Supervisory Fuzzy Gaussian Neural Network Design for Mobile Robot Path Control

This paper aims to propose an efficient control algorithm for the mobile robot path control. A supervisory fuzzy-Gaussian-neural-network (SFGNN) controller is proposed. This controller includes a fuzzy-Gaussian-neural-network (FGNN) controller and a supervisory controller. The FGNN controller is constructed in a form of neural network with a Gaussian-type fuzzy membership function; and the parameters of the membership function are on-line tuned by the derived adaptive laws. The supervisory controller is designed to compensate for the approximation error between the FGNN controller and an ideal controller. This combination of FGNN controller and supervisory controller can achieve favorable control performance and can reduce the required neurons and tuned-weights of the SFGNN control system. The path control of mobile robot for two paths, a circular path and a square path, are used to test the effectiveness of the proposed design method. Simulation results demonstrate that the proposed SFGNN controller can achieve better control performance than an FGNN controller and a PID controller for the mobile robot path control.