An eye on...

Definition of automatic control (Pierre Bernhard, 2010)

“Automatic control is the alliance of a science and an art, connected thanks to a half-science/half-art discipline:

  • dynamic models science, providing answers to rigorous mathematical questions, the only way to surpass the ordinary,
  • designing dynamic models art, according to questions we want to ask them. This art can only be performed with a solid knowledge of (category of) a model's mathematical properties,
  • validation discipline, by comparing mathematical properties of models with reality. It may lead to a reworking of the model.”


What automatic control deals with (Pierre Bernhard, 2009)

“Automatic control is neither about electronics, nor mechanics, nor process engineering nor about more recent fields of application like finance, epidemiology, behavioral ecology or fisheries science. As a matter of fact, how could it be possible to be a good automation expert if it were to require mastering all of these disciplines, each one as noble as the others?

Automatic control deals with:

  • designing models in each of these areas of use. Not knowledge-based model which are determined by respective specialists, but action models, sufficient to fill a practical need. The cleverness of not taking secondary phenomenon into account — subject to further examination —is an integral part of our profession, but not the least.
  • introducing signals into models. Signals that go in the considered system then go out, in order to get their informative content.
  • scrutinizing mathematical properties so as to deduce the required formulas to obtain representative numbers for a model’s physical quantities.
  • finding a way to validate the global approach… or question it if needed.

To sum up, automatic control is not more about physical object than mathematical properties. In this context, “signals” stand for information. Moreover, as an applied science, automatic control depends on the computational power available, whatever the field is.”