Adaptive systems are dynamic units that learn their environment while make their decisions. Within this broad framework, the main research areas of the department are:
Dynamic decision making (DM) maps knowledge into DM strategy, which ensures reaching DM aims under given constraints. Under general conditions, Bayesian DM, minimizing expected loss over admissible strategies, has to be used.
Advanced control strategies based on LQ and predictive algorithms are significant for different industrial applications. The aim of the research is a fixing of control theory in this area and developing of complete computer-aided design of adaptive controllers. The design arises from raw data, measured on a real controlled system; user's knowledge; and user demands and it results into a completely pre-tuned and verified controller.
Sampling methods are known for being computationally expensive, however recent research and increasing performance of computers improved applicability of these methods in such a way that they represent a strong alternative to traditional approximation methods.
Linear systems form a well-developed core of advanced controllers. Consequently, their understanding and even minor improvements have a deep impacts on the field.
The general objective of the project is the enrichment of the complete design line of LQG controllers so that it will cover steps related to state estimation, ideally with mixed-type (continuos and discrete) states.
This research deals with Bayesian learning using models with bounded noise and physically constrained quantities.
Investigated Research Topics:
Knowledge extraction maps extensive data sets on lower dimensional objects. Its results always serve to a subsequent, often dynamic, decision making. Decision-making quality is substantially influenced by the mapping used. This simple fact is relatively rarely respected by many elements in the overwhelming arsenal of existing mappings. A complete solution of decision making problems that includes explicitly the discussed mapping are severely limited by computational complexity.
Complex technical and societal systems are often managed by human beings (operators, managers, medical doctors ...) who badly need help to reach high standards of their acting. Conceptual solution, formalization, algorithmization and implementation of such advising systems have been addressed. The resulting system is able to cope with dynamically changing incompletely known multi-attribute environment, to learn and optimize dynamic decision-making strategy realized either by human being or automatically.
Single decision-making unit like the advising system or non-linear adaptive controller reach relatively soon its applicability barrier, mostly caused by computationally complexity and limited reliability. Then, a distributed solution is needed. The theory and algorithms covering design and cooperation Bayesian decision-making units (participants) are inspected. They respect limited abilities of such units, incomplete knowledge and random nature of the surrounding environment. The problem is scientifically challenging with an extreme applicability width.