| With the development of the automobile industry, automobile has become the necessary traffic tool in people's everyday work and life. It brings human not only convenience but also traffic accidents and environment pollution at the same time.The automobile performance's problems are caused by many reasons. Automobile is a complex mechanical-electronic system. As the mileage increasing, some technical status will change, such as lower dynamic performance, more fuel consumption and worse security,which have great influences to economy and efficiency of automobile. Moreover, it will be a threat to safety. So the former automobile performances should be kept by all means. To check the automobile performances regularly is one of the means. But if the test equipment is imprecise, the exact situations of automobile can not be obtained. It is a pity that the problems exist widely in our country, which brings more dangerous factors and hidden troubles. The test equipment's test data are taced to the national standards regularly to make sure the right status of the test equipments are kept so that we can trust the test results. So it makes great sense to calibrate the automobile test equipments regularly according to calibrating standards.Wheel alignment failure is the commonest one of troubles. It brings on tyres abrasion, front-wheels sway and sideslip, which influence controlling stability and driving safety of the automobile. Wheel alignment detection is the most important test item of automobile safety tests.Wheel alignment parameter detection system is appropriative equipment that inspects vehicle's four-wheel positioning parameters. The instrument detects positioning parameters including wheel toe-in, camber, caster, SAI, thrust angle, steering toe-out and wheelbase difference based on the establishment of the wheel geometric model. Requirements for the whole vehicle equipment in GB18565-2001 "The general performance requirements and test methods of the working vehicle" (National Standards) are: vehicle and equipment should be complete, good, effective, with well-fastening connecting parts. In JJF1154-2006 national standard "Four-wheel positioning calibration standards", toe-in angle's measurement range is -3°-3°with±3′accuracy, camber angle's range is -10°-10°with ±5′accuracy. Caster's measuring range is -15°-15°with an accuracy of±10'. Transport industry-standard "Four -wheel aligner" JT/T505-2004 also regulates that SAI angle's measuring range is -20°-20°, testing accuracy is±6' within 0°-18°range, the rest of the scope of accuracy is±10'. It can be seen that the test items of wheel alignment parameters have been paid more attention with higher measurement accuracy demands.Based on the above background, the study of automotive detection systems of binocular stereo vision-based wheel alignment parameters is proposed to achieve fast and accurate measurement of parameters. Vision-based three-dimensional detection system is designed and developed aiming at the detecting vehicle wheel alignment parameters characteristics in this thesis. The automotive measurement of vehicle wheel alignment is put out using binocular stereo vision system. Based on the three-dimensional vision theory such as target feature extraction and image matching, camera calibration and vision-based technology, the kinematics model of the wheel in the detection process and the mathematical model of wheel positioning data are set up and studied by experimental method. In order to verify the accuracy of the system, the existing four-wheel aligner calibration apparatus'testing functions are analyzed. The existing installation problems at SAI and caster in testing are pointed out, too. Based on the movement relationship between wheels and toe-in, a four-wheel aligner calibration apparatus is developed with the function that is able to test all wheel alignment parameters such as wheel toe-in, camber, caster, SAI and so on with a five-point cross structure, which traces all the detection system testing parameters to national measurement standards.Main research contents of this article are:1.The computer vision theory-based wheel alignment parameters testing methods of detection system is studied. Firstly, measurement goal of the system including toe-in, camber, caster and SAI and their effect on vehicle working performance are analyzed. Wheel alignment parameters'requirements and detection system's performance target are explained. Secondly, the basic principles of computer binocular vision testing are expounded, and computer vision-based automotive wheel alignment measurement system is designed rationally considering the vision inspection system testing aim. System's optical structure sizes'impact on the measurement accuracy and the structural parameters applicable for measurement design are investigated.2.Image feature extraction and matching methods research. Key technologies of wheel alignment detection system including Gamma correction, the target feature point extraction, stereo matching are mainly studied. Firstly, Gamma correction function is set up to improve the image quality based on the basic principles adopting the nonlinear function superposition method. Secondly, the traditional corner point extraction algorithm is analyzed. The characteristic corner points are picked up based on Harris operator that suppresses the maximum value to extract a certain number of the scope of the best corner points, using target feature points'symmetry features and Guassian weighted symmetry operator. Finally three-dimensional visual matching technology is studied and feature points matching are achieved using epipolar and the