Toward Intelligent Generation of System Architectures
Oliver von Heißen, Fabian Hanke, Isaac Mpidi Bita, Aschot Hovemann, Roman Dumitrescu · 2024
Designing new systems involves collaboration across mechanics, electronics, and computer science, often through multiple iterations. We introduce a method using large language models to streamline architecture creation, incorporating the requirements, functional, logical, and physical (RFLP) concept, thus speeding up the process and reducing errors. A proof of concept is demonstrated within Cameo Systems Modeler, applied to a remote-controlled autonomous car project. Feedback from model-based system engineering experts confirm the approach's relevance and provides direction for enhancements.