Controller Design Libraries

Frameworks for model- and AI-based controller development

The design of control systems for complex mechatronic applications such as autonomous vehicles requires both robust model-based methods and modern data-driven approaches. To address these demands, the Department of Vehicle System Dynamics and Control provides specialized controller design libraries that support the entire development workflow—from modeling and simulation to training and validation. The VDC-Workbench is a Modelica-based library offering a unified platform for developing, testing, and validating vehicle dynamics controllers and energy management strategies, featuring multi-physical component modeling and baseline path-following controllers that can be extended with machine learning methods. Complementing this, train_fmu_gym is a framework that enables the training and evaluation of reinforcement learning agents using Functional Mock-up Units (FMUs), allowing engineers to leverage existing simulation models for deep reinforcement learning. Together, these libraries facilitate the seamless combination of classical control engineering with AI-based approaches such as residual reinforcement learning, accelerating the development of high-performance controllers for multi-physical systems.

Community Edition Libraries