Bayesian Toolbox for Dynamic Distributed Decision Making

The paper outlines a new challenging area of control, namely distributed decision making under uncertainty. Presence of many decision-making units, called the participants is the main distinction from the classical control theory. The problem is analysed from Bayesian point of view. Each participant has the abilities of classical adaptive controller, i.e. learning of the environment and design of the control strategy. What is new is the ability to communicate with its neighbours. The effect of this ability on learning and control strategy design is studied in the paper. Methodology for dealing with the problem is introduced.