A new intelligent approach for mobile robot navigation

Abstract

In recent times computational intelligent techniques such as fuzzy inference system (FIS), artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS) are mainly considered as effective and suitable optimization methods for modeling an engineering system. In this paper an efficient hybrid technique has been applied for mobile robot navigation using multiple adaptive neuro-fuzzy inference system (MANFIS). ANFIS has taken the advantages of both fuzzy inference system and artificial neural network. First, we design an adaptive fuzzy controller with four input parameters, two types of output parameters and three parameters each. Next each adaptive fuzzy controller acts as a single Sugeno-Takagi type fuzzy inference system where inputs are the different sensor based information and output corresponds to the velocity of the mobile robot. The implementation of the proposed navigational controller is discussed via numerous simulation examples. It is found that such an adaptive neuro-fuzzy controller is successfully and quickly finding targets in an unknown or partially unknown environment. © Springer-Verlag 2013.