Visual Servo System Design Of Mobile Robot

In this paper,we apply the vision sensor to the robotic system design.Using the visual information obtained by the visual sensor to complete the close-loop control of robot is called visual servo.Vision servo involves many disciplines such as robotics,digital image processing,control theory,and electronic information,etc.And it is one of hot research topics,which is the basis for robot positioning,robot navigation and other technologies.In this paper,we mainly study the image-based visual servo,which uses the visual information obtained by the airborne camera to realize the function that the airborne camera can point to a fixed target with high accuracy during the movement of the mobile robot.In this paper,we firstly introduce the hardware platform of mobile robot vision servo.And then two methods of image detection are introduced,which are circle detection based on RGB color space and Hough transform,and real-time comparison experiment of two methods is conducted.A camera model is established and the relationship between the target's error in the image plane and its position in the 3-dimensional space is deduced.Then the camera is calibrated based on MATLAB and the camera's internal parameters are determined.After that,the system identification of the robot joint is conducted by using the least squares method and subspace method,and we obtain the transfer function of the object.Due to the high order of the model,model reduction is performed.Then the double loop model of the visual servo system is introduced.The outer loop is the image loop,which uses image information to close the loop.The inner loop is the robot joint and the encoder data of the joint is used to close the loop.In order to improve the response speed and control accuracy of the inner loop,an output feedback and PI hybrid controller is designed for the identified model of position loop.The image loop uses a PI controller,and suitable controller parameters are obtained through experiment.Based on this,since the visual servo system is a dual-rate control system,dual-rate high order holds is designed.In addition,because the mobile robot is always in motion,there is a steady-state error in the visual servo system.To eliminate the error,we design the motion compensation scheme through the mechanism modeling method.Finally,the relevant experimental verifications are carried out on the self-built mobile robot hardware platform,and the visual servo of the mobile robot is realized.The results show that the control scheme designed in this paper can guarantee the high pointing accuracy of the airborne camera,and its performance is better than the traditional PID controller.