Research group profile

The Identification and Decision Making Research Group (IDM) aims at fundamental and applied reseach in fields of identification and system control, estimation of unknown variables, and decision, all under determenistic or uncertainty settings. The main research areas compose of system identification, unknown parameters estimation, variables prediction, decision making for fault or change detection, optimal control, adaptive control and adaprive systems for signal processing, information fusion and optimalization in estimation, control and decision areas.

Main research areas

The main research area contains the following areas:

  • Adaptive dual control
  • Fault detection
  • Nonlinear filtering
  • Information fusion

Our goal in these areas is a research of new approaches, new formulations and problem unification, and finding problem solution. The main motivation is the fact that for a quality and detailed description of real problems and real systems is necessary to work with nonlinear models and with uncertainty that reflects inaccuracies in measurements, fault behavior, and better description between a model and reality than by using linear and deterministic models.

Theoretical and practical application

The outputs in the area of unknown variable estimation can be widely used in various fields, specifically in technical process modelling, industry, aviation, navigation, transportation, health system, economy, etc. For example the algorithms for position, velocity, and orientation estimation can be used in the fields of aviation and navigation for vehicles and satellites. Working with uncertainty enables to find new solutions that bring higher quality. The research outputs in the area of detection consist of algorithms for change detection of system behavior and eventually for fault detection. These algorithms find application in monitoring, identification, and control of various systems such as modern buildings (air-conditioning, heating), etc. The research outputs in the area of optimal and adaptive control consist of algorithms that provides quality system identification and control. The adaptive algorithms enables to control systems with not only unknown parameters, but also with unknown structure. In the area of information fusion, the IDM research group designs methods that provide efective treatment of information comming from number of sources (e.g. different type of sensors implemented on the devices). The data and estimation fusion represents a tool for decentralized information treatment and enables to identify and control the system even when the centralized approach is not possible to use.

Research and development aims in the fundamental research

  • Research and development in the identification of stochastic dynamic models.
      The interest is aimed both on the systems that provide the estimations in form of probability densities of state under measurement and the systems that provide point estimates of unknown variables. The efforts are mainly focused on improving a quality of estimates or lowering the computation demands.
  • Research and development in the detection of changes and faults of monitored systems.
      A goal in this area is esentially a systemathic research of a new generation of detectors that provides not only a decision about system change of fault, but generates signals that probes the monitored system, and therefore enables to increase the quality and speed of detection.
  • Research in the system control under uncertainty.
      The basic idea is to design a control laws that enable to optimal corrections of system model and maintain the desired characteristic of controlled system at the same time.

Cooperation offer

Systems for estimation, decision making and control are applied in technical and non-technical areas, i.e. for technology process control, navigation, tracking and can be also utilized in signal processing, bioengineering, in operations research, geophysics, econometrics, ecology, communication systems, for prediction of future behaviour of quantities, e.g. technology quantity, navigation trajectory, economic indicators.

The research group is able to solve the problems of fundamental and applied research in the areas of system identification, unknown parameter and variable estimation, decision making for change and fault detection, optimal control, adaptive systems for signal processing, and information fusion. Among the problems of fundamental research the research group can process and implement the outputs in form of practical algorithms for an instant use. The team can offer its capacities for cooperation on various research tasks, both with research teams at universities and research institutes and companies. The IDM research group can also work on contracts with companies and institutions (contract research). The IDM research group offers professional training and cooperation in organizing seminars.

The research group offers topics for bachelor, diploma and dissertation thesis in the mentionned areas of fundamental and applied research. The students can directly cooperate on specific projects or eventually cooperate with companies on the topics.

What we can specifically offer

The research group offers a design of mathematical model of statistic and dynamic systems through system identification, using parametric and nonparametric methods of identification. this includes cophisticated designs of filtering algorithms that enable to estimate parameters and state of the parametric models. The uncertainties are implemented in the mathematical model to provide the best connection between mathematical model and reality considering system dynamics and system measurements. The proposed estimation algorithms can even implement a priori information about unknown variables, e.g. technilogical and physical constraints of variables. This enables better similarity between model and reality and a better combination of a priori information and the information from the measurements in real time. Moreover, in the case of decentralized estimation the information fusion is a necessary element of system identification designs and should be considered as well.
The user can specify his or her demands on estimation quality. This influences a selection of the actual method, a form of the outputs, and a complexity of algorithms. A quality and detailed system model and estimation of its unknown variables respecting uncertainties of the real world are the key elements of a correct decision making, prediction and a design of control algorithms, where the quality of regulation depends on the quality of model and estimation.

Important references

  • References:
    • design of algorithms for monitoring of vehicle queue length in populated areas (ELTODO, dopravní systémy, s.r.o.)
    • development of algorithms for state and parameter estimation for cold rolling systems
    • development of algorithms for identification and state estimation in the problem of tracking of a moving object
    • long experiences of the group in research and development of algorithms for identification and estimation as published in the world respected journals and conferences
  • Partners:
    • Academic
        Uppsala University, Air Force Research Lab, FEL ČVUT v Praze, ÚTIA AV ČR, FEL ČVUT Praze, ÚTIA AV ČR, FEKT, VUT Brno
    • Industry
        AŽD Praha s.r.o., Honeywell spol. s r.o., COMPUREG Plzeň, s.r.o.