HYBRID NEURAL NETWORK CONTROL OF MOBILE ROBOT SYSTEM VIA ANTI-CONTROL OF CHAOS

Zahra Yaghoubi, Hassan Zarabadipour

Keywords

Neural network controller, Anti-control of chaos, Mobile robot system, Feedback linearization controller

Abstract

In this paper, a new method for controlling a mobile robot system is proposed. Utilizing chaotic features, this paper presents a new method to synchronize a mechanical system with a chaotic system. To optimize the energy consumption and reduce the costs in this method, a hybrid neural network controller is designed using anti-control of chaos. The objective of this paper is to introduce a method in which a mobile robot system is being controlled through chaotification via neural networks as its identifier and controller. First, the feedback linearization controller is used to trace the desired path, and then in parallel with this classic learning controller, the neural network is used. Finally, this neural network is replaced and trained on-line using the back propagation algorithm with sample data collected from feedback linearization controller. In this study, anti-control of chaos means that the tracking error, instead of assuming to approach zero, is synchronized with small ratio of chaotic gyroscope system amplitude. To show how the feedback linearization and neural network are applied, the trajectory tracking in presence of Noise is examined and the performance of the proposed method is estimated through simulation and cost function calculation.

Important Links:



Go Back