Fuzzy-net control of non-holonomic mobile robot using evolutionary feedback-error-learning

Abstract

Standard approaches to non-holonomic control design deal only with the kinematic steering system, ignoring the actual vehicle dynamics. Recently a stable control algorithm that considers the complete vehicle dynamics has been developed using back stepping kinematics into dynamics. According to this approach the dynamics of the actual cart has to be completely known. However, exact knowledge of the mobile robot dynamics in many cases is unattainable. A solution to this problem requires implementation of robust-adaptive control methods combining conventional approaches with new learning approaches in order to achieve good performance. This paper deals with the control problem of a non-holonomic wheeled mobile robot in the tracking mode. A locomotion control structure based on the integration of a kinematic controller and an adaptive fuzzy-net torque controller is presented. An evolutionary feedback-error-learning method for automatic elicitation of knowledge in the form of fuzzy if-then rules is developed. The proposed adaptive fuzzy-net torque controller can deal with unmodeled bounded disturbances and/or unstructured unmodeled dynamics of the vehicle. The control architecture developed is simulated and its effect on the trajectory tracking performance of a non-holonomic mobile robot cart is evaluated.