AN INTRODUCTION ON QUALITY INSPECTION

It doesn’t matter if a company is producing boxes, cars, electrical components or raw materials; production quality defines how efficient and cost-effective a production line is. Inconsistent production quality can quickly reduce customer satisfaction and destroy a company’s reputation in almost no time at all.

Financial costs associated with preventing, detecting, and remediating product issues related to quality – technically known as quality costs  are as high as 15-20% of annual sales revenue, with a few companies reaching even 40% of total operations.

These expenses can be drastically reduced through the automation of a visual quality inspection process, implementing an AI-based system such as Konica Minolta’s Box Defect Detection, that can autonomously detect cracks, scratches, missing or incorrect components, leakages, foreign objects, and deformations.

THE LIMITS OF MANUAL VISUAL QUALITY INSPECTION

Most manufacturing and logistic companies still rely on manual quality inspection, a lengthy, time-consuming task that typically consists of a human physically inspecting the product.

The drawbacks to such an approach are evident; the main one being a general lack of consistency. In reality, an operator’s attention decreases during the day, making it harder to detect cracks and anomalies that can sometimes be as small as a hair. Moreover, negligence, or the simple fact that an operator has different perceptive skills and experience compared to another, makes traditional quality control significantly more prone to errors. According to research, visual inspection errors typically range from 20% to 30% (Stallard, MacKenzie, Peters, 2018).

To mitigate quality inspection issues, companies have adopted software-based solutions that claim to completely solve the problem. Even if, on paper, software can help to standardize and monitor the process, it must be programmed for each product every single time. This means that these solutions can’t adapt to even the smallest product change, and so require constant updates to the code, usually carried out at a high cost to the seller.

Imagine that a company that produces building materials sets out to make a smaller version of a metal fitting. If an item is extremely small, zooming in or out to categorise the defect type will affect the speed of inspection. Furthermore, the cameras used are usually not up to date, meaning that the resolution may just not be enough sufficient to spot small imperfections or a missing component.

These drawbacks are the reason why we set out to develop a better way to perform quality control, one that is both consistent and adaptable.

WHY AI-BASED QUALITY INSPECTION IS A GAME-CHANGER

A deep learning-based solution such as Konica Minolta’s Visual Quality Inspection, solves all previously encountered problems, making quality inspection really transform into a tool that helps production. The proven benefits include:

• Productivity increases, by reducing the number of faulty products and maintaining higher quality standards;
• Produces data, invaluable in order to make long term analyses;
• Reduces costs, by removing the need for new inspectors and lowering production wastes and delivery times;
• Integrates seamlessly, thanks to the ability for the algorithm to be customized and tailored to specific needs.

We’ve already discussed in our previous article on Box Defect Detection how boxes and trays are one of the fields we cover, and if you want to know more, you can watch this video.

But, thanks to its flexible design, Visual Quality Inspection is applicable to a whole range of industries, products and different types of anomalies.

Some examples include:

• Automotive, including materials and parts, resin parts and fabrics;
• Electronic parts, including electrical components and panels;
• Building materials, including wooden boards, sashes, metal fittings and tiles;
• Nonferrous and ferrous metals, including wire cables, aluminium and stainless steel;
• Raw materials, including chemical fibres, rubber, glass, paper, and pulp.

To find out how these solutions could apply to your industry, get in touch with our researchers today.