Synthesis of fuzzy, artificial intelligence, neural networks, and genetic algorithm for hierarchical intelligent control-top-down and bottom-up hybrid method

Autonomous robots, which perform tasks without human operators, are required in many fields. They have to be intelligent to determine their own actions in unknown environments by themselves based on sensory information. In advance, human operators can give the robots knowledge and skill in top-down manner, but when the robots perform tasks in unknown environment, the knowledge may not be useful, In this case, the robots have to adapt to their environments and acquire new knowledge by themselves through learning. This process proceeds in bottom-up manner. This paper introduces a control scheme for autonomous robots, hierarchical intelligent control. It consists of three levels: adaptation level, skill level and learning level. To link the three levels, the scheme uses artificial intelligence (AI), fuzzy logic, neural networks (NN) and genetic algorithm (GA). Each technique has advantages and disadvantages. In order to overcome the disadvantages, this paper introduces synthesis techniques of them. Those are key techniques for intelligent control of robots. This paper describes advantages and disadvantages of each technique, and explains how to construct the hierarchical intelligent control.