Self-Sensing Calibration Method for Robotic Assembly Error in Ultrasonic Conformal Inspection

To address the assembly error between the robotic manipulator and the test object in ultrasonic conformal inspection—where the test object is grasped by the robot (TOGR)—this paper proposes a closed-loop iterative calibration method based on ultrasonic self-sensing. The core of this method lies in utilizing endogenous echo signals from ultrasonic testing as self-sensing data and reconstructing the point cloud of the clamped object by fusing these signals with the synchronously acquired pose data of the manipulator. The system integrates trajectory planning, ultrasonic self-sensing data acquisition, and a point cloud reconstruction algorithm, enabling two primary functions: forward free-form surface trajectory planning and inverse point cloud reconstruction. The calibration matrix relative to the theoretical model is computed using the Iterative Closest Point (ICP) algorithm, thereby establishing a closed-loop iterative calibration process. Calibration and inspection experiments conducted on an aero-engine blade containing prefabricated defects demonstrate that, after five iterations of self-sensing calibration, the root-mean-square error (RMSE) between the reconstructed point cloud and the theoretical model is reduced to 0.57 mm, and the maximum deviation in ultrasonic time-of-flight is reduced to 0.57 µs. The quality of the C-scan images is significantly improved, enabling clear identification of flat-bottomed hole defects with diameters ranging from 0.12 mm to 0.25 mm inside the blade. This effectively enhances both the conformity of the scanning trajectory and the accuracy of ultrasonic follow-up inspection for the blade in the clamped state.