TRansferability of experiential knowledge in industrial inspection by explainable AI

Purpose and goal:

The project fulfilled the main goal to create the Explainable AI system which suggests the product quality based on the given cutting parameters. The system is based on the Computer Vision and AI techniques which enables the intelligent analysis of the drilling-induced defects. The collaboration between academic and industrial partners provided unique combination of expertise need to develop the AI solution. The results has been published and presented in national and international conferences.

Expected results and effects:

As planned, project delivered a robust solution which enables the transferability of the experiential knowledge in industrial inspection. The developed AI solution is based on the results on the manual inspection and contained all unique knowledge from industrial experts regarding the product quality and defect formation. The AI solution as it is can be used and expended by the unexperienced users in both industrial R&D and academia. The AI system is passed validation and field tests which confirmed its reliability.

Approach and implementation:

The delivered AI system, which provides the transferability of industrial experience is based on the set from both mechanical engineering and computer science field. The combination of different approaches to the data generation and processing provided efficient and highly accurate training of AI using experimental data. The Validation of the trained AI was done in the various environments which confirmed its versatility and stability.