A Machine Vision-Based Method for Pose Estimation of Pin Shafts in Oil Derrick Structures

As a core component in oil drilling operations, the derrick structure of a drilling rig traditionally relies on manual handling during assembly and disassembly, which is complex, inefficient, and poses significant safety risks. To enhance the level of automation, this paper proposes a machine vision-based method for pose recognition of pin shafts in derrick structures. A structured light camera is employed to capture on-site 3D point cloud data, which is then aligned with pre-defined workpiece templates for recognition. The registration process begins with a coarse alignment using the Sample Consensus Initial Alignment (SAC-IA) algorithm based on Fast Point Feature Histograms (FPFH), followed by fine pose estimation through the Iterative Closest Point (ICP) algorithm accelerated by a KD-tree structure. Experimental results demonstrate the effectiveness and practicality of the proposed system in identifying and locating typical components. This method offers reliable technical support for the automated assembly and disassembly of derrick structures and holds promising value for engineering applications.