Research On Visual Servo Trajectory Tracking Of Mobile Robots

With the development of the mobile robot technology,more and more fields such as industry,service,medical and domestic use wheeled mobile robots to complete various tasks.Due to the nonholonomic constraint of the wheeled mobile robots,the design of the controller is quite difficult and challenging.In recent years,using visual information to control the movement of the robot,namely visual servoing,has become very popular in the field of robot control for its many merits like low cost,flexibility,contactless.Visual servoing tracking control of wheeled mobile robots leads the robot to move along a given reference trajectory to the position and posture of the target.It can be easily combined with other methods such as trajectory planning,obstacle avoidance,feature preservation and so on.Although there are already many research results in the domain of visual servoing trajectory tracking of wheeled mobile robots,but the practical application of this kind of task still has many problems like the missing of the scene depth due to the imaging model of the camera and the unknown intrinsic and extrinsic parameters of the camera-robot system.This thesis designs relevant control strategies to deals with the aforementioned problems in the visual servoing trajectory tracking control of mobile robots.Concretely speaking,the works in this thesis are:(1)Aiming at the visual servo traj ectory tracking task of mobile robots,a control strategy which can simultaneously identify the depth information of the scene is proposed.Because of the inherent nature of the camera imaging principle,it is impossible to recover the depth information of the scene from the image information.This part studies how to obtain the three-dimensional information of the scene from the two-dimensional image information to complete the reconstruction of the three-dimensional information of the scene.This method,using a concurrent learning framework,can recover depth information during the operation of the robot through a series of historical and current image data by the designed adaptive update law.Compared with the traditional visual servo trajectory tracking method,this method overcomes the defect of lack of depth information.The Lyapunov method is used to prove that the system consisting of trajectory tracking control module and depth identification module is stable as a whole.The simulation and experimental results prove the validity of the method.(2)This dissertation presents a visual servo trajectory tracking controller with unknown translation parameters between camera and robot.Firstly,the kinematics model of the visual servo system is determined when there are translation parameters between the camera and the robot.Then,the camera pose information is obtained by decomposing homography matrix and tracking errors are designed.The adaptive control laws are used to compensate for the unknown depth information and extrinsic parameters information,and the designed controller can drive the robot complete the trajectory tracking task.Lyapunov method proves the stability of the control strategy.The validity of this method is verified by simulation and experiment results.(3)A non-calibration method of camera parameters for trajectory tracking task of mobile robots is proposed.In the scenario of trajectory tracking task,the method firstly chooses an initial estimated value of focal length,calculates the expected and current projective homography matrices by using this the estimated value of' focal length,then decomposes the projective homography matrix to obtain the coordinate component of the plane normal vector of the feature point.After that,the method obtains the non-linear equation of cost function according to the relevant properties,and solves the non-linear equation to get the focal length of the internal parameter matrix.In the case that there are errors between the principal coordinates and its true values,it can be seen that the calculated focal length is within the acceptable error range.The simulation results show that the method is practical and effective.