Current students

Control theory is now a rapidly evolving field. Data are having a substantial impact on modern research in the automation field. In many applications, retrieving a model is considered a challenging task. Learning a model that fits available data is the crucial passage required in model-based control design. Moreover, the model is not obtained with a control-oriented purpose, and it is a time-consuming task. For these reasons, data-driven methods have known significant interests. They allow to skip the model identification step and directly exploit data for the controller design. The work revolves around studying methods to exploit data for the direct design of controllers and reference governors. Many real-world systems have performance limitations, such as safety constraints or actuators' saturations, which have to be appropriately accounted for in the design of efficacious controllers. Few data-driven methods are available to exploiting data for directly designing controllers while explicitly accounting for constraints. In this context, the present work is devoted to proposing methodologies to account for constraints in the direct design of controllers. The focus will be kept on both the algorithmic viewpoint as well as the application side.