Welding robots have been widely applied in welding industry to improve the automation level of welding production and the labor condition of workers because of precision, reliability and stability. Welding robots mainly work in teaching-and-playing mode nowadays, which lack adaptability for welding condition. In process of welding, the torch mounted on welding robots often deviate from the seams because of the error in positioning and the warp and distortion caused by thermal expansion, which deteriorates the quality of welding seams. Welding robots are expected more adaptability for the complexity of welding production to detect the seam position of work pieces and rectify the deviation to realize seam tracking. Research on automatic welding robots based on vision sensor has widely promising applications and significance. In this dissertation, the structured light image processing, features extraction and vision control of welding robot are studied. Firstly, robot vision sensor and vision measure system are designed based on the analysis of laser reflection on smooth surface and the imaging effect of the position and pose relationship of the camera, laser plane and work pieces. Secondly, structured light images processing and the extraction of laser stripes are researched in the case of disturbances of strong reflection. . Thirdly, a recognition method of joint type is proposed based on model matching. Hausdorff distance is adopted to measure matching degree between joint stripes and standard models. Fourthly, the shape feature is selected as one of image features. The welding joint types are divided into three typical categories: groove with slopes, butt joints, and lap joints; and the features extraction algorithms are developed respectively. Fifthly, structured light vision method is applied to aperture measurement and positioning. Sixthly, seam tracking vision control of pipe welding robot is studied. A vision control method of pipe welding robot is proposed based on analysis of vision control of welding robot with five freedom degree, and the seam tracking vision control based on images without calibration is realized. Finally, the obtained research results of dissertation are summarized and future work is addressed.
修改评论