Research On Pipeline Welding Evaluation And Tracking System Based On Passive Binocular Vision

The pipelines welding accounts for a considerable proportion of work in laying operations.Therefore,it is important to improve welding automation and reduce costs.The traditional welding tracking uses methods with high cost and low adaptability.This paper is based on the passive optical camera composed of binocular vision to identify and track the weld seam.The main work is as follows:(1)Binocular vision obtains the depth information of the weld seam through the principle of parallax which has high accuracy and good adaptability.This paper firstly deduces the parallax principle of binocular camera to obtain depth,analyzes the characteristics of the weld video,and uses it to design the recognition algorithm to identify each part of the weld,which can identify the upper and lower bevel edges,the melt pool,and the wire during the welding process,and track them.(2)Next,a basic mobile platform robot was designed that could carry the camera in motion and was carried out for the camera.(3)The software to control robot was written,and the software also carried the weld identification algorithm to achieve integrated operation of identification and control.(4)A test environment was eventually built to simulate pipeline welds,and the robot platform simulated the tasks of pre-weld evaluation and weld-time tracking in the test environment.Analysis of the final test results showed that the designed weld seam recognition algorithm was applied well,as well as the mobile robot platform and software could also meet the task requirements normally,which is of reference significance for the subsequent robot design required for actual production.