Matlab/Octave toolbox for structurable and robust output-feedback LQR design

Paper i proceeding, 2018

In this paper, a structurable robust output-feedback infinite horizon LQR design toolbox for Matlab and Octave is introduced. The aim of the presented toolbox is to fill the gap between available toolboxes for Matlab/Octave by extending the standard infinite horizon LQR design (from Matlab/Control System Toolbox, Octave/Control package) to robust and structurable output-feedback LQR design. The toolbox allows to design a robust infinite horizon output-feedback controller in forms like proportional (P), proportional-integral (PI), realizable proportional-integral-derivative (PID), realizable proportional-derivative (PD), realizable derivative (D), dynamic output-feedback (DOF), dynamic output-feedback with integral part (DOFI), dynamic output-feedback with integral and realizable derivative part (DOFID), and dynamic output-feedback with realizable derivative part (DOFD). In addition, the controller structure for all supported controller types is fully structurable. The toolbox relies on Yalmip (A Matlab/Octave Toolbox for Modeling and Optimization) and on linear matrix inequality solvers like SeDuMi, SDPT3, etc. Notions like "simple", "highly customizable", and "user-friendly" have been used and considered as main terms during the development process.