Abstract: The pressurized delivery pipeline, as a main component of the launch vehicle system, serves the function of stably conveying propellant from the tanks to the engine, directly influencing the flight reliability of the launch vehicle. To address issues such as the complexity of remote assembly of intricate pipelines and difficulties in repeated measurements, a method utilizing a binocular tracker combined with visual targets is proposed to achieve high-precision pose replication by robots in pipeline assembly. Experiments were conducted using industrial robots and a binocular vision tracking system with multiple repeated trials. The process initially involved coordinate system construction and alignment mapping, followed by the application of a vision-based multi-robot closed-loop control algorithm for real-time position correction. Finally, the effectiveness of the method was verified through on-site robot pose replication experiments. The results indicate that the maximum error during dynamic tracking was 0.01 mm, with an average error of only 0.007 mm, demonstrating good real-time performance and positioning accuracy. This research method holds practical application value in the field of complex pipeline assembly.