SCOPE

AI-empowered MBSE (Model-based System Engineering)" is increasingly advocated as a promising approach to overcome MBSE challenges and to promote high-quality systems. At the same time, recent research projects such as the program "Confiance.ai" https://www.confiance.ai/ advocate an MBSE-empowered AI as a way to build trustworthy AI systems.
This workshop will be the opportunity to discuss how to choose, apply, evaluate and adapt AI techniques to support MBSE and how to adopt an MBSE approach to produce high-quality and reliable AI systems. Thus, “AI for MBSE” and “MBSE for AI” are both in the scope of the proposed workshop.

TOPICS OF INTEREST

KEYNOTE SPEAKERS

AI Requirements for an Effective Adoption in an MBSE Context

Jean-Michel Bruel IRIT, University of Toulouse France

Brief Bio
Jean-Michel Bruel has been head of the SM@RT team of the IRIT CNRS laboratory until September 2021. His research areas include the development of software-intensive Cyber-Physical Systems, and methods/model/language integration, with a focus on Requirements and Model-Based Systems Engineering. He defended his “Habilitation à Diriger des Recherches” in December 2006 and obtained in 2008 a full professor position at the University of Toulouse . He has been Laboratory Representative for the Toulouse 2 Jean Jaurès University from 2016 to 2020. He has been head of the Computer Science department of the Technical Institute of Blagnac in 2009-2012 and 2021-2023 and has been a member of the Strategic Research Committee of the IRIT CNRS laboratory since 2021. He is the holder of the Chair of Model-Driven Systems Engineering between AIRBUS and Toulouse 2 Jean Jaurès University since 2022 and joined the International Research Laboratory in AI in 2023.

Abstract
Although recent advancements in artificial intelligence (AI) and large language models (LLMs) have shown promise, there remain inherent challenges that hinder their widespread adoption in critical systems development and operation. This talk will outline the essential criteria that AI approaches must meet to be effectively integrated into Model-Based Systems Engineering (MBSE).

AI-Enhanced Machine-Executable V-Model for MBSE

Stephan Rudolph Institute for Aircraft Design, University of Stuttgart Germany

Brief Bio
Stephan Rudolph holds the academic degrees of Dipl.-Ing., Dr.-Ing. and Priv.-Doz. in aerospace engineering from the faculty of Aerospace Engineering at the University of Stuttgart, Germany. Stephan is Head of the Design Theory and Similarity Mechanics Group and teaches several courses on digital engineering in the MSc specialization "Information Technology in Aerospace Engineering". In 2021 Stephan has been appointed adjunct associate professor at Swinburne University of Technology in Melbourne, Australia. Besides, Stephan is co-founder and associate of a small German engineering company Ingenieurgesellschaft für Intelligente Lösungen und Systeme mbH, a high-tech SME, specializing in digital design automation. Stephan’s research interests include formal methods in Model-Based System Engineering (MBSE) and formal engineering design synthesis methods using graph-based design languages, automatic model generation and design evaluation methods.

Abstract
Applications of Artificial Intelligence (AI) to engineering design of complex cyber-physical systems (CPS) are getting currently more attention than ever despite the fact that the AI known today resembles more a collection of distinct methods than for some sort of closed theory of AI. The talk starts therefore with a scientific engineering needs analysis of a digital, machine-executable V-Model for MBSE and continues with an analysis of the different spaces which occur along the machine-execution of the V-Model. Then, based on this space analysis, the suitability of the different AI techniques is explored and illustrated using the current implementation of the Design Compiler 43 (IILS, 2017) as a further step towards a future “intelligent” design compiler. The key idea of a „design compiler“ as a machine which automates engineering design is around long since (Ward 1989, MIT) but long not believed to be feasible (Whitney, 1996, MIT). While in the electronic design automation (EDA) community so-called „silicon compilers“ are today the key element for the design and the production of new chips in a 3-month rhythm, the idea of digitazing and automating engineering design tasks in an intelligent design compiler still seems to be an open question for many.
The talk illustrates how an AI-enhanced machine executable V-Model for MBSE for the “intelligent” automation of complex engineering design tasks is achieved by means of graph-based design languages based on a UML-like notation. These graph-based design languages consist of ontologies (as UML class diagrams), rules (as UML model-to-model transformations or JAVA code) and production systems (as UML activity diagrams) and can be automatically compiled by means of the Design Compiler 43 into a design graph (UML instance diagram). This design graph represents a holistic and consistent abstract digital representation of the engineered product and is mapped via various design compiler plug-ins into various executable domain-specific language (DSL) models such as CAD-, MBS-, FEM-, and CFD-models. Several enhancements by AI methods in the V-Model are emphasized and illustrated with examples such as the design of satellite propulsion systems, the design of a FORMULA student racing car, automated packaging, piping and routing of subsystems during the system integration phase. (talk duration approx. 30-50 min, depending on details, tbd).
References
Ward, A.: “A Theory of Quantitative Inference Applied to a Mechanical Design Compiler”. PhD thesis. Massachusetts Institute of Technology, 1989.
Whitney, D.: “Why mechanical design cannot be like VLSI design”. Research in Engineering Design 8, 3 (1996), 125–138.
IILS Whitepaper “Total Engineering Automation with Graph-based Design Languages and the Design Compiler 43”, IILS mbH, Trochtelfingen, 2017.

IMPORTANT DATES

WORKSHOP PROGRAM COMMITTEE

PAPER SUBMISSION

PUBLICATIONS

AUTHOR'S KIT

SECRETARIAT CONTACTS