FAQs

What is AI?

Artificial Intelligence (AI) is a branch of computer science that involves creating smart machines capable of performing tasks that typically require human intelligence. These machines are usually purpose-built and are significantly faster and more reliable than humans at their specific task.

How can I use AI in manufacturing?

Every aspect of manufacturing stands to gain from AI. In manufacturing, AI, particularly in visual inspection and predictive maintenance, enhances efficiency by automating quality checks, identifying defects, and predicting when equipment maintenance is needed. AI also has applications in predictive analytics and human-machine interface simplification.

How does AI improve visual inspection in manufacturing?

AI algorithms can be trained to see similar to like humans learn to see and identify objects. Unlike traditional quality control where feature-engineering was used to define defective units, AI algorithms learn to recognize them by being shown examples. Instead of identifying detailed and specific measurements and programming your system to make the pass/fail decision based on these measurements, the algorithm learns from examples, e.g. pictures or other physical measurements, and then determines whether a new product it is shown is good or defective.

Why is defect detection with AI superior to traditional approaches?

The advantages of AI vs manual inspection or traditional automated approaches like machine vision are reliability, high accuracy and high-speed defect detection 24/7 of every single manufactured product, even in high-speed manufacturing. 

Compared to manual inspection, AI is faster, mopre accurate and reliable. It also addresses challenges hiring personnel for quality inspection jobs. 

Compared to machine vision, AI models can be trained more quickly, the hardware is generally significantly cheaper and the models can be retrained quickly if product or process parameters change. They can also flexibly adjust to changes in the environment e.g. changes in lighting conditions (time of day/ season/broken light bulb).

What type of defects does AI detect?

Algorithms can be trained to detect any defect that can be recognized visually or by other physical measurements. This ranges from large and obvious defects, such as a wine bottle without a cork, to the small and subtle, such as a hairline crack in the glass of the wine bottle.

Can AI classify different defect types or classes?

Yes. When trained properly, AI algorithms can differentiate between different types of defects and identify defective units accordingly, e.g. wine bottles a) without corks, b) with the label misaligned, c) with cracks in the glass or d) insufficient fill volume. To correctly classify a defect, the algorithm needs to be trained with enough examples (the training set) of this defect so it can “learn” to spot the defect reliably.

AI models are also capable of identifying new classes of defects, that previously couldn’t be detected.

What is a training set?

To learn an AI-based algorithm needs to “see” examples of both defective and good units. The set of pictures or other sensor data that is used to teach the algorithm to differentiate between good and defective products is referred to as training set. A training set can also be referred to as training library.

How is a training set generated?

Existing sensor data, e.g. images showing good and defective products are collected. These pictures are tagged good/OK or defective/NOK by an experienced team member. These tagged data are then used to train the algorithm.

If defect categorization is needed, the training set has to contain sufficient tagged images showing the defects categories the manufacturer wants to differentiate. Based on that training library the model will learn to identify good and categorize defective units.

What is a validation set?

After the algorithm has been trained using the training set, it is shown a set of pictures or sensor data it has never seen before. This set is used to gauge the performance of the algorithm. If the desired accuracy has not been achieved yet, additional training is needed.

How large do training sets need to be?

This depends very much on the rarity and type of defect(s) and the required accuracy. The rarer a defect, the more training data are needed so the model “sees” enough examples to learn. A large, obvious defect, e.g. a wine bottle without a cork, requires only a few hundred sets of input data such as pictures, for more subtle defects, e.g. tiny scratches on a metal surface thousands of pictures might be needed.

If defects have to be classified into different categories more input data are required than are needed for a simple pass/fail decision. Accuracy also impacts training set size, the more accurate the answer has to be, the more training is required.

What are the quality requirements for a training set, e.g. pictures?

A good training set contains a sufficient number of high-quality, relevant pictures. Just like a human, what the model doesn’t see, it doesn’t recognize. In the case of pictures, good quality means good resolution, lighting, sharpness and relevance. If critical features are outside the plane of focus or insufficiently lit, the algorithm will not be able to learn their relevancy.

What is predictive maintenance, and how does AI play a role in it?

Predictive maintenance involves using data and AI algorithms to predict when equipment is likely to fail so that maintenance can be performed before it breaks down or causes defects. AI analyzes historical data, monitors current conditions, and predicts potential issues, minimizing downtime and optimizing maintenance schedules.

How does the implementation of AI in manufacturing impact overall production efficiency?

AI can significantly improve production efficiency by reducing defects, minimizing unplanned downtime, and optimizing maintenance schedules. This leads to increased productivity, cost savings, and improved overall equipment effectiveness (OEE).

What types of manufacturing processes benefit the most from AI?

Industries with complex manufacturing processes, high-volume production, and stringent quality requirements, such as automotive, electronics, and pharmaceuticals, benefit the most. However, AI applications can be tailored to all manufacturing sectors.

How can manufacturers transition to AI-powered systems without disrupting existing operations?

Implementing AI in stages, starting with pilot projects and gradually scaling up, allows manufacturers to adapt without major disruptions. Collaboration with experienced AI solution providers, employee training, and effective change management strategies contribute to a smooth transition.