Research On Vision Recognition Algorithm Of Wheel Pose And Position Parameters In Bench Testing

The pose and position parameters are one of the most basic group of parameterswhile studying the objects'dynamics characteristic. When tracing some movingbody's dynamics characteristic, such as pose, velocity, acceleration or overload, andanalyzing their errors, the pose and position parameters should be accuratelyrecognized in time.In bench testing, if the pose and position parameters of the moving wheel canbe found, we can trace out its sport orbit, and work out wheel alignment parameters.Therefore,when turning to steering system experiment, they can be used to analyzethe work character of the whole steering system. And when testing the suspensionsystem, such as testing of the suspension elastic kinematics character, testing andstudying of active or semi-active suspension character, the wheel pose and positionparameters can be regarded as the basic terms of the elastic kinematics modeling andthe dynamics modeling.However, to some mechanism in the space, if finding the one of the parts'poseand position parameters with the direct position analysis, the results received are notaccurate. This is because that the algorithms of the direct position analysis are notperfect at present. So it is still a difficult problem that how to find the pose andposition parameters of some mechanism in the space, and it leads to the difficulty ofcarrying out the closed-loop control.Based on above-mentioned reasons, the dissertation proposes a research subject:research on vision recognition algorithm of wheel pose and position parameters inbench testing. The algorithm not only applies to the steering testing and suspensiontesting, but also can be used for recognizing the pose and position parameters of themulti-freedom-degree machine in the robot field, such as the manipulator on theproduct line, the grab at the construction spot, and the platform with multi degrees offreedom, and the results will supply the base of the analysis and revise of themoving error and the realization of the closed-loop control. Therefore, the fruit ofthe dissertation can applies to the scientific research and the production practicesextensively, and possesses high utility value. From the scheme of the paper, the content can be divides as following parts: (1) At the first part, the paper explains the conception and the importance of thepose and the position parameters for the moving rigid body, then introduces thefunction of the recognition of the wheel pose and position parameters in the benchtesting, thus describes the meaning and the purpose in which the paper studies. Thenthe paper introduces the development and the current situation of the relevantdisciplines, which are the data close-range photogrammetry and the computer vision.And then introduces the level the similar subject reached. Finally, the paperdescribes the research contents. (2) Then, the paper confirms it as the camera model of the paper with thepinhole camera model with lens distortion. Though the model is more complicated,it is much closer to the course for real formation of image of the camera comparedwith the general model, and its precision is relatively high. And then each parameterof the camera model is found with the camera calibration testing. Meanwhile, eachsystem of coordinates is set up that the appearance recognition system in thelocation needs, and the conversion relationship is derived between the systems ofcoordinates. And the material, size, form and pattern of the artificial mark aredesigned. At last, using the point data cloud which is scan with the measuringapparatus of non-contact-type optics, and dealt with in the CATIA software, thethree-dimensional space coordinates of the artificial mark center are work out, whichis the character point. The above, the preparation that the algorithm realizes has beenfinished. (3) Various kinds of equipment packages of the recognition system hardwareare Chose, and be made use of with C ++ programming. Afer comparing variouskinds of methods, the image processing flow is conformed: creating the gray picturewith picking up the color channel t(t = R + G ? a ? B), sharpening the edge withSobel operator, and getting two-value picture with thresholding. The Least MedianSquare algorithm is chose to fit the edge of the artificial mark, and then the imagecoordinates of the character points are obtained. So the pose and position parametersof the moving wheel can be calculated through POSIT algorithm with the...