Leveraging semantic interoperability in Industry 4.0: an opportunity for automated task planning for discrete manufacturing processes

Abstract

Proprietary data formats in manufacturing and supply chain ecosystems impose significant manual effort to harmonize cross-system communication, spanning from shop-floor sensor data over manufacturing execution systems (MES) to enterprise resource planning (ERP) systems. This paper introduces a semantic interoperability framework powered by the Asset Administration Shell (AAS), a standardized digital twin model designed to unify heterogeneous systems. By leveraging AAS, the framework enables resilient, real-time production planning at the shop-floor level, facilitating dynamic process planning, synchronizing available resources, and automated deviation detection via seamless integration into a planning/execution service. In parallel, initiatives like Aerospace-X extend this paradigm in order to address broader supply chain challenges such as demand-capacity bottlenecks, quality management, and circular economy goals. These efforts establish federated data ecosystems which utilize standardized submodels, such as digital nameplates and digital product passports, to enhance visibility across multi-tier supplier networks, mitigate disruptions from material shortages, improve collaborative decision-making, and instantiate a market place for central services. The idea is that transitioning from rigid, inflexible supply chains to multilateral, AAS-enabled supply networks reduces lead times, optimizes inventory management, and strengthens resilience against global disruptions. To show the potential of AAS-driven manufacturing we explored the feasibility to use AASs with their inbuilt capability and operation submodels to perform a discrete assembly task. In the FoF-X (Factory of the Future – eXtended) project we used these models to demonstrate autonomous task planning as well as autonomous error recovery. By combining planning via classic symbolic artificial intelligence (e.g. by using Planning Domain Definition Language (PDDL)) with the modeling power of AAS submodels and perspectively formal semantic ontologies (e.g. OWL, RDF) for capability descriptions defined as AAS submodels, the framework achieves an agnostic and autonomous approach to planning, execution, non-conformity detection, and re-planning. The current implementation focuses on a toy problem to evaluate the technology. An application in an aerospace context is planned in the continuation project in ASPIRO (Aerospace production using intelligent robotic systems). The results regarding implementation, caveats, and future perspectives are shown in this paper. The results of this part of FoF-X emphasize the transformative role of semantic standardization in bridging vertical integration (shop-floor to ERP) with horizontal collaboration (supplier to OEM), offering a scalable blueprint for Industry 4.0 digitization, which enables higher efficiencies in distributed supply chains as well as higher resource utilization on the shop floor. This work underscores the critical importance of adopting AAS as a universal language for enabling intelligent and adaptive manufacturing ecosystems.