Developmental cognitive map learning and behavior acquisition by a mobile robot through a self-organizing incremental neural network

Needs for making of the robot intelligence have risen. The acquisition of the symbol is one of the most important tasks in the process of making of the robot intelligence. A lot of methods of the intelligent robot´s using the neural network in the process of an autonomous problem solving through the interaction with environment are proposed. In the method based on these functional approximations, a huge computational complexity is required to learning a higher dimension input like the image sensor. Moreover, it doesn´t correspond to an additional input. Self-organizing incremental neural network (SOINN) is robust in a higher dimension and additional input. e propose the model by which the change in the action and the perception input is learned based on SOINN in the form of the if-then rule as a symbol. We confirmed the effectiveness of the proposed method by the navigation experiment which human specified the state of the goal. The outcome of an experiment showed the possibility of the achievement of the mobile robot based on the pattern base reasoning.