Self-organizing control based on strategy diversity

In order to identify the most suitable learning parameter which adapts to the dynamic environment, we define two parameters. One is a strategy diversity which means an average of the summation which is calculated from the difference of the strategy between all agents. Another is a self-organizing velocity which means a time average from the maximum value of the strategy diversity to the convergence value of that when the strategy diversity is reduced. By using these parameters, we realized to identify the most suitable parameters for the behavior which adapts to the dynamic environment in design technique of the! distributed autonomous robotic systems.