Slip And Tire Force Estimation Of Mobile Robot

This thesis introduces a method to estimate wheel slip and tire forces for a fourwheeled mobile robot on rigid terrain. The method, based on extended Kalma n filter(EKF), is also good at tracking with vehicle's dynamic states such as body velocity andangular accelera tion. The vehicle dynamic structure is modeled in a universa l way tosuitable for most four wheel driver robot equipped with an IMU, wheel encoders, andinclinometer. With adding pitch rate into dynamic model, it can be more preciselyreflecting the truth mecha nism of vehicle running. Such a dynamic model is also suitablefor a complia nt framework or suspension structure of mobile robot.An empirica l tire-terrain model from Pacejka's Magic Formula is introduced here torepresent for the truth model of contact patch force. Simula tion results show that EKFalgorithm is successful tracking longitudina l and latera l forces when robot is traveling onthe flat surface. And the estimated line covers slip-force curves effectively. However,parameters in ma gic formula are determined by norma l force. Once pitch rate changing,the norma l forces acting on front wheels and rear wheels are various from time to time.EKF is also proved to be converged and a good reflection of slip-force relationship. Slipdetector algorithm proposed in this thesis is not only suitable for flat surface conditionbut also verified on slope surface, which mea ns future applica tion in complia ntframework mobile robot.Great efforts also worked on determining measurement sensor covariance forapproaching to a real system. All simulation parameters are close to hardwareconfiguration. The methodology in this thesis is demonstrated effective and valid insimulation. All the proposed work in this thesis can provide a suitable foundation forfuture works.