Adaptive Visual Servoing Tracking Control For Mobile Robots

The working environment of mobile robots is complex and diverse.Although visual sensors can be used to obtain rich information,how to use the feature information extracted from the captured environment images to realize the pose control of the three-dimensional space of mobile robots is a complex problem.The lack of depth information in monocular camera images makes it difficult to reconstruct complete 3D pose information and motion information through image features.This creates uncertainty in the system model.On the other hand,a mobile robot is an underactuated system with nonholonomic constraints,and its controllers are mostly designed assuming that all states can be accurately measured.Considering that vision sensors bring uncertainty to the system,it brings greater difficulties and challenges to controller design.Based on the adaptive control method,this paper studies the visual servoing tracking control problem of mobile robots.The main research work and contributions are summarized as follows:Firstly,a new tracking control method based on adaptive dynamic programming is proposed to solve the optimal control problem of visual servoing trajectory tracking for mobile robots.The image of the coplanar feature points is captured by the camera mounted on the mobile robot,so as to obtain the current image,the desired image and the reference image information.Using the homography technique,the current pose information and the desired pose information(ie translation and rotation angle)of the mobile robot are obtained,and then the open-loop error signal of the system is constructed by the difference between the current and the desired translation and rotation.And combined with the kinematics model of the mobile robot,the open-loop error system model is obtained as a timevarying affine nonlinear system.Next,for the optimal control problem of the visual servoing trajectory tracking system of mobile robots,an evaluation neural network approximation function is used to continuously learn to approximate the solution of the Hamilton-Jacobi-Bellman(HJB)equation.There is a time-varying term,which causes the HJB equation to also contain a time-varying term.Therefore,a neural network approximation function with a time-varying weight structure is designed.Finally,using the Lyapunov stability theory,the convergence of the neural network weights and the stability of the closed-loop system under the control method proposed are proved.Secondly,the control problem of visual servoing adaptive tracking control with input saturation constraints.By introducing a class of continuous bounded functions,it is applied to the design of the controller to solve the problem of control input saturation constraints.By using the feature point image information captured by the camera on the mobile robot,the three-dimensional pose of the mobile robot is first reconstructed.The homography technique is used to obtain the current pose and expected pose information of the mobile robot.Then,the mobile robot tracking control system is modeled.However,since the depth of the image is unknown,the 3D pose reconstructed using image information is a scaled pose.Aiming at this problem,an adaptive update law is designed to compensate the unknown depth and improve the control performance of the system.Furthermore,an adaptive visual servoing tracking control method is proposed for the established open-loop error system.By appropriate choice of control parameters,the control input signal can always be kept within the saturation constraints.Finally,the Lyapunov stability theory is used to prove the stability of the closed-loop system under the action of the proposed control method.