A New Machine Vision Method for Target Detection and Localization of Malleable Iron Pipes: An Experimental Case

Abstract

Malleable iron pipes are widely used in construction, manufacturing, aerospace, and many other fields. Cast malleable iron pipes need to be treated flat to meet the needs of different shapes and sizes. This process is usually completed manually, which is low efficiency and is subject to potential safety risks. To solve this problem, a machine vision method is proposed to detect and localize malleable iron pipes. Point cloud images of malleable iron pipes are obtained by the Random Sample Consensus (RANSAC) algorithm, and precise matching is completed by the Iterative Closest Point (ICP) algorithm to obtain more accurate positions, so as to realize robot grasping. The grasping experiments of malleable iron pipes with the same and different specifications were carried out using a specially designed experimental platform. The results show that malleable iron pipes can be identified effectively and that the corresponding grasping success rate is more than 85 %. The target detection and localization method can obtain the three-dimensional (3D) position of malleable iron pipes to improve grasping efficiency, which provided a certain theoretical basis and guiding significance to improve production efficiency in practice.