GB2605989A

GB2605989A - Device for simultaneous NDE measurement and localization for inspection scans of components

Info

Publication number: GB2605989A
Application number: GB2105651.0A
Authority: GB
Inventors: Garriga Casanovas Arnau; Khalili Pouyan
Original assignee: Sonobotics Ltd
Current assignee: Sonobotics Ltd
Priority date: 2021-04-20
Filing date: 2021-04-20
Publication date: 2022-10-26
Also published as: GB202105651D0

Abstract

Apparatus 2 to localise the position and orientation of a non-destructive evaluation (NDE) probe 4 in order to perform scans of components 6 and encode the relative probe pose, comprises; an ultrasonic transducer having a transducer housing 4; one or more cameras 10 attached to the transducer housing, orientated relative to an active face of the transducer; a computation unit adapted to receive transducer ultrasonic signals and camera signals; wherein the ultrasonic and camera signals are generated by the transducer and cameras in a time synchronised manner. The images from the cameras are processed by a localisation algorithm, which may include simultaneous localisation and mapping (SLAM) to determine their position and orientation with respect to the environment.

Description

DEVICE FOR SIMULTANEOUS NDE MEASUREMENT AND LOCALIZATION FOR

INSPECTION SCANS OF COMPONENTS

INTRODUCTION

This invention relates to a device and method to localise the position and orientation of a non-destructive evaluation (NDE) probe in order to perform scans of components and encode the relative probe pose. The invention relates particularly to NDE probes using ultrasound for inspection, and specifically includes but is not limited to probes known as electromagnetic acoustic transducers (EMATs).

BACKGROUND

Non-destructive tests, which are also referred to as non-destructive evaluation (NDE), are performed regularly in industry to inspect components that include but are not limited to pipes, plates, storage tanks, and structural elements such as beams. A common inspection technique involves the use of elastic waves, such as ultrasound waves, to obtain inspection measurements, which include but are not limited to the component thickness, and the presence of defects which include but are not limited to cracks or pits. A particular type of probe that excites and receives elastic waves is an EMAT probe. Other types of probe that excite and receive ultrasound waves are probes with piezoelectric elements.

Encoding the pose of these inspection probes, including EMATs and all other type of ultrasound probes, is a common challenge when performing the inspection. This is particularly difficult when performing in situ inspections, since the inspected components may not have clear grids or references on them to scan them following a specific pattern.

In addition, the inspected components can have curved and complex geometries, which further complicates any encoding.

Encoding the pose of the probe while inspecting is important in order to determine the coverage of the inspection performed, and to generate maps of the scanned region.

Encoding is especially important when performing line scans (B-scans) or surface scans (C-scans) of components. The resulting inspection maps obtained when encoding the probe pose contain the inspection measurements, which can include thickness or presence of defects, combined with the probe pose. These inspection maps provide a way to determine the inspection coverage of the component of interest, and synthesize the display of the inspection data. In addition, they provide information for inspection staff on the location of any potential defect or region of interest in case this needs to be reinforced or repaired.

In addition, encoding the full pose of the probe is also of specific importance when performing inspections using EMAT probes. This is due to the fact that EMAT probes generate polarized shear waves, and the orientation of this polarization influences the interaction with potential defects in the component such as cracks. As such, the knowledge of the full EMAT pose from the encoding provides information regarding the type of defect in the component, and can also be used to inform future inspections of the same region by recommending a specific probe polarization.

The challenge of encoding the probe pose during inspection applies to both manual, semi-automated, and automated or robotic inspections. In all cases, a robust device and method to encode the probe pose while scanning is necessary. Proposed solutions exist, but they present limitations that make them unsuitable in a variety of scenarios.

The majority of existing solutions are based on rotary encoders, which are arranged or mounted in different manners to achieve encoding of the probe position.

One common existing solution consists of a set of two cables attached to the probe on one end and anchored at two different points on the component in their other ends. Rotary encoders then are placed on these cables to measure their extension. The combination of the two encoded cable length can be used to encode the probe position. This solution is effective when scanning planar components such as plates and components with simple geometry, but presents issues when scanning complex three dimensional geometry components. In addition, it does not provide information regarding the probe orientation.

Another solution involves the use of wheels with encoders mounted on a probe, the wheels are arranged perpendicularly, and thus provide 2-dimensional encoding of the probe position. The solution has the limitation that it requires complex manipulation to ensure constant contact between the wheels and the component. In addition, it has the issue that the wheels can slip. Moreover, this solution does not provide information regarding the probe orientation.

Another existing solution is to use a mechanical arm. It consists of a set of rigid links and articulated joints holding the probe at one end, and anchored at its other end. The arm joints contain encoders, and therefore can be used to determine the probe pose at all times by relying on the direct kinematics of the arm. This solution presents the drawback that it is complex, costly, and relatively large and heavy. In addition, it needs to be anchored, which complicates its deployment.

SUMMARY OF THE INVENTION

The invention described in this document is a device that consists of one or more cameras mounted on an NDE probe (which can be an EMAT probe), and uses the visual images from the camera as the method to localise the probe. The cameras record images from their environment and use a localisation algorithm, which can include but is not limited to simultaneous localisation and mapping (SLAM), to determine their position and orientation with respect to the environment. SLAM refers to an existing set of algorithms that are used to determine the position and orientation of one or more cameras, which are commonly referred to as monocular SLAM or stereo camera SLAM, from visual information from the cameras, and simultaneously generate a map of the environment [1], [2], [3], [4]. The simultaneity is due to the fact that the problems of localisation and mapping are interconnected, and solving one requires simultaneously solving the other, and vice versa. In the solution described in this document, SLAM is one of the algorithms that may be used to localise the position and orientation of the cameras that are part of the device described. The aim of using SLAM in the solution described in this document is primarily to localise the camera and by extension the device described in this document, and the mapping part of the algorithm is secondary.

Then the pose of the cameras is combined with the knowledge of the geometry of the attachment between the cameras and the probe to determine the pose of the NDE probe at its base, which is where it scans the component. This localisation method is used to encode the pose of the probe as it is moved over a component to scan it. This allows one to generate a line or a map of the scanned component showing features that include but are not limited to the component thickness, and the presence of defects which include but are not limited to cracks or pits.

The invention described in this document is robust and capable of encoding the full pose of the probe when scanning it over a component of any geometry. It has the advantage that it does not involve moving parts, and its setup is simple. In addition, it is relatively lightweight (it can be less than 1 Kg), it is more portable than existing solutions such as the robot arm, and it does not require mechanical anchoring.

In the case of using EMAT probes as the NDE probe in the device, the magnets in the EMAT probes can generate a significant attractive force towards any ferromagnetic components being inspected. As a result, it can be difficult to drag the inspection device described in this document over the component. The use of wheels or rollers can be desirable. In particular, in some cases the device described in this document may be augmented with wheels or rollers to create a cart that can be easily scanned over a component. This cart may include motors as actuators to move it over the surface of the component, thereby creating an inspection robot.

BRIEF DESCRIPTION OF THE DRAWINGS

Embodiments of the invention will now be described, by way of example only, with reference to the accompanying drawings in which: Fig1. Schematically illustrates a system for encoding pose of an EMAT probe when used to generate/receive ultrasonic vibrations.

Fig2. Schematically illustrates the components of a system for encoding/localizing