Optimization of a Neural Dynamics Based Controller for a Nonholonomic Mobile Robot Using Genetic Algorithms

In this paper, a neural dynamics based controller for a nonholonomic mobile robot is proposed. The turn angle of the robot in the proposed model is characterized by a biologically inspired shunting equation derived from Hodgkin and Huxley’s membrane equation. This model is capable of generating smooth steering velocity command that drives the robot to track desired paths. Some parameters in the proposed neural dynamics based controller need to be selected. A genetic algorithm is designed to optimize the model parameters that can guarantee the convergence of tracking error of the mobile robot. Simulation studies of a fourdegree-of-freedom mobile robot are conducted, which demonstrate the effectiveness of the proposed motion controller.