Vehicle Modeling

This research area focuses on developing standardized modeling libraries and interfaces that facilitate modular and interchangeable vehicle subsystem models. It addresses the challenges of integrating diverse vehicle components—powertrain, chassis, controllers—across simulation environments to enable whole-vehicle system modeling with configurable architectures. The ability to selectively instantiate or bypass components via conditional connectors supports adaptive model complexity. These methodologies enable flexible assembly of multi-domain vehicle models and promote interoperability across different simulation platforms.

This theme investigates the decoupling of driver behavior from vehicle dynamics to model diverse driver types and vehicle automation levels, from human-driven to fully autonomous vehicles. Separating driver guidance and vehicle response enables simulation frameworks that flexibly combine human drivers, driver assistance systems, and vehicle propulsion types. Such models support analysis of mixed traffic flow containing internal combustion, electric, and autonomous vehicles, providing insights into driver-vehicle interactions and their influence on traffic patterns.

This theme explores advanced computational techniques such as finite element analysis (FEA), deep learning, and surrogate modeling to enhance aerodynamic evaluation and energy consumption prediction for vehicles. These methods aim to reduce reliance on costly physical experiments and enable rapid design iterations. Data-driven approaches using signed distance fields and convolutional neural networks facilitate approximate drag coefficient predictions for arbitrary geometries without explicit parameterization, extending design freedom. Additionally, surrogate modeling supports real-time energy optimization in electric vehicles, improving control strategies.