Model-Based System Engineering Framework for Verification and Validation (V&V) of Cyber-Physical Vehicle Systems

Abstract

Autonomous vehicles are highly complex cyber-physical systems that require a multitude of design choices spanning vehicle platforms, software architecture, and algorithmic frameworks. These design choices inherently encapsulate the modularity of modern autonomous vehicles, necessitating systematic capture and analysis to optimize performance, safety, and reliability. This study introduces a digital twin-based meta-modeling approach to develop a modular framework for Model-Based Systems Engineering (MBSE) in autonomy development. By leveraging this framework, we enable structured integration, rapid prototyping, and validation of autonomy components across different operational scenarios. To demonstrate the effectiveness of this approach, we apply the framework for an off-road autonomous vehicle path planning problem over a custom terrain, utilizing MATLAB Simulink, NVIDIA Isaac Sim, and AutoDRIVE Simulator as key simulation environments. We rapidly prototype variant design choices through MathWorks System Composer and establish a verification and validation workflow with complete requirements traceability across the digital thread. This allows us to systematically assess system performance and refine decision-making at different levels of abstraction. Furthermore, we conduct a comparative performance analysis of various path planners across multiple key performance indicators to evaluate their effectiveness in satisfying mission objectives in diverse off-road conditions. These findings also emphasize how simulator choice affects fidelity, influencing precision, reliability, and real-world deployment suitability while enabling rapid testing and refinement of early-stage algorithm developments.