Adapting robot paths for automated NDT of complex structures using ultrasonic alignment

Riise, Jonathan and Mineo, Carmelo and Pierce, S. Gareth and Nicholson, P. Ian and Cooper, Ian (2019) Adapting robot paths for automated NDT of complex structures using ultrasonic alignment. AIP Conference Proceedings, 2102 (1). 040006. ISSN 0094-243X

[img] Text (Riise-etal-AIP-CP-2019-Adapting-robot-paths-for-automated-NDT-of-complex-structures-using-ultrasonic-alignment)
Riise_etal_AIP_CP_2019_Adapting_robot_paths_for_automated_NDT_of_complex_structures_using_ultrasonic_alignment.pdf
Final Published Version
Restricted to Repository staff only until 8 May 2020.

Download (1MB) | Request a copy from the Strathclyde author

    Abstract

    Automated inspection systems using industrial robots have been available for several years. The IntACom robot inspection system was developed at TWI Wales and utilizes phased array ultrasonic probes to inspect complex geometries, in particular aerospace composite components. To increase inspection speed and accuracy, off-line path planning is employed to define a series of robotic movements following the surface of a component. To minimize influences of refraction at the component interface and effects of anisotropy, the ultrasonic probe must be kept perpendicular to the surface throughout the inspection. Deviations between the actual component and computer model used for path-planning result in suboptimal alignment and a subsequent reduction in the quality of the ultrasonic echo signal. In this work we demonstrate methods for using the ultrasonic echo signals to adapt a robotic path to achieve a minimal variation in the reflected surface echo. The component surface is imaged using phased array probes to calculate a sparse 3D point cloud with estimated normal directions. This is done through a preliminary alignment path covering approximately 25% of the total surface to minimize the impact on overall inspection time. The data is then compared to the expected geometry and deviations are minimized using least-squares optimization. Compared to manual alignment techniques, this method shows a reduction in surface amplitude variation of up to 32%, indicating that the robot is following the surface of the component more accurately.