Dual-station intelligent welding robot system based on CCD

Abstract

Welding is a manufacturing technique that joins metal or other thermoplastic materials in a heated, high-temperature, or high-pressure manner. Harmful dust and the light from welding can cause great harm to the human body. In addition, to improve welding quality and efficiency, intelligent welding has become an urgent need in the manufacturing industry. Herein, a dual-station intelligent welding strategy is designed based on an industrial charge-coupled device (CCD) visual detection system and welding control system. This solves the problem of expensive laser vision sensors and the poor detection effect of the visual system under the conditions of a short distance and strong light intensity. In the actual environment, there is the problem of radial distortion that affects the edge of the image, and the problem of the optical axis of the camera not being consistent with the installation plane. In this study, the camera coordinate system and hand–eye coordinate system are calibrated separately. The abovementioned problems are solved, and the images obtained by the CCD are acquired and processed in real-time. Image segmentation was performed using neighbourhood average filtering, iterative threshold segmentation, and local information extraction. The edge amplitude image was obtained by the Prewitt operator. Under the Hough transform method, the recognition of the weld seam and the extraction of weld feature points are realised. We designed an intelligent weld detection system that contains a friendly human–computer interface. Through numerous repeated experiments on circular and square workpieces, the control error of this intelligent welding system is within 0.2 mm, and the time of single seam feature extraction is 0.8 s.