In order to achieve automated weld seam inspection of nuclear power facility pipelines and reduce the safety risks of manual operations while improving work efficiency, a robot autonomous operation software system and related algorithms that integrates spraying, wiping, brushing, scrubbing, environmental perception, and scene modeling were proposed. The working environment and functional requirements of the work robot were analyzed, the robotic arm and environmental perception module was selected, a 3D modeling and parameter identification plan was developed, and a welding seam detection operation system was built. A scene modeling algorithm based on visual and depth information perceived by robots was proposed, which can output detection of the position of the working surface, 3D model, and the range of the working area. An algorithm for job trajectory planning based on 3D reconstruction model and job area range was proposed, and a simulation platform was built to verify the correctness of the motion trajectory. Experiments on pipeline positioning and 3D reconstruction were conducted, as well as trajectory planning. The results of the experiments proved that the proposed pipeline positioning method can achieve accurate point cloud data collection and high-precision pipeline parameter identification. The robotic arm can achieve high accuracy and good continuity of movement, meeting the requirements of pipeline weld detection operations.