The next part of the article series „AI in Industry“ deals with quality management in the automotive industry. Since enormous costs arise from sorting and revocation in virtually all industries, an increase in product quality is immensely significant.
In the automotive industry, such a quality management system can be implemented relatively easily. But first, data is needed with which an AI can be trained. For this purpose, employees can easily take pictures of different components with standard cameras – the higher the resolution, the better the result! Another employee then marks irregularities or errors on the images – a few hundred images are sufficient for this, so the effort is rather small than the benefit! Afterward, the images can be transferred to the AI system to learn independently to recognize deviations. This is possible without a lot of programming experience because the AI is trimmed on its own initiative. Therefore, no specialized personnel is required to configure or maintain AI.
Now the system is already prepared for the real environment and can be used. The processing of the productive images takes only a few hundred milliseconds so that no delay in serial production is to be expected. In this time span, hundreds or thousands of existing images are matched to obtain the highest possible results. If necessary, every step of the final assembly can be checked in the car factory, from the correct and scratch-free assembly of a door to the correct positioning of the first aid kit in the trunk. An AI application in the production line makes sense; only its dimension is up to each individual.