Two-Wheeled Differentially Drive Mobile Robot Motion Model And Control Research

Along with the computer, control and artificial intelligence technology dramatic changing, the application of TWDDMR(Two-wheeled Differentially Driven Mobile Robot) is turning into more and more widely and its intelligent degree is becoming higher and higher.As the applied fields of TWDDMR continuing to expand, it will be more work in a dynamic, unknown and unstructured environment. It puts forward higher request for motion control of the system. TWDDMR is a kind of typically nonholonomic system, and its motion control problem can be divided into three categories: path tracking, trajectory tracking and point stabilization. Trajectory tracking control is the basis of TWDDMR motion control and hot spot of the research, and it has extremely important applications in many industry and broad prospects. But TWDDMR system is a complicated multiple input multiple output(MIMO) nonlinear system, and also as a kind of typical nonholonomic system, which has time-varying, strong coupling and nonlinear dynamic characteristics, and TWDDMR has the influence of uncertain factors such as load change, friction, road conditions and interference in practice, which will bring great difficulties to TWDDMR trajectory tracking control.Aiming at the trajectory tracking control problem of the TWDDMR, the main research of this paper are as follows:① Putting forward a trajectory tracking controller of TWDDMR, which has a Novel Two Stage and Double Layer Structure(NTSDLS).The controller is composed of two stage cascaded structure, the cascaded structure includes kinematic dynamic parameter adaptive and dynamic decoupling control. And structure of each stage is composed of two layer structure respectively.1)Kinematic dynamic parameter adaptiveAiming at the varying load, strong disturbance and uncertainties and time-varying characteristic of the systems in the process of the real-time motion control of TWDDMR,the kinematic controller of the Back-stepping control parameters self tuning is proposed in this paper,which is based on the improved RBF neural network.As the NTSDLS trajectory tracking controller of the outer ring,it plays a leading role,mainly used for high precision tracking.Kinematic dynamic parameter adaptive level is composed of RBF neural network layer and Back-stepping controller.Specifically, is the first use of integral Back-stepping method to design the kinematic controller, and using the improved RBF neural network to control the gain of the self-tuning,overcome the change of load and strong disturbance load and other uncertain factors to control the effect of optimal controller.2)Dynamic decoupling controlFor the multi variable, MIMO and strong coupling and other dynamic characteristics of TWDDMR, and put forward the adaptive PID dynamic decoupling controller based on the improved RBF neural network, and it as the inner loop of whole NTSDLS trajectory tracking controller, is used to overcome the negative effects of nonlinear, strong coupling TWDDMR system dynamic characteristics of the system.Dynamic decoupling control is composed of RBF neural network identifier and controller of PID two layer structure.Specifically, through using SR-UKF algorithm, starting from the perspective of observation, to design SR-UKF state online observer, and then online estimating the amount of state variables of TWDDMR system that can not be measured directly, and then through the output equation providing output data of TWDDMR system for RBF neural network identification system. The RBF neural network identifier is used to identify the nonlinear time-varying information of the TWDDMR system, and to automatic adjustment of PID control parameters, so as to obtain the corresponding control system variables, to eliminate the coupling effect between the variables, the final realization of the intelligent decoupling control in TWDDMR system.3)Describes the structure and principle of the control system, and the stability analysis shows that the application of Lyapunov theory on criterion of stability analysis controller is proposed in this paper.② Established TWDDMR equivalent coupled motion model and conducted optimization on it, as well as put forward state-parameters online joint estimation method of algorithm model based on the SR-UKF.This paper carried out supplement and optimization for class equivalent coupling motion model of TWDDMR. In addition, it conducted supplement and deduction in the situation that is not considered each drive motor system in the nonlinear saturation lower limit workplace at the early time and set up a more perfect, more press close to actual system class equivalent the coupling movement model of TWDDMR.According to previous least square method and genetic algorithm identification methods need get plenty of measurement data for off-line identification model parameters, resulting in the identification period for a long time and big workload. It took the SR-UKF method to identify the status parameters online for the optimization class of equivalent coupling motion model of TWDDMR, real-time estimate the unknown time-varying parameters, and improve the information of the current model for the system. These works made the mathematical model can more accurately reflect the actual situation of the system.③ Set up simulation experiment platform based on the class equivalent coupling motion model of TWDDMR, and conducted simulation experiment of trajectory tracking control for TWDDMR.Through a series of comparative experiments, which proved that NTSDLS controller can well inhibit uncertainties of the model parameters and the influence of external disturbances on the system, ensure the stability of the closed-loop system, and realize intelligent decoupling control implementation for TWDDMR system. It made the TWDDMR system after decoupling have good dynamic and static characteristics, as well as well anti-interference system characteristics and strong robustness and well tracking performance for TWDDMR system. The experiment proved that NTSDLS trajectory tracking controller that is presented in this paper is effective and correct.