AAS Clause Samples

AAS. In this chapter the current status of the AAS semantic models of the MAS4AI framework are explained. We go into how they are designed and provide a functional level description of the used shells. As the modelling process is ongoing, not all AAS models are of equal maturity level, however all agents defined in WP1 have a provided rudimentary model. 4.1 Design process The AAS models developed in MAS4AI aim to be both generically applicable to all current and future MAS4AI use cases. Furthermore, they should be rich enough to be used in their standard form, or with very minor additions. Recently, an approach outlining how to create AAS models was published as a result from the DIMOFAC project [Bout21]. This methodology (shown in Figure 8) is geared specifically towards use in industry and can therefore be used to achieve the abovementioned goals. The methodology focusses on iteratively building a model while keeping in mind three distinct general modelling phases. First is a conceptual (knowledge representation) phase. It focusses on what is needed from the model and which concepts it should contain. Within this phase there has been close collaboration with the use case owners to determine how they intend to use the agents and what the use case generic concepts are versus the use case specific components. After the conceptual level is agreed upon the models enter the formalization phase (alternatively called Industry 4.0 compliance phase). In this phase the ideas and high-level models from phase one are broken into separate AAS submodels, which are filled with the specific elements and then formalized. Finally, the Semantic identifiers are aligned with pre-existing models and concepts. This process, like the one in phase one can be performed in an agile way where insights gained during the process feed back into the next iteration of the development cycle. Within this phase a lot more collaboration with the RTO’s and agent developers in the project took place. As they can provide more detailed input on which properties are needed for a specific agent. The final phase in the used methodology is the evaluation phase. In this phase the models are used for the development of agents, and feedback is gathered. Depending on this feedback, the design process may go back to an earlier phase to improve the model. Evaluating the MAS4AI models is scheduled for the upcoming period, and no model has yet completely gone through the design process. Within the third phase we...