Real-world autonomous systems often have dynamics that are best described by infinite dimensional systems. For example, PDE-constrained systems that require aerodynamic modeling (UAV control) or structural deformation modeling (soft robotics). Current frameworks for design and simulation of infinite dimensional systems, such as computational fluid dynamics (CFD) and finite element methods (FEM), are well established. However, they often fall short in the context of control applications due to their heavy computational complexity. This project investigates the use of reduced order models to address this shortcoming, especially for use within the framework of model predictive control.
Keywords: Model Predictive Control, Dimensionality Reduction, Reduced Order Modeling
Students: Joe Lorenzetti, Benoit Landry