Smart Manufacturing Coordination with Integrated Large Language Models

SMaiLL aims to develop a data ecosystem that captures information from various sources of all the different phases of the product and production system life cycle and provide to the worker real time insights and alerts via user interfaces (e.g. graphical user interface, mixed reality interface, voice user interface, etc.). The data would be collected from the manufacturing environment by the use of standardized Asset Administration Shell (AAS) models, providing scalability and interoperability. Providing access to this large amounts of data from the manufacturing environment, LLM aims to coordinate advanced AI tools, and take real-time decisions, which would be proposed to the decision makers through a natural language-based UI. Moreover, Figure 2 illustrates the use of the data space, where the aim is not only to share a trained LLM but also to enhance the collaborative training approach (federated learning) across different companies, technology providers, and technology consumers.