Design of a Tracking Controller for an Unmanned Bicycle Using Neural Networks

摘要

Non-honolomic constraint of a bicycle and its complex dynamical model, have made the control of an unmanned bicycle a difficult task. This paper is one of the few articles that propose an analytical approach to control the roll angle and to track a desirable path by the bicycle, simultaneously. In this paper, first, using the back-stepping method the roll angle is controlled to follow a time-varying reference. Then, using a neural network, the relation between the desired roll angle and the steering variable is modeled, and also by calculating the necessary steering variable to track an arbitrary path, a tracking controller is proposed to make the bicycle to track a predetermined path. Finally, the performance of the proposed method is shown by computer simulations.