ModAI: Development of an on-device AI platform for manufacturing

Short description of the project

The MODAI project aims to develop an on-device AI platform for manufacturing. It will utilize standards-based technology so that interoperability between different systems is ensured. Furthermore, the project promotes the dissemination of intelligent digital twin technologies that strengthen the manufacturing sector internationally. The solutions developed serve as flagship projects for advanced real-world applications. MODAI supports rolling, machining, and pressing industries both nationally and internationally by offering cloud-based platform services.

Project goals

Fraunhofer IOSB is tasked with the integration of reactive, data-sovereign digital twins based on the Asset Administration Shell (AAS) to enable interoperability in the AI platform. With the Eclipse FA³ST Tools an open-source solution is deployed to operate reactive digital twins. The integration into standardized data spaces like the International Data Spaces (IDS) is considered to promote secure data-sovereign sharing of manufacturing data and AI models.

Our contribution

Fraunhofer IOSB proposes an approach to simplify the integration of manufacturing data inside a data space to train AI models. Since On-Device AI devices will be used, Eclipse FA³ST is a natural fit with its Asset Connection feature to connect with sensors. Our colleagues from Fraunhofer IWU provide the SmartNOTCH and SmartTOOL sensor systems. The Submodel Template Processor feature allows to utilize the AI Model / Dataset Submodel Templates, which describes AI models within the AAS model. Fraunhofer IAO is developing the business model. Bringing Eclipse FA³ST and IDS connectors to the korean testbeds promotes an international data space based on standardized technologies.