Special Session organized by

Luis Miguel Vieira da Silva, Helmut Schmidt University Hamburg, Germany, Gernot Steindl, University of Applied Sciences Burgenland, Austria, Prerna Juhlin, ABB Corporate Research Center, Germany, Felix Ocker, Honda Research Institute Europe, Germany, Ander Garcia Gangoiti, Vicomtech, Spain, Gero Strobel, University of Duisburg-Essen, Germany, Alexander Fay, Ruhr University Bochum, Germany.

---

Download Call for Papers

Click here to download the special session cfp.

---

Focus

Industrial automation must adapt quickly and reliably while leveraging heterogeneous data across systems. Semantic models provide structured representations and reasoning but are costly to engineer. Generative AI, particularly Foundation Models (FMs), unlocks unstructured information and natural interaction but raises concerns regarding formal guarantees. This Special Session targets Hybrid AI that combines both towards flexible, reliable, trustworthy industrial systems: Using semantic models to ground and verify generative AI behavior, and generative AI to automatically create and evolve semantic artifacts.

---

Topics under this special session include:

• FMs for Semantic Model Engineering: Automated generation of models (e.g., ontologies, AAS, PDDL), natural language interfaces for querying, automated mapping between different representations (OWL, OPC UA, AutomationML, etc.)
• Semantic Models for FM Grounding and Control: Validation and constraint checking of outputs, Verification of generated control logic or process plans, knowledge graph-enhanced retrieval (graphRAG) for planning and maintenance
• Hybrid Reasoning and Planning: Integrating symbolic reasoning and planning (e.g., PDDL, SMT) with FM-based approaches, semantically-guided decision-making, FM-assisted task decomposition with model-based capability matching
• Trustworthy AI through Formal Grounding: Explainable AI (XAI) using formal models, Traceability and provenance tracking for AI decisions, Validation ensuring consistency between FM outputs and domain models
• Industrial Applications and Integration: FM-based control, multi-modal integration, natural-language interfaces for HMI, Integration of FMs and emerging conventions (e.g., MCP, A2A), digital twins enhanced with generative AI capabilities
• Engineering Methods and Tools: Model-driven development, benchmarks and evaluation frameworks, domain-specific fine-tuning of FMs on industrial knowledge bases, automated code generation, formalization of engineering artifacts