer choice, and the manufacturer competes against larger players with much deeper marketing budgets.
In addition, operating in the food and beverage market brings another set of risks. Approximately 60% of companies in the market experience a recall. While recalls related to food quality generate media headlines, and can significantly damage a brand’s reputation, typically one-third of recalls from the US Food and Safety Inspection are related to packaging and labelling errors. These misbranded or incorrectly labelled products may not impact consumer safety, but they can result in costly shipment delays and rework for a manufacturer.
For the distillery, a main concern is maintaining a consistent brand appearance so they can ensure a premium spot on a store shelf. The distillery uses a bottle fashioned after an old fashioned milk bottle, with distinctive and eye-catching labelling. It’s what the distillery owners call a “shelf talker”. The bottle tells a bit about the dairy background of the business, with unique packaging that stands out for the consumer as they peruse the shelves at their local store. A good looking product and helps build cosnumer brand awareness and loyalty.
The bottle has three brand elements, with a main label and cap label applied by machine. A human then has to accurately place an emblem logo that visually aligns with corresponding brand elements on the other labels to ensure a consistent and appealing shelf display. With multiple products and short manufacturing runs, it’s uneconomical for the distillery to fully automate its labelling process.
Over a shift, the emblem placement would begin to shift as the operator got tired or was focusing on other tasks. Mistakes were often not noticed until the final packaging, when staff was then tasked with manually removing and replacing labels. This resulted in downtime, production delays, and additional costs. Worse, there was always the risk a poorly labelled product could reach the store shelves and negatively impact a consumer buying decision.
AI-based visual inspection provides decision-support for the operator to help eliminate labelling errors. The system includes a camera, edge processing, display panel, and pre-packaged AI plug-ins from common inspection requirements. Pre-packaged inspection skills are easily trained to verify components, inspect labelling, and check assembly, or customized for specific requirements.
Without requiring any programming skills, the distillery quality manager and operator trained the image compare plug-in to add decision-support for its labelling process. With just one image of a known good product – a “golden reference” – the system automatically identifies the key brand elements on the bottle. The plug-in has been customized to then add a graphic overlay on the visual display that highlights and guides the correct placement for the medallion label for the operator.
AI-based visual inspection ensures brand consistency and accuracy for the distillery, as well as cost-savings as labelling does not have to removed and replaced due to human error. The technology is also being used by the manufacturer as a training tool for new operators, so they can quickly understand the proper positioning of brand elements on the bottle and the difference between “good and bad” products.
With expanding production, the quality manager or operator can easily update the visual inspection system with additional “golden references” to provide labelling guidance for new bottles, labelling, and packaging. The operator simply chooses the correct plug-in for the product to be inspected. An additional image save plug-in could also be used to capture images of products at various stages of production for batch tracking. This will also provide the manufacturer with key data related to their manual assembly and inspection processes for root cause analysis and productivity management.
As the distillery adds more automation to its production, the visual inspection system can provide a valuable quality control (QC) check for in-process or finished goods to ensure all machines and humans are operating in sync. With this approach, the operator is using the AI system to inspect a number of bottles in various stages of production to ensure brand accuracy. This helps remove stressful subjective decision-making for operators, and will increase production as errors can be identified well before final packaging.
Electronics Inspection and Image Compare
One of the fastest ways manufacturers can add AI decision-support into processes is with image compare. The visual inspection system compares the manufactured product with the “golden master”, and visually highlights differences and deviations on a monitor.
DICA is an electronics contract manufacturer located just outside of Ottawa, Ontario that services an expansive list of healthcare, industrial controls, telecommunications, security, and digital imaging companies located in “Silicon Valley North”. The company specializes in high quality electronic assembly services for the small-to-medium volume market.
Serving a high-value, lower volume market can pose inspection challenges for the company, as not all products are well-suited to automated processes. As a result, a number of products are primarily inspected by human operators. The company prides itself on its exemplary record for product quality, and views the automated visual inspection system as a method to add decision-support for its inspectors.
Like the distillery, the electronics manufacturer has trained the image compare plug-in with a known, good image of a final product. Operators and quality control staff have trained multiple image compare plug-ins to inspect different products. The AI capabilities are used to match the approved layout and final production for electronic assemblies. The system quickly compares the placement of components on the circuit board, and highlights differences and deviations for the human inspection before it moves to the next step in the manufacturing process or to final packaging.
The system is also used by the manufacturer for quality checks on incoming components from suppliers. In addition, the manufacturer uses the system to capture and save an image of every printed circuit board. This data is shared with traceability systems for inventory and shipment management and batch tracking, and helps reduce root time analysis time and costs if a potential error is detected in the field.
For both the distillery and the electronics manufacturer, AI-based visual inspection systems reduce subjective decision-making for the human operator to help ensure consistency and accuracy. Errors can be detected during different phases of production, and the system can be quickly scaled for other products with no programming skills required. In lower volume or custom manufacturing applications, visual inspection systems are a key way that operators quickly and cost-effectively leverage new AI technologies to ensure quality.