Design Of Tensor Product Model Transformation Controller For Mobile Two-wheeled Robot

Two-wheeled robot is one of the classic models in underactuated systems,which has the characteristics of multivariable,unstable,nonlinear and strong coupling.Therefore,two-wheeled mobile robots are used as one of the typical platforms to verify the quality of control algorithms.In this thesis,the two-wheeled robot is taken as the controlled object,the quasi-linear parametric variable model of the robot is extracted,the research on the tensor product fuzzy control method is carried out.The main work of this thesis is as follows:1.A nonlinear disturbance model of a wheeled robot is constructed.Aiming at the disadvantage of low robustness of the controller designed by the fixed-point linearization model,the physical modeling of the two-wheeled robot is carried out by using the Lagrangian modeling method.At the same time,the cases where the left and right wheels of the robot are subject to the same disturbance and different disturbances are considered,the quasi-linear parametric variable model of the robot is obtained.2.Design of variable universe interval type II fuzzy logic controller based on tensor product(TP).Aiming at the problem that the fuzzy control method has many fuzzy rules and large amount of calculation,the controller is designed based on the linear variable parameter technique and the variable universe interval type II fuzzy method.Linear matrix inequalities are used to account for variable error and error derivative gain,which simplified controller design.It can be seen from the simulation results that the improved control method has the advantages of fast response and short stabilization time.3.Design of variable structure tensor product fuzzy controller.The equidistant sampling method of TP model is easy to cause the loss of local extreme value information.The convex hull manipulation method of the entry function partition is refined by adding local extrema to the model with variable input space,that is,the method is designed based on the variable input quasi-linear parametric variable state space model,which reduced the amount of calculation based on the high approximation model.At the same time,a non-fixed time step sampling method is designed for TP model transformation,the Hammersley sampling method and the uniform design method are optimized respectively by the minimum volume simplex method,so as to realize the design of the non-uniform sampling method.It can be seen from the simulation comparison results that the newly control method has strong robustness.4.Design of optimal path tracking controller in virtual reality scene.Existing distributed compensation controllers use linear matrix inequalities to find feasible solutions,the optimization of controller parameters has not been studied in depth.The gain parameters of the distributed compensation controller based on TP model transformation are optimized through the commonly used bionic algorithm,it is showed that the ability of the algorithm to optimize globally.Then,the parameter optimization interface of newly controller is built to improve the convenience of the optimization algorithm.It can be seen from the simulation comparison that the optimization algorithm improved the control performance of the controller.It can be concluded that the path tracking speed is the fastest and the displacement overshoot is the smallest.5.A test platform for two-wheeled robot is built based on the Raspberry Pi,the above-mentioned improved and designed control method is experimentally verified.At the same time,the 3D modeling of the Googol’s two-wheeled robot was completed,the effectiveness of the new controller was verified through the virtual reality scene.