Wheeled Mobile Robot Based On Neural Networks Robust Sliding Mode Control

In recent years, the theory and application research of nonholonomic mobile robot has gained more and more attention in the control community. Generally speaking, any mechanical system with wheeled mobile driving device such as lorry belongs to the category of nonholonomic robots. So it has a broad real application background. On the other size, because this kind of system is constrained by the nonholonomic non-slipping and pure rolling conditions, the control of nonholonomic system becomes very difficult. One can't realize system's asymptotic stability by continuous or integrable pure state feedback. Moreover, one can't also realize system's global linearization by nonlinear transformation. Consequently, the research of robust control of mobile robot has great significance in areas of theory and application. Analysis, kinematic and dynamical modeling and trajectory tracking controller designing for a mobile robot are studied in this thesis.Firstly, a kind of kinematics model of two-wheel driven mobile robot is analyzed. Aimed at its trajectory tracking control problem, the sliding-mode controller is designed. The effectiveness of the proposed controller is shown through the simulation of two-wheel driven mobile robot tracking circularity and surve trajectory under the controller.Secondly, the review of producing cause, weakened methods of chatering in sliding-mode control. According to the physical significances of the reaching law parameters and the qualitative relationships among them in variable structure control are analyzed. A reaching law control method is put forward based on fuzzy rules. On the aim of improving control quality, the reaching law parameters are selected by fuzzy rules.Finally, nonholonomic mobile robot system is a complicated control system, and during its running, there are various uncertainties, which do great effects on its performance. Usually, the normal control strategies cannot get ideal results. Neural network (NN) has recently applied to mobile robot control due to its universal mapping property, and the result will be satisfactory. So, we design a trajectory tracking algorithm based on NNs, considering both kinematic and dynamical model, and the simulation results demonstrate that this algorithm has good robustness.