order consistency of feature points constraint. The results show that the matching technology can achieve higher matching rate which meets the requirements for the experimental system.3.The vision inspection system calibration of the wheel alignment parameters. First of all, the camera imaging model is presented. The world coordinate system, camera coordinate system, the definition of the image coordinate system and the transforming relationship between them are described. Thus definitions of camera's internal and external parameters are given. Then, camera calibration methods based on image distortion correction are adopted after analyzing typical camera calibration algorithm in detail. Distortion parameters are solved using LM (Levenberg-Marquardt) that is non-linear iterative optimization methods, and various specific methods realizing algorithm are described, too. Finally the calibration algorithm for the binocular stereo vision is experimentally studied. Experiments show that the calibration process used is simple and accurate. The camera's inside and outside parameters of detection system and sensor structural parameters calibration can be achieved accurately.4.Research on the wheel alignment parameters measurement model. First of all, the measurement model of an existing typical four-wheel-positioning system is introduced and its kingpin angle measurement error is deduced. Then, the detection system instructs three-dimensional reconstruction of target feature points through the binocular stereo vision measurement model and realizes the reconstruction test of the target feature points with different angles. Finally, wheel kinematics model of the detection system is derived. The target translation and rotation matrix are got using SVD method with high anti-noise. Wheel toe-in, camber, caster, SAI and wheel steering angle's measurement model is built by vehicle kinematic model and all auto's detection of positioning parameters were studied.5.The study of the wheel position parameters detection system. Firstly, SAI and caster's theoretical test error is derived based on analyzing the existing four-wheel aligner calibration apparatus'test function. Then, the principle proposed that a calibration device should be followed is kingpin angle tested is generated independently in vertical and horizontal plane. Three four-wheel aligner calibration apparatus of five-axis cross at the point that is able to test all front-wheel positioning parameters including wheel toe-in, camber, caster and SAI are studied according to the movement relations among the testing wheels, targets and kingpin. The device can simulate geometry relations of vehicle wheel positioning system as while as represent its real detection process. The device conducts comprehensive inspections. Once again the test device's accuracy is studied. The accuracy of the test comes down to angle's geometric relationship test and the installation's geometric relationship. The device's calibration program is proposed aiming at high accuracy requirements. The test parameters and national standards are consistent. The multi-faceted prism and autocollimator are used to calibrate the angle. Calibration rod and level instrument device calibrate the installation position. The four-wheel aligner calibration apparatus'testing results'accuracy is ensured by tracing through the above-mentioned program. Finally, the calibration system carries out calibration experiments, whose results show that the accuracy of detection system meets the standard's requirements.Main innovation points in the research:1.The wheel alignment parameters'testing methods are proposed based on binocular stereo vision theory. In the research on extracting the characteristic points of image, arithmetic methods of corner points'extraction are put out combining Harris corner points and weighted Guassian symmetry operator. It realizes more precise characteristic points'extraction of the target image.2.Camera calibration methods are researched in detail. Camera calibration arithmetic methods based on image distortion correction using linear perspective transformation invariance are put forward. The arithmetic methods are made use of by binocular stereo camera's real-time calibration, which has got satisfying calibration precision.3.For the first time the principle that the tested kingpin angle is generated independently in vertical and horizontal plane is proposed, which the calibration device of the wheel alignment parameters'detection system should follow. 3 kinds of four-wheel aligner calibration apparatus that are able to simulate geometry relations of vehicle wheel positioning system as while as represent its real detection process. The device conducts comprehensive inspections. The consistency precept of the test parameters and national standards is advanced, which realizes comprehensive and accurate calibration of the wheel position parameters'detection system such as wheel toe-in, camber, caster and SAI.Thesis'research and development work is an actual application of combining vision detection technology and practice. Some research fruit is achieved in some areas such as research realization technology, arithmetic, precision guarantee, improving the level of stereoscopic vision detection technology and so on. It has theoretical and practical application value for improving the level of stereoscopic vision detection technology. Meanwhile, vehicle inspection technology is promoted and testing process automotion is realized in automobile field by the introduction of testing the wheel position detection parameters using the binocular stereo vision.
|
PDF Full Text Request