Stabilization Of Nonholonomic Mobile Robots With Active Disturbance Rejection Strategy

Wheeled mobile robot has attracted a large number of researches over the years for its simple mechanical structure,high flexibility and good operating performance.Due to the nonholonomic constraints,the standard nonlinear control methods are not suitable for the stabilization of the mobile robot.Furthermore,the working environment of most mobile robots is unstructured where has various unknown disturbances.The mobile robot system itself also has several problems such as model uncertainties and non-ideal skidding and slipping.How to realize the stabilization of the mobile robot under the condition that the nonholonomic constraints and various uncertainties exist in the system? This has been a research hotspot in the field of robotics and nonlinear control theory.Active Disturbance Rejection Control(ADRC)is a method to overcome system nonlinearity,large-scale uncertainty and external disturbances.It has inherited and carried forward the advantage of model-free from PID.The nominal integral chain of ADRC transcends the boundaries of linear and nonlinear,time-invariant and time-varying systems.The concept of total disturbance is defined that lumps the vast kinds of uncertainties,including all the internal uncertainties and external disturbances of the system.Extended State Observer(ESO)as the key part of ADRC is designed to estimate the total disturbance.In recent years,lots of theoretical researches and practical applications have shown the promising performance of ADRC.However,ADRC is mainly used in single-input single-output systems and multi-input multi-output systems with equal input and output dimensions.The research of ADRC in the nonholonomic systems with different input and output dimensions is still open.In this thesis,three components of ADRC including Tracking Differentiator(TD),ESO and Nonlinear State Error Feedback(NLSEF)are used to deal with the nonholonomic constraints and uncertainties from different perspectives.The nonholonomic mobile robot can be asymptotically stabilized,and the application scope of ADRC is expanded.The main research works are summarized as follows:(1)The motion planning method for nonholonomic mobile robots based on the differential flatness is restricted by a fixed time condition.Moreover,it is always used for tracking control of the mobile robot systems.For the uncertain mobile robot system,a motion planning method using TD is proposed to transform the stabilization problem into a tracking problem,which overcomes the impact of nonholonomic constraints.First,TD serves as a transient profile generator for the step reference input of an ideal mobile robot system.A smooth desired trajectory is generated by the TD which converges to the desired pose asymptotically and is free from the fixed time constraint.Afterwards,an ESO-based tracking error feedback controller is synthesized for an uncertain mobile robot to tracking the desired trajectory.Then the mobile robot will converge to the desired pose asymptotically.(2)Nonholonomic mobile robots suffer from the dead-reckoning problem caused by the internal sensor such as wheel odometry.An useful methodology is to directly apply an external visual sensor to obtain the visual information and perform the feedback control.It also provides a new solution from the perspective of visual features to deal with the problem that the dimensions of the system inputs and outputs are unequal.The external visual sensor is effective to increase the robustness of the closed-loop mobile robot system to the environment.While it brings the system various uncertainties because of imprecise or even uncalibrated camera parameters and the lack of depth information.Therefore,a new image-based visual stabilization of nonholonomic mobile robots is proposed with both uncalibrated intrinsic and extrinsic camera parameters.First,system states are defined in the image plane by a coordinate transformation of the feature points.The kinematic system model with equal input and output dimensions is developed.Then,a dual-loop control structure is presented.In the inner loop,a generalized linear ESO is designed to estimate the system uncertainties caused by the uncalibrated camera parameters and unknown depth.Then a nonlinear switching controller is established in the outer loop to make the system errors piecewise asymptotically converge.Last,simulations and real-world experiments are given to validate the proposed scheme.(3)Uncertain nonholonomic systems with two inputs represent a large class of mechanical systems including mobile robots.Most existing researches on the perturbed chained form nonholonomic system require precisely known boundary functions of the uncertainties.Moreover,the external disturbance is not considered.Therefore,an ADRC strategy is proposed for a class of uncertain nonholonomic systems with both drift uncertain nonlinearity and external disturbances.First,the input-state scaling technique is used to obtain a controllable lower triangular subsystem which is a integral chained form with drift uncertain nonlinearity.The second transformation is carried out for the subsystem to achieve the nominal integral chain with the equivalent total disturbance.Then the ADRC design procedure is well developed for the uncertain subsystem.Afterwards,the performance analysis of the closed-loop system indicates that the subsystem and the ESO can be asymptotically stabilized by tuning the bandwidth of the ESO.