Coordinator: Francesco Carravetta
Description
MCISCO is involved in theoretical and applied research in the wide-sense field of Control Systems, which includes, besides devicing strict sense control systems, communication and detection systems as well, the latter being often embedded into and devised at once with an overall control system.
Research Activities
Nonlinear systems
Nonlinear systems are mathematical descriptions of natural or man-made (engineering) systems, where the main difficulty in the analysis
is essentially algebraic: a bias from linearity in the algebraic expression of the system, typically difference and/or differential
equations. Even though not limited to them, an intense research activity is devoted to
Differential systems, and many meaningful results has been obtained. One of the most fruitful research lines is concerned with the
‘algebrizzation’ of the equations of a nonlinear system, i.e. searching for a polynomial differential equation having the same solution
of a given more
complex nonlinear differential equation. The research activity is devoted as well to numerical integration, with application
to biological systems.
Filtering and Identification
Filtering and identification is the field of research aiming to working out devices for estimation, of the state or of a parameter, for
stochastic systems. As above sketched, these are ofted embedded into a control system, as a separate part, even though they are devised
at once with the controller. Filters and identification devices are used in a stand-alone form as well, as communication-detection devices,
for instance: as a man-machine interface. They are used as well in cyber-security (or to attack cyber-security devices), and/or in fault
detection systems. The MCISCO group has a long time record of research results in filtering and Identification for
nonlinear/nonGaussian systems. Applications in wheater forecast have also been developed and still are active research lines.
Artificial Intelligence
As a general definition, Artificial-intelligence devices aims to simulate/reproduce some tipically human capability, like for instance:
vision and speech-based communication.
We group under this title all research lines involved with stochastic systems of a particular type, not owning the usual
analytic-algebraic structure of a difference and/or differential equation. This is the case of the so-called syntactic systems, used
for automatic translation/understanding of natural languages, which have been studied and developed within the MCISCO research activities.
We also group under this title all activities aiming to design Markov-field based devices, such as bayesan filters, for which the ‘system’,
i.e. the problem data, reduces to just a probabilistic description of a signal.
Nonlinear control
In the framework of difference/differential systems, two ways are the most used to define ‘control’ in a mathematical
sense: 1) as an optimization task
(optimal control) 2) as an ‘inversion problem’. The ‘stochastic counterpart’ of these notions is defined as well, but in a quite
restrictive sense, and is indeed still an open research problem (in which MCISCO is involved). As for the deterministic case, MCISCO leads
research mostly on inversion problems, where the main difficulty has an algebraic nature and is strictly related to what described above at the
item ‘nonlinear systems’.