Design decisions in vehicle engineering are complex, time-critical, and expensive. Even a minor modification to a single component can affect the entire system—from vehicle mass to lap time. This is precisely where the STRIKE-FS research project, conducted within the framework of the InnovationsCampus Future Mobility (ICM), comes in.
How Algorithms Help Design a Race Car
- News
Explainable AI in Vehicle Engineering
At the Institute of Aircraft Design (IFB) at the University of Stuttgart, Dr.-Ing. Stephan Rudolph and Julian Borowski areexploring how artificial intelligence can systematically analyze these highly complex interdependencies. A Formula Student race car serves as a real-world demonstrator for a central question: Can a large language model help engineers make better decisions and faster?
The methodological foundation is a machine-executable V-model based on Model-Based Systems Engineering (MBSE). Using a graph-based design approach, the vehicle’s components and their functional and physical relationships are represented as an interconnected system model. This makes it possible to systematically capture and make explicit even those dependencies between subsystems that are typically embedded implicitly in design data. A key distinguishing feature of this approach is that the MBSE model is not merely descriptive but fully machine-executable. The process ultimately produces a digital product model represented as a graph. This graph ensures digital model consistency, process continuity, and interoperability across software tools—preventing errors caused by outdated or incompatible data throughout the design process.
The result is not an autonomous decision-making system, but an intelligent engineering assistant. If developers change the material of a control arm, for example, the system automatically analyzes the potential impact on weight, stiffness, or aerodynamics. It highlights trade-offs and evaluates design options using measurable performance indicators such as vehicle mass, aerodynamic drag, and lap time.
The practical benefits are clear: With design automation, development cycles are reduced essentially to algorithm execution time. Powered by modern computing hardware, this translates into acceleration by several orders of magnitude—from several months to just a few days or even minutes. This capability is enabled by novel graph-based design languages that are compiled into executable engineering models. Explainability is supported by an integrated Large Language Model (LLM), allowing complex model relationships to be communicated in an accessible way.
The project has also demonstrated that the methodology is transferable to a wide range of model-based systems engineering applications. It is therefore not limited to traditional vehicle development, but is equally relevant to the development of manufacturing systems in the context of the digital factory, as well as to aircraft and satellite systems engineering.
Further Information
- Research Area “Design Theory and Similarity Mechanics,” IFB, University of Stuttgart
- Open Access Publication „Managing Design Variants in Formula Student Race Cars: A Digital Engineering Approach Across Multiple Teams.“
- Formula Student Racing Team, University of Stuttgart
- Project website Bottom-Up-Project "STRIKE-FS"
Contact
Expert Contact
PD Dr.-Ing. Stephan Rudolph
Head of the "Design Theory and Similarity Mechanics" Research Group
Institute of Aircraft Design, University of Stuttgart
rudolph(at)ifb.uni-stuttgart.de
Marketing und Communications
Teresa Mittner
Marketing and Communications, InnovationCampus Future Mobility