Flexible intelligent system based on fuzzy neural networks and reinforcement learning

Intelligent systems have been proposed for control, recognition, man machine interfaces and other applications. In order to apply intelligent systems, the system must have the flexibility for environmental change and tasks. Recently, fuzzy systems, neural network applied systems and heuristic approaches have been utilized in intelligent systems. These systems are eagerly researched by many researchers. However these intelligent systems without hierarchical structure of intelligence would be difficult to adapt for environmental changes and various tasks. The flexible intelligent system proposed is based on a hierarchical intelligent system architecture. The top layer generates the control trajectory or strategies, the middle layer manages the skills of the tasks, and the bottom layer handles the controlled objects.<<ETX>>