Research On Data Driven Model Free Adaptive Control Of An Omnidirectional Mobile Robot

In recent years,mobile robots have been used more and more widely in various fields such as civil,military,and commercial.Among them,Due to the independent and simultaneous translational and rotational movements,omni-directional mobile robots are more suitable for narrow spaces where mobility is required,such as factories,warehouses and hospitals.Therefore,how to realize the accurate,reliable and stable trajectory tracking control of omnidirectional mobile robots has become one of the research hotspots in the field of robotics.In this paper,the omni-directional mobile robot is taken as the research object.However,robots modeling has several problems,such as complex modeling process,time-consuming parameters calibration process,model uncertainty,external disturbance,etc.Therefore,this paper takes model free adaptive control as the main method,and studies the trajectory tracking problem of omni-directional mobile robot without any mathematical model and parameter information.The main research contents are as follows:Firstly,this paper briefly introduces the omnidirectional mobile robot platform built by the laboratory.And in order to better verify the effectiveness of the proposed control method in simulation,the kinematics and dynamics model of the omnidirectional mobile robot is derived.Then,due to the complex and time-consuming robots modeling process,the existence of uncertainties in models and the existence of external disturbances,a new model free adaptive control method is proposed.Specifically,a new rotating coordinate is first defined to apply the model free adaptive control method for the wheeled mobile robot.Then the coordinate transformation matrix and terminal sliding mode control are introduced in the typical model free adaptive control method to ensure the system convergence.Then the stability analysis of the proposed control method is carried out.Note that,a method for setting the initial value of pseudo-Jacobian matrix is given with explicit physical interpretation.In addition,the effectiveness of the proposed method is valued according to simulation and experiment.Finally,in view of the problems of model-free adaptive control methods: pseudoJacobi matrix estimation errors and low-order data models cannot contain all nonlinear dynamics,a model-free adaptive control method based on neural networks and concurrent learning compensation is proposed.That is,a low-order dynamic linear data model is first established,and then a neural network is used to estimate the pseudo Jacobian matrix and the non-inclusive nonlinear dynamics.Then a model-free adaptive controller based on neural network and concurrent learning compensation is given.In addition,the stability of the proposed control method is analyzed,and the effectiveness of the proposed control method is verified in simulations and experiments.