Stabilizing Control With Obstacle Avoidance Of Nonholonomic Mobile Robots

Wheeled mobile robots (WMRs) belong to nonholonomic control systems, which are also typical strong nonlinear systems. In practice, WMRs always move to a desired point independently in the presence of obstacles. Therefore, the development of the feedback controller which makes WMR move to a desired point avoiding obstacles is promoted by the myriad of practical applications that can be addressed by mobile robots.In this dissertation, noholonomic systems, obstacle avoidance and the stabilization problems of WMRs are systematically investigated. The main contents are as follows:Firstly, after analyzing the construction of WMR, four mathematics model of WMR are summed up: the kinematic model, the dynamic model, the kinematic model described by Lie Basket and the linearization of the kinematic model.Secondly, steady obstacles description. In the environment with wall-shape or round-shape obstacle, the detailed descriptions are derived by the navigation function constructed by the mathematic method. The navigation function has only one minimum—the goal configuration. Then, the problem of local minimum in Artificial Potential Fields is solved.Thirdly, the feedback controllers, which make WMR move to a desired point avoiding obstacles, are presented. In Chapter 4, the navigation function of the obstacle environment is verified as a CLF (Control Lyapunov Function). Then, the expression of the global asymptotic stabilizing control law is obtained for the kinematic model of the WMRs. By the combination of CLF and backstepping methods, an asymptotic stabilizing controller is designed for the dynamic model of the WMRs and the actual torques control inputs are derived. In Chapter 5, based on the linearization of the kinematic model, a MPC (Model Predictive Control) controller is designed, and the real-time information of the system and constrains is used to optimize the controller on-line.The methods above are validated by SIMULINK, and it can be realized that WMR moves from an initial configuration to the origin without collision. In the first method, the actual torques control inputs would use in the actual system directly. And the second method absorbs the virtue of MPC and makes great effects. Key words: nonholonomic control system; obstacle avoidance; feedback...