Neuro-fuzzy systems for intelligent robot navigation and control under uncertainty

Abstract

This paper describes neuro-fuzzy systems for intelligent robot navigation and control under uncertainty. First, we present a new neuro-fuzzy system architecture for behavior navigation of a mobile robot in unknown environments. In this neuro-fuzzy system, a neural network is used to process range information for understanding distribution of obstacles in local regions; while fuzzy sets and a rule base are used to quantitatively formulate reactive behavior and to coordinate conflicts and competition among multiple types of behavior. Second, based on open-loop responses of a simplified model, we present a new method for designing a neuro-fuzzy controller for a manipulator with nonlinear dynamics or with unknown structure. The parameters of the fuzzy controller, related to the second-order systems, are off-line optimized, and a neural network is used to train the mapping relationship between the open-loop responses of the second-order systems and the optimized parameters of their corresponding fuzzy controllers. >