Weld line recognition and path planning with spherical tank inspection robots

Abstract Periodic inspections are required for the safe operation of large pressure vessels such as spherical tanks. Inspection robots have been applied in large pressure vessels due to their low cost and high efficiency. This paper presents a robotic system for the inspection of spherical tanks, which can identify and track weld lines on the shortest running route. Two‐dimensional (2D) weld maps were prepared for robot path planning on the basis of the actual distribution of weld lines. In 2D weld maps, indispensable repetitive lines were added to form an Eulerian circuit that traversed all weld lines. In addition, an improved Fleury algorithm was proposed to solve Eulerian circuit and plan an optimal running route for robot inspection. To accurately identify weld lines, deep learning networks were constructed and trained with weld line data sets, which were captured by the camera mounted in the front of the robot. The laboratory experiments indicated that the inspection robot could identify weld lines within 0.2–0.25 s and track weld lines with a maximum offset of ±20 mm. The experiment results demonstrated that the robot could plan the shortest path to traverse all weld lines on the experimental platform. In the field tests, the virtual simulation of weld path planning on spherical tanks was explored in detail. The field tests of a spherical tank (3000 m 3 ) verified that the robotic system could improve the efficiency and stability of inspection operations and replace manual inspection with automated weld line recognition and weld path planning.