Case Study

Improving industrial processes: AI automates quality control

4min read

A large chain of industrial production comes to a sudden halt. The reason: one faulty piece in a machinery of over a thousand parts. Finding the problem costs time and money. It also leads to frustration. Why not use computers to help us solve these issues in production? That’s what MoonVision thought. We’ve developed a system that identifies flaws in a production chain before they can cause a problem, saving money and simplifying upkeep through object tracking and artificial intelligence. Our solution automatically sounds the alarm when maintenance or replacement is necessary.

After perfecting our object counting system for the food and beverage industry, our platform has now been adapted to work in industrial settings. Kamil Kula, Managing Director of MoonVision explains: “What started as a system to identify various dishes in a restaurant-setting has now been further developed. Our platform can now detect even the slightest flaw on the surface of a piece of machinery, all through Computer Vision. It’s easy to use and precise while being cost effective.”

Finding our Bearings

Recently, MoonVision began a cooperation with a large bearings producer. Bearings reduce friction and limit motion to specific parts of a machinery. Bearings guarantee safety in cars or power plants. They are constantly in use, leading to wear and tear. The upkeep of every part of the production chain is essential in industrial settings. One deficient small part, being a bearing or something else, can halt production indefinitely and increase costs. To keep the machinery running, it is necessary to constantly oversee the condition of all its parts, taking action before a problem emerges. In the worst case, the deficient piece may not be identified until flaws in quality show up further down the line, leading to larger problems than originally anticipated. The bearings producer we cooperate with is responsible for assisting all clients in the upkeep of their bearings, ensuring quality and functionality. This is where MoonVision comes in.

Our Approach

We install a camera that oversees the relevant processes and parts. We then teach our system what the part should ideally look like and it learns to identify even the slightest difference. MoonVision’s platform cannot only identify if there is a flaw in the machinery but also how large the problem is: a small scratch or a faulty piece. It provides a status update and informs the person in charge via App, saving time and money. Our technology is easy to use and reliable. We reach an almost 100% accuracy and can identify flaws no human eye would be able to spot, increasing the quality of any process. Automatically categorizing flaws means that once the information comes in via a status update, everyone can take appropriate action without consulting experts to determine the next steps: the system offers solutions.

The Patented Solution

Computer vision works similar to the human eye. Albeit fully automated and with increased precision, by default being less prone to errors. In the past, this took weeks and consisted of taking thousands of pictures to teach the system what it is supposed to do for each possible scenario.

We have revolutionized this technology and have developed a way to efficiently teach machines to see. By implementing artificial intelligence, our system detects objects via camera and through visual characteristics can deduce necessary steps in real-time. It’s cost-effective and can even be done with a cell phone camera. It all happens in real-time, meaning the system starts to learn as soon as a video feed is set up. It only takes our system one day to learn all essential information. No specific knowledge or programming skills on your part are necessary. The collected data can steer machines, and provide information, making your work easier and adding further quality controls.

Our technology has tracked and identified dishes at the Oktoberfest in Munich, has decreased the checkout time at the employees’ cafeteria at A1 and has detected containers at Audi. Through our work in industrial machinery we have now made the lives of individuals working in an additional field easier with the help of AI.