Simultaneous pose control and defect imaging by integrating eddy current array sensor and robotic arm

Automatic inspection of metallic pipe surfaces is essential for both manufacturing quality control and in-service maintenance. However, real pipes often exhibit local deformations and high surface reflectivity, resulting in unknown surface profiles that hinder non-contact defect imaging. To address this challenge, this study integrates a UR5 robotic arm with a 16-coil eddy current (EC) array sensor to enable adaptive smoothing scanning and defect imaging of unknown pipe surfaces in a single pass. EC signals from three coils located over defect-free areas are used as feedback to control the sensor pose in real-time, while signals from the remaining coils are employed for defect imaging. An adaptive smoothing surface scanning strategy is introduced, which dynamically adjusts the scanning step size according to sensor pose errors to balance measurement accuracy and scanning efficiency. Experiments on an aluminum pipe with machined defects demonstrate a surface coverage rate of approximately 1 cm 2 / s , a reconstructed surface RMSE of 0.51 mm, and defect radius estimation deviations below 0.2 mm. These results indicate that the proposed method enables efficient, non-contact inspection of unmodeled and reflective pipe surfaces, providing a practical solution for automated industrial non-destructive evaluation.