In order to realize autonomous manufacturing systems in environments characterized by high dynamics and high complexity of task, it is necessary to improve the control system modelling and performance. This requires the use of better and reusable abstractions. In this paper, we explore the metamodel techniques as a foundation to the solution of this problem. The increasing popularity of model-driven approaches and a new generation of tools to support metamodel techniques are changing software engineering landscape, boosting the adoption of new methodologies for control application development.