Research On Model Predictive Control For A Wheeled Mobile Robot

As wheeled mobile robots(WMRs)have strong adaptability in unknown environments,researchers have paid more and more attention to the application of WMR in real life.Since WMRs are usually used in the complicated and uncertain environment,intensive research efforts have been devoted to improving the performance.In related research topics,the motion control is one of the most critical and basic issues since it is essential for all kinds of tasks.In addition,the WMR is a kind of typical nonholonomic and nonlinear system with multiple inputs and multiple outputs and strong couplings.Because of these characteristics,many classical control methods cannot be directly applied to nonholonomic WMRs.Therefore,there still exist many research challenges for the motion control of WMRs.In this thesis,according to the current research on WMRs,a comprehensive literature review on the development and motion control of WMR is given.We present the kinematic model of the WMR,and then design the nonlinear model predictive controller.Finally,the proposed controller is tested on the experimental platform-Qbot developed by Quanser Company.The main results of this thesis are as follows:(1)The uncertain and nonholonomic constraints of the WMR are introduced.By establishing the kinematics model of the WMR in Cartesian coordinate,the tracking error dynamic model of the WMR is obtained.(2)The nonlinear model predictive control(NMPC)algorithm is designed based on the error dynamics of the WMR.The stability of the closed-loop system is analyzed.(3)The modified continuous residual algorithm(C/GMRES)is employed to reduce the computational load of the NMPC algorithm.Since the C/GMRES method cannot be directly used for the constrained optimization problem,in this work,the original optimization problem arising from the NMPC problem is transformed to an unconstrained problem by using the barrier function method,and the C/GMRES algorithm is employed to solve the unconstrained optimization problem effectively.Based on the Quanser Qbot model,the NMPC tracking algorithm is designed,and then the modified C/GMRES method is employed to solve the NMPC problem.The simulation results show that the algorithm shows good performance and the required computation time is much lower than the sampling time.(4)The local path planning and the global path planning methods for the WMR are also studied.For each panning method,three different obstacle avoidance methods are tested on the Quanser Qbot platform;advantages and disadvantages for each obstacle avoidance method under different path planning strategies are analyzed based on the experimental results.When there are unpredictable dynamic obstacles in the known and static environments,the hybrid obstacle avoidance experiment is performed by combining the local and global path planning methods,and finally a smooth security path is obtained.The experimental results show that hybrid obstacle avoidance can follow the predefined global path.