In terms of safety and convenience, an Unmanned Aerial Vehicle (UAV) offers significant benefits when conducting remote NDT evaluations by mitigating hazards and inefficiencies associated with manned access. Traditionally, UAV remote inspections rely on high-resolution cameras, providing a visual overview of surface condition. This photogrammetric inspection, however, cannot distinguish minute discontinuities or deformations beneath a surface coating. Ultrasonic inspection is a Non-Destructive Testing (NDT) method conventionally used in corrosion mapping. Surface contacting ultrasonic transducers offer the potential for internal inspection of an industrial asset, providing enhanced structural integrity information. However, manually piloting a UAV with sufficient surface proximity to perform a detailed, contact-based examination requires a highly developed skillset and intense concentration. Limitations of payload mass and electronic interference also represent significant challenges to be overcome. Addressing such issues, this paper demonstrates the implementation of an autonomous UAV system with an integrated ultrasonic contact measurement payload. The prototype is autonomously guided and undertakes the contact thickness measurement process without manual intervention.

Remote photogrammetric inspection is a Non-Destructive Testing method used to quantify surface integrity and detect external discontinuities. The mobility and size of an unmanned aerial vehicle (UAV) offer the flexibility to quickly deploy remote photogrammetric inspections for large-scale assets. In this paper, the results of a photogrammetric inspection are presented as a 3D profile, reconstructed from UAV captured images. Experiments were conducted indoors using a wind turbine blade section obtained from a recently decommissioned asset. The naturally occurring surface features representative of environmental wear were augmented with a small number of artificial features to aid in the visualisation of inspection quality. An autonomous UAV system for photogrammetric inspections is demonstrated and the influence of image parameters such as environmental light levels, motion blur and focal blur quantified in terms of their impact on the inspection accuracy. Over the range of parameter values studied, the poorest scenario was observed to cause a degradation in reconstruction error by a factor of 13 versus the optimal. Reconstruction quality when employing a laser range scanner to maintain standoff distance relative to the object during flight was also investigated. In this schema, the controller automatically generated a real-time adaptive flight path to follow the outer profile of the wind turbine blade and, consequently, demonstrated improved image quality during close-range inspection of an object with complex geometry. Inspection accuracy was quantified using the error of the photogrammetric reconstruction as compared to a model acquired using independent metrology equipment. While utilising the laser-based adaptive path, error in the reconstructed geometry was reduced by a factor of 2.7 versus a precomputed circular path. In the best case, the mean deviation was below 0.25 mm. Instances of wind turbine blade damage such as edge crushing, surface imperfections, early stage leading edge erosion were clearly observed in the textured 3D reconstruction profiles, indicating the utility of the successful inspection process. The results of this paper evaluate the impact of optical environmental effects on photogrammetric inspection accuracy, offering practical insight towards mitigation of negative effects.

No abstract

The mobility of an Unmanned Aerial Vehicle (UAV) offers significant benefits when deploying remote Non-Destructive Testing (NDT) inspections of large-scale assets. Ultrasonic inspection is primarily a contact-based NDT method, that grants the opportunity to remotely monitor the structural health of an industrial asset with enhanced internal integrity information. Presented in this paper is an implementation of an autonomous UAV system, equipped with an ultrasonic thickness measurement payload. This system is designed to conduct ultrasonic inspections of non-magnetic facilities and industrial infrastructure where surface adhesion cannot be achieved magnetically. Operating within a laboratory environment, this system autonomously positioned the transducer on a vertically mounted, unpainted, aluminium sample and completed an ultrasonic thickness measurement without manual intervention. An onboard laser scanner provided instantaneous UAV alignment and standoff error measurements versus the sample's surface normal vector. While inspecting a region of the aluminium sample with 12.92 mm nominal thickness, the UAV system demonstrated a measurement error of 0.03 mm. During this process, the standard deviation of the craft's positional error was recorded to be below 63.26 mm, accompanied by an angular alignment error versus the surface normal vector of below 2.71°. The accuracy of the UAV deployed inspection, including thickness measurement accuracy and positional accuracy, depends on many factors. As such, transducer alignment constraints, electrical noise and UAV stability are investigated and discussed. Findings from this paper may be taken to inform future research regarding autonomous airborne ultrasonic inspection of constructed infrastructure and industrial facilities.

The creation of unwrapped stitched images of pipework internal surfaces is being increasingly used to augment routine visual inspection. A significant challenge to the creation of these stitched images is the need to estimate the pose and position of the camera for each frame, which is often alleviated through the use of a mechanical centralizer to ensure the camera is held in the center of the pipe. This article proposes a novel method for image centralization and pose estimation, which is particularly beneficial to circumstances where mechanical centralization is impractical. The approach involves post-inspection centralization of the captured video, by first estimating the probe's position relative to the pipe, using an integrated laser ring projector combined with the camera sensor, and then using this position to unwrap the image, so it produces an undistorted view of the pipe interior (equivalent to unwrapping a centralized view). These unwrapped images are then stacked to produce a stitched image of the pipe interior. In this paper pose estimation was successfully demonstrated to have a 90% confidence interval of ±0.5 mm and ±0.5° in translation and rotation changes. This pose estimation is then used to create stitched images for both a visual test card image mounted inside a pipe and an aluminum pipe sample with artificial defects, in both cases demonstrating near equivalent results to those obtained using traditional mechanical centralization.