[en] Based on visual measurement techniques, the real-time robotic welding tracking system achieves real-time adjustment for robotic welding according to the position and shape changes of a workpiece. In system design, the sensor design technique is so important that its performance directly affects the precision and stability of the tracking system. Through initiative visual measurement technology, a camera unit for real-time sampling is built with multiple-strip structured light and a high-performance CMOS image sensor including 1.3 million pixels; to realize real-time data process and transmission, an image process unit is built with FPGA and DSP. Experiments show that the precision of this sensor reaches 0.3mm, and band rate comes up to 10Mbps, which effectively improves robot welding quality.With the development of advanced manufacturing technology, it becomes an inexorable trend to realize the automatic, flexible and intelligent welding product manufacture. With the advantage of interchangeability and reliability, robotic welding can boost productivity, improve work condition, stabilize and guarantee weld quality, and realize welding automation of the short run products [1]. At present, robotic welding has already become the application trend of automatic welding technology. Traditional welding robots are play-back ones, which cannot adapt environment and weld distortion. Especially in the more and more extensive arc-welding course, the deficiency and limitation of play-back welding technology becomes more prominent because of changeable welding condition. It becomes one of the key technology influencing the development of modern robotic welding technology to eliminate or decrease uncertain influence on quality of welding such as changing welding condition etc [2]. Based on visual measuring principle, this text adopts active visual measuring technology, cooperated with high-speed image process and transmission technology to structure a tracking sensor, to realize real-time measurement of the space location and posture information of the work piece and reliable and accurate tracking of the welding seam