Robust backstepping and neural network control of a low-quality nonholonomic mobile robot

A robust motion controller based on neural network and backstepping technique is proposed for a two-DOF low-quality mobile robot (MR). There are two main components in the motion controller. One is the tracking controller, which guarantees the MR follows the reference trajectory; the other one is the wheel-level inverse NN controller, which compensates the dynamics of the MR. Simulation results are provided to validate the proposed controllers. Experiments with a real low-quality MR, which were built from cheap drivelines, have been used to verify the effectiveness and robustness of the motion controller. (C) 1999 Elsevier Science Ltd. All rights reserved.