Implementation of a neuro-fuzzy sensor based robot navigator

This paper presents a fuzzy based neural network navigation method, which utilizes fuzzy logic and neural network learning for navigation of mobile robot in uncertain environments. There are five fuzzy logic controllers represent the five basic motion behaviors. These five fuzzy logic controllers generate training data for five neural networks called behavior controller. In order to select an appropriate behavior controller at the present instant, an additional neural network, called behavior selector, concludes the current situation and selecting different behavior controller to perform different action. The effectiveness of the proposed method is verified by a series of MATLAB program simulations. Copyright (C) 2001 IFAC.