[3B4] Fiducial marker-based augmented reality positioning system for non-destructive testing
Y Wang¹, T Meng¹, W Yin¹, J Skelton², J Watson² and A Peyton¹
¹University of Manchester, UK
²Jacobs, UK
Augmented reality (AR) technologies have gained interest for a variety of applications, particularly in entertainment, education, security and medicine. Two particular features of AR technology are the ability to track positions in the real-world scene and display ancillary information in the user’s field of view. Both these features are of potential interest in non-destructive testing (NDT) applications, particularly when hand-held probes are used. Electromagnetic or ultrasonic inspections are typical examples, which collectively account for 60% of the NDT service market. Increasingly, these probes are being adapted for robotic tools (fixed or mobile) and scanner frames, underlining the growing need for position-indexed NDT data. Positional data is crucial to digitalisation, for referencing against prior inspections, computer-aided design (CAD) files and simulation software. Nevertheless, a significant fraction of NDT inspections can only be implemented by skilled operators with hand-held probes. In such cases, accurate positional information may be non-existent or rudimentary, ie tracked in 1D or 2D with an encoder wheel attached to the probe or an array of sensors providing data.
A fiducial marker-based AR positioning system for eddy current NDT is presented in this paper. The AprilTag marker and OpenCV library were used for positioning and AR implementation in the NDT application. The AprilTag detection and pose estimation were tested in various conditions; the results show that the marker’s pose estimated an error of the order of 1 mm in practical conditions when using readily available low-cost consumer hardware. Then, the markers were attached to a testing probe and the sample to indicate the positions and the AR technique in the eddy current testing (ECT) application was implemented. An eddy current scan of a stainless steel sheet was taken, where the sheet had a range of electric discharge-machined slots of varying depth. The ECT defect map was overlaid on the live captured image from a camera. By using the positioning system, the trace of the probe was shown virtually, which demonstrated the possibility of AR application for real-time and offline display in NDT.
A fiducial marker-based AR positioning system for eddy current NDT is presented in this paper. The AprilTag marker and OpenCV library were used for positioning and AR implementation in the NDT application. The AprilTag detection and pose estimation were tested in various conditions; the results show that the marker’s pose estimated an error of the order of 1 mm in practical conditions when using readily available low-cost consumer hardware. Then, the markers were attached to a testing probe and the sample to indicate the positions and the AR technique in the eddy current testing (ECT) application was implemented. An eddy current scan of a stainless steel sheet was taken, where the sheet had a range of electric discharge-machined slots of varying depth. The ECT defect map was overlaid on the live captured image from a camera. By using the positioning system, the trace of the probe was shown virtually, which demonstrated the possibility of AR application for real-time and offline display in NDT.
