6D Pose Estimation of Stator Windings from CT-Measurements

In the manufacturing of electric motors, the quality of wire windings is a critical factor determining motor performance and reliability. Defective windings typically result in motor rejection during production, necessitating effective quality control methods. Non-destructive evaluation via computed tomography (CT) offers a promising approach for assessing internal components, although data analysis presents significant challenges. In this study, we introduce a novel method to extract the complete 6D pose of stator windings from CT measurements. The method applies a topology-preserving thinning algorithm to reduce segmented wires to their medial axis, from which positional and curvature data are derived. Local twist about the wire’s axis is quantified by analysing image moments in orthogonal cross-sections along the winding. Validation with synthetic CT data and physical motor replicas demonstrates high precision in estimating wire positions and rotations, including intricate bending and twisting behaviours. Future work will focus on optimizing the method to enhance its robustness against the noise inherent in real-world CT scans.