TABLE OF CONTENTS Control design requires some knowledge about the dynamic behaviour of the plant. This knowledge is structured in a mathematical model of the plant. Nevertheless, it is impossible to obtain models that accurately describe the dynamic behaviour of the plant. No matter the effort dedicated to obtain a model, there always exists a discrepancy between the behaviour predicted by the model and the behaviour of the real system. This discrepancy between the mathematical model and the real system is known as model uncertainty.
Since the controller is designed using the model of the plant, stability and performance features of a closed-loop system will naturally be affected by model uncertainty. The purpose of robust control is to formulate analysis and design tools that can guarantee stability and performance conditions despite modelling errors. In LTI models, one of the first robustness measures were the phase and gain margins, which indicate the distance between the Nyquist plot and the critical point ?1. However, these stability measures are useful just in very particular cases of modelling errors and cannot be used in multivariable systems [67, 75].
In robust control theory, the dynamic behaviour of a system is described by a set of possible models. This family of models, denoted by 
, is characterised by a nominal model G 0 and a bounded uncertainty set ?. Although there are several sources of uncertainty, they can be classified into parametric and dynamic ones.
In the case of parametric...
