Research On The Technique Of Robot Autonomous Calibration Based On Vision Assistant Location

Robot with high absolute positioning accuracy is the premise when using off-line programming technique and completing high precision task.However,due to the influence of manufacturing and assembling,there is always error between the actual structure size of robot and its designed size,which decreases the absolute positioning accuracy.Therefore,calibrating the robot to achieve its actual structure size is of great significance.Currently many calibration methods are dependent on the imported measuring devices,which are always very expensive,and inconvenient to be applied at industrial production sites.Comparatively,the autonomous method based on planar constraints costs much less and is convenient to operate.Taking the industrial robot,Staubli TX60 L,as the research object,study on the technique of autonomous calibration was implemented in this paper.First of all,due to the problem that the current two popular D-H models,which were respectively proposed by Paul and Craig in sequence,were easy to be mixed.A research on the two methods was implemented and the essential differences between them were pointed out.The kinematic model of the industrial robot Staubli TX60 L was built according to the two methods respectively.Programs were written by using the Robotics Toolbox based on Matlab,and it showed that the kinematic model was built correctly.Due to the problem that the value of D-H parameters changes drastically while the axes of two joints are parallel,a parameter ? rotated around the Y-axis of coordinate system was introduced,which optimized the kinematic model.Secondly,the differential transformation of kinematics was studied.Based on the kinematic model built using Paul-method,the end-pose error model of robot was established,and was proved to be correct by simulating based on Matlab.Furthermore,the kinematic parameters error identification model was also simulated and proved to be reliable.Due to the situation that only the end-position is achieved,the end-position error model was built and was proved to be correct by simulating.In addtion,the technique of robot autonomous calibration based on planar constraints and contact-type probe was studied.Due to the drawback or somewhere needs to be improved in the planar-constraint error model based on the normal vector of plane as well as the equation of plane.A calibration method was proposed,which reduced the requirement for experiment and simplified the error model.Experiment showed that the proposed method enhanced the absolute positioning accuracy of robot to some degree.Then,in order to improve the automaticity of calibrating,a method based on vision assistant location for constraint-planes and automatically completing the measuring process was proposed.The stereo vision system was studied and established,and the camera imaging model,intrinsic parameter model,extrinsic parameter model,eye-in-hand parameter model,3D locating model were built in sequence.Using the Calibration Toolbox by Zhang Zhengyou based on Matlab,the intrinsic parameters of cameras were calibrated.The theory of Kronecker product was studied,and the eye-in-hand parameters of two cameras were successfully achieved.Due to the narrow public sight of stereo vision and in order to simplifying digital image processing,several target spots were established on constraintplanes,which transformed the location of planes into the location of target spots.The theory of digital image processing was studied,and programs were written to extract the pixel coordinates of target spots center.After substituting the pixel coordinates into the 3D locating model,the 3D coordinates of target spots in the base coordinate system were calculated,and the absolute positioning accuracy of stereo vision was evaluated by experiment.Finally,based on Visual Studio and adopting C# to write programs,the class library mainly used for relevant calculation of robot calibration was packaged,in which were some basic functions such as calculating the forward kinematics,transforming the rotating matrix into RPY angle,calibrating the tool parameters,identifying the kinematic parameters error,and simply dealing with some exceptions,which avoided the crashing or deadlock of software when being used.Essential parameters were opened on the software interface,so that the software can be used to solve the forward kinematics and calibrate the kinematic parameters error for any 6-DOF serial industrial robot with revolute joints.Taking the common 6-DOF serial industrial robot with revolute joints as the research object in this paper,the kinematic model,end-pose error model,end-position error model,and kinematics parameters error identification model were built in sequence.A calibration method based on planar constraints was proposed,which enhanced the absolute positioning accuracy of robot.An autonomous calibration method based on vision assistant location was proposed.Taking the eye-in-hand stereo vision as the research object,the camera imaging model,intrinsic parameter model,extrinsic parameter model,eye-in-hand parameter model,3D locating model were built in sequence,which laid the foundation for improving the automation of robot calibration.