AIQAS

develops an intelligent system for advanced inspection in industrial production lines

The main outcome of the AIQAS project was the development of a system that enables multi-defect and multi-material inspection of parts produced through continuous production line industrial processes.

Challenges

To detect, at a single station, various defects and anomalies in real time across different materials.

To ensure that the system can learn incrementally and flexibly through semi-supervised labelling of detected anomalies.

To deploy AI models on computing hardware physically located on the production line (IoT Edge Computing).

Solution

The developed solution is based on real-time analysis of multiple types of defects simultaneously, resulting in reduced costs, and it can be implemented across different types of flat materials.

By using deep learning techniques (for example, convolutional neural networks), the system can process images holistically and automatically learn the most effective features for defect characterisation, without the need for pre-specification.

This represents a major advancement compared with the solutions available so far, which involve separate analyses for different defect types and are specific to each material.