MoDEVVa will take place on Sunday, October 5th, in Michigan, USA (co-located with MODELS 2025)
Keynote Speaker: Prof. Betty Cheng (Michigan State University, MI, USA)
Model-Driven Approaches to Addressing Assurance for AI-enabled Autonomous Systems in the Face of Uncertainty
Abstract: This presentation will overview several research projects that explore how model-driven requirements engineering can been used to model, analyze, and mitigate uncertainty arising in three different aspects of high-assurance autonomous systems. First, uncertainty about the physical environment can lead to suboptimal, and sometimes catastrophic, results as the system tries to adapt to unanticipated or poorly-understood environmental conditions. Second, uncertainty in the cyber environment can lead to unexpected and adverse effects, including not only performance impacts (load, traffic, etc.) but also potential threats or overt attacks. Finally, uncertainty can exist with the components themselves and how they interact upon reconfiguration, including unexpected and unwanted feature interactions. Each of these sources of uncertainty can potentially be identified and mitigated at design time and run time. All of these factors are further highlighted by the increasing role of AI, which we will also explore in this talk. Based on a number of collaborative projects involving industry applications, we share lessons learned and identify research challenges to applying model-driven requirements engineering to address uncertainty posed by the changing roles of humans, computers, and their collective ecosystem.
Bio: Betty H.C. Cheng is a professor in the Department of Computer Science and Engineering at Michigan State University. She has also been the Industrial Relations Manager and senior researcher for BEACON, the National Science Foundation Science and Technology Center in the area of Evolution in Action. Her research interests include self-adaptive autonomous systems, safe use of AI-enabled systems, requirements engineering, model-driven engineering, automated software engineering, and harnessing evolutionary computation and search-based techniques to address software engineering problems. These research areas are used to support the development and maintenance of high-assurance adaptive systems that must continuously deliver acceptable behavior, even in the face of environmental and system uncertainty. Example applications include intelligent transportation and vehicle systems. She collaborates extensively with industrial partners in her research projects in order to ensure real-world relevance of her research and to facilitate technology exchange between academia and industry. She has collaborated with Ford, General Motors, ZF, Motorola, and Siemens. Previously, she was awarded an NASA/JPL Faculty Fellowship to investigate the use of new software engineering techniques for a portion of the NASA space shuttle software. She currently has projects in the areas of assured autonomy (systems with machine learning components), model-driven approaches to autonomous systems and digital twins, cyber security for automotive systems, and feature interaction detection and mitigation for autonomic systems, all in the context of operating under uncertainty while maintaining assurance objectives. Her research has been funded by several federal funding agencies, including NSF, AFRL, ONR, DARPA, NASA, ARO, and numerous industrial organizations. She serves on the journal editorial boards for ACM Transactions for Autonomous and Adaptive Systems, as well as Software and Systems Modeling; she has served as Co-Associate Editor-in-Chief and two terms as an Associate Editor for IEEE Transactions for Software Engineering and Requirements Engineering Journal. She was the Technical Program Co-Chair for IEEE International Conference on Software Engineering (ICSE-2013), the premier and flagship conference for software engineering.
She received her BS from Northwestern University and her MS and PhD degrees from the University of Illinois-Urbana Champaign, all in computer science. She may be reached at the Department of Computer Science and Engineering, Michigan State University, 3115 Engineering Building, 428 S. Shaw Lane, East Lansing, MI 48824; chengb@msu.edu; https://www.cse.msu.edu/~chengb
Models are purposeful abstractions of systems and their environments. They can be used to understand, simulate, and validate complex systems at different abstraction levels. Thus, the use of models is of increasing importance for industrial applications. Model-Driven Engineering (MDE) is a development methodology that is based on models, metamodels, and model transformations. The shift from code-centric software development to model-centric software development in MDE opens up promising opportunities for the verification and validation (V&V) of software. On the other hand, the growing complexity of models and model transformations requires efficient V&V techniques in the context of MDE.
The workshop on Model Driven Engineering, Verification and Validation (MoDEVVa) offers a forum for researchers and practitioners who are working on V&V and MDE. The main goals of the workshop are to identify, investigate, and discuss mutual impacts of MDE and V&V.
For the 2025 edition of the MoDEVVa workshop we would like to encourage papers addressing the use of AI techniques such as machine learning, to help address the challenges of model-based V&V, Process Engineering and Quality Assurance, while continuing to welcome work in all areas in the intersection between MDE and V&V.
Modelling is a powerful technique for handling the complexity of software and hardware artifacts, and their respective environments. Model Driven Engineering (MDE) provides efficient tools for building and working with models, from the requirements specification of a system to code-generation, testing, configuration and deployment. Through the systematic use of digital models, which can be processed automatically by programs, MDE offers the opportunity to verify and validate every step in the life cycle of a system. Thus, the first motivation for MoDEVVa is the integration of verification and validation (V&V) techniques into MDE.
While V&V can be seen as an enabler in MDE, it presents a set of challenges of its own. These challenges includes issues of usability and integration with MDE processes as well as the technical difficulties of performing V&V tasks.
One way of addressing these challenges is by taking ad-vantage of MDE itself in V&V tasks, for example by means of domain-specific modelling languages (DSMLs) to capture requirements, system properties, specifications and system de-sign, and leveraging all MDE has to offer such as abstraction, refinement, model-transformations and other techniques, to help perform V&V tasks. Thus, the second motivation for MoDEVVa is the integration of MDE techniques into V&V.
Another way of addressing the challenges posed by V&V in MDE is to leverage novel techniques from AI. The advent of practical machine learning techniques and frameworks opens the way for novel approaches to model-based V&V, which are poised to improve the usability and range of V&V. Thus, the third motivation for MoDEVVa is the integration of novel approaches to the challenges presented by V&V and MDE.
Both MDE and V&V intend to help solve “real-world”problems. Real-world problems and systems are complex.Both MDE and V&V propose approaches to tackle such complexity. Thus, the fourth motivation for MoDEVVa is the applicability of MDE and V&V to complex, real-world problems.
The overarching objective of the MoDEVVa workshop is to bring together researchers and practitioners in the domain of V&V and MBSE/MDE so that the key issues in the integration of MDE and V&V can be identified and solved.
More concretely, MoDEVVa's main objectives are to address the following questions:
How can V&V tools and techniques be integrated into MDE in such a way that expertise in V&V is not required in order to obtain the benefits that V&V offers?
How can MDE be leveraged to facilitate V&V tasks?
How can novel approaches such as Machine Learning be leveraged to facilitate V&V in MDE?
How can the combination of MDE and V&V help to address the development of complex real-world systems?
How can MDE be leveraged to facilitate industry to acquire certification for their systems or qualification of their development processes and tools?
How to deploy V&V in ``lightweight'' modeling environments that do not use explicit metamodeling or heavy modeling infrastructures?
How MDE and V&V help in increasing confidence in modern systems involving more and more AI components?
We are pleased to announce that the best papers from MoDEVVa and the SAM Conference will be invited to submit extended versions jointly published in a special issue of Innovations in Systems and Software Engineering: a NASA Journal (ISSE) published by Springer Nature !!
We welcome contributions in all areas at the intersection of MBSE/MDE and V&V. Papers addressing the following topics are particularly welcome:
V&V in MBSE/MDE
Theoretical frameworks and approaches for integration of V&V in MBSE/MDE.
Formalisms and theories for the specification and verification of models.
Formal approaches to models, modeling languages, including DSMLs and MDE in general.
Modeling relations for checking model conformance and/or refinement.
The application and combination of different V&V techniques (e.g., classical testing, static analysis, model checking, deductive approaches, runtime verification) to MBSE/MDE artifacts.
V&V in “lightweight” modeling environments that do not use explicit metamodeling or heavy modeling infrastructures
MDE in V&V, Certification and Quality Assurance
Use of MDE abstractions (models, meta-models, model transformations) in V&V tasks.
Use of model-evolution approaches to enable incremental V&V.
Industrial case studies for application of MDE for quality assurance.
Model-based process engineering to acquire certification.
Process engineering to support V&V activities.
Tools, usability, and applications
Integration between modeling tools, IDEs and V&V back-ends.
Innovative approaches for model-based V&V of “real-world” systems.
Tools and techniques that help reduce the semantic gap between models and back-end formalisms used in V&V tasks.
Case studies and applications of V&V in MBSE/MDE.
AI-related topics for V&V activities
Use of Machine Learning (ML) to assist model-based V&V activities (e.g., testing selection, generation and prioritization)
AI-enabled model inspection
AI-enabled frameworks/processes for model-based testing
Current practices/case-studies/experience reports on applying ML-assisted model-based V&V.
Use of MDE and V&V in systems that involve AI components
Saad Bin Abid (CARIAD SE/Alten, Germany)
Juergen Dingel (Queen's University, Canada)
Jens Kosiol (Philipps-Universit ̈at Marburg, Germany and Universit ̈at Kassel, Germany)
Rakshit Mittal (University of Antwerp - Flanders Make, Belgium)
Iulian Ober (ISAE-SUPAERO, Université de Toulouse, France)
Ernesto Posse (Lumenix/Zeligsoft, Canada)
Contact: modevva@gmail.com