A Simple and Efficient Tube-based Robust Output Feedback Model Predictive Control Scheme
The control of constrained systems using model predictive control (MPC) becomes more challenging when full state information is not available and when the nominal system model and measurements are corrupted by noise. Since these conditions are often seen in practical scenarios, techniques such as robust output feedback MPC have been developed to address them. However, existing approaches to robust output feedback MPC are challenged by increased complexity of the online optimization problem, increased computational requirements for controller synthesis, or both. In this work we present a simple and efficient methodology for synthesizing a tube-based robust output feedback MPC scheme for linear, discrete, time-invariant systems subject to bounded, additive disturbances. Specifically, we first formulate a scheme where the online MPC optimization problem has the same complexity as in the nominal full state feedback MPC by using a single tube with constant cross-section. This makes our proposed approach simpler to implement and less computationally demanding than previous methods for both online implementation and offline controller synthesis. Secondly, we propose a novel and simple procedure for the computation of robust positively invariant (RPI) sets that are approximations of the minimal RPI set, which can be used to define the tube in the proposed control scheme.