Systematic Error Correction in Robotic-OCT Inspection of Hard-to-Reach Industrial Parts

Hard-to-reach industrial parts have complex geometries, such as internal holes and deep cavities, posing challenges for accurate 3-D measurement. Robotic-arm-assisted optical coherence tomography (robotic-OCT) is a promising method for its ability to access hard-to-reach spaces with a compact probe, high resolution, and flexibility. However, probe misalignment and nonuniform rotation in robotic-OCT systems introduce systematic errors and sacrifice measurement accuracy. In this article, we report a technique for achieving precise self-alignment between the OCT probe and workpiece by utilizing the data feedback mechanism inherent in OCT scanning. We also propose a nonuniform rotational distortion correction method to rectify distortion errors. Experimental results showed that the proposed method improved point cloud completeness to 98.7%, surpassing uncorrected (77.6% ) and manual correction (96.8% ), and reduced errors by 84.1% compared to uncorrected and 13.0% compared to manual correction, while greatly enhancing efficiency over manual correction. The robotic-OCT system, with reduced errors, achieved a measurement accuracy better than 4.5 $\boldsymbol {\mu }$m for internal holes with diameters from 5 to 100 mm.