Explainable AI for Manufacturing and R&D: AI Regio Project It is in fact a research and development initiative in the field of artificial intelligence (AI) which is part of the I4MS (ICT Innovation for Manufacturing SMEs) initiative of the European Commission.
In questo contesto, Intellico propone “Grapho“, una tecnologia che impiega tecniche di intelligenza artificiale avanzate basate su GNN (Graph Neural Networks) per rendere più efficiente e accurato il processo di identificazione e riutilizzo dei componenti esistenti nelle nuove generazioni di prodotti. Its purpose is to improve the productivity and competitiveness of European manufacturing companies by speeding up the process of identifying common platforms and therefore reducing variants.
Our ambition at Intellico is to achieve this goal as part of the European AI-Regio project. The European project aims to create a European network between the initiatives already active on the topics of the Digital Transformation of enterprises and production processes (e.g. Vanguard Initiative Pilot, S3 Platforms, …), with Data-driven or AI-Driven techniques. The project is part of the I4MS (ICT Innovation for Manufacturing SMEs) initiative promoted by the European Commission to expand the digital innovation of manufacturing SMEs in Europe, in order to increase their competitiveness. The initiative funded 30 regional application experiments focusing on the adoption of AI technologies in specific industrial cases, including Intellico’s “Grapho” project.
The industrial context of Intellico’s project: the challenge of ETO/MTO companies
ETO/MTO (Engineer to Order/Make to Order) companies in manufacturing can process hundreds of CAD projects per week. The challenge of enhancing process efficiency is to find out which products or components already exist and can be reused or adapted in the initial phase of a new generation or product variant (design by re-use).
It is a common challenge for many manufacturing companies that have a large database of CAD product models containing years of engineering experience. However, it is often difficult to perform targeted database queries to find specific design information, and this can slow down the design process, making it difficult to re-use existing components.
In this sense, the use of innovative Explainable AI solutions, such as the one proposed by the Intellico team, can help manufacturing companies overcome these challenges and optimize the product design process, improving the productivity and competitiveness of the entire company.
Intellico’s proposal for AI Regio
The solution proposed by the Intellico team is based on an innovative Graph Neural Network (GNN)-based model for extracting class-specific feature vectors from CAD product models.
This model is able to perform efficient and accurate geometric similarity identification between CAD profiles, facilitating similarity search and helping users to quickly identify and reuse existing components in new product generations.
The model is able to represent CAD profiles within a graph, where the nodes are characterized by product components and the edges represent the similarity relationships between product components identified by the AI. The model then uses a GNN architecture to process the graph and generate a feature vector for each product component. The feature vectors generated by the model can then be used to perform geometric similarity identification between the components of different CAD product models. This capability also greatly aids in the process of archiving.
Why choose Intellico’s Explainable AI solution?
The solution proposed by Intellico has several benefits for ETO/MTO companies and in general for companies managing product “platforms”:
- Increased reuse of past designs: thanks to efficient and accurate geometric similarity identification between components of different CAD product models, the system enables rapid identification of existing components that can be readily reused or modified in the design of new products.
- Increased standardization and efficient management of CAD profiles: supports the correct classification of design into specific families based on similar characteristics, improving the existing archiving system, and facilitating the efficient management and retrieval of CAD profiles.
- Increased design department productivity: ability to quickly identify existing components that can be reused or adapted in the design of new products, reducing design time and increasing overall department productivity
- Reduced computing time compared to traditional software: artificial intelligence techniques allow for faster and more efficient identification of similar profiles. They also enable more accurate and efficient identification of geometric similarity between components of different CAD product models compared to traditional software (e.g. based on pixel-to-pixel comparison).
Conclusions
The AI Regio project led to the creation of an Explainable AI-based solution to improve decision-making in product design. As we have seen, Intellico developed a GNN-based model to extract class-specific features from CAD product models, enabling efficient and accurate geometric similarity identification and improving the productivity and competitiveness of ETO/MTO companies.
The solution could be used by other manufacturing and engineering companies to improve decision-making in product design, increasing the company’s productivity and overall competitiveness. Grapho can also facilitate the research and development of new products by identifying “platforms” against which the design of new items can be initiated.
The solution is currently in the consolidation phase and will be applied within a real-world context involving Phoenix International SpA, a leading multinational company in the aluminum die extrusion sector.
For more information, please do not hesitate to contact us via the CTA at the bottom of the page.
