Research On Segmented Visual Servo Method Based On Six Axis Manipulator

Under the background of manufacturing industry's pursuit of flexible production,agile manufacturing and intelligent manufacturing,the traditional industrial robot's design of production process by means of artificial teaching has been unable to meet the requirements.Visual servo is a feasible scheme,which adds a visual sensor to the robot for real-time feedback.Visual servo is mainly divided into position based and image-based visual servo.Position based method is sensitive to model error,while image-based method is easy to lose tracking target.In this thesis,a segmented visual servo scheme is proposed.In the first stage,eye-to-hand system is used to measure the world coordinates of the center of gravity of the object,and then the reasonable initial pose of the manipulator is calculated,so that the object is in the field of view of the camera at the end of the manipulator at the beginning.In the second stage,the manipulator is first moved to the initial pose determined in the previous step,then eye in hand system starts to work,extracts the pixel coordinates of several feature points on the surface of the object,controls the movement of the manipulator according to the trajectory of the feature points designed in the image space,makes the current feature points close to the desired feature point coordinates,and finally makes the end actuator reach the ideal working pose.In order to achieve the above scheme,the camera and the six axis manipulator are modeled and analyzed respectively,and then the image processing algorithm of feature point extraction is introduced.On this basis,taking the positioning of small cube wood block by mechanical arm as an example,the overall workflow and relevant details of the system are described.Finally,the servo control experiment is designed to verify the effectiveness of the method in this thesis.In the first phase of the experiment,the average detection time is about 0.4s and the average detection error is less than 3mm.This shows that the speed and accuracy of the detection algorithm meet the basic requirements.In the second stage of image-based servo control simulation experiment,the algorithm converges fast,the motion of each axis of the manipulator is relatively stable,and the steady-state error is small,which shows that the system has good dynamic performance and control accuracy,and there is no singularity of Jacobian matrix,and the tracking process does not lose the target,which proves the feasibility of this scheme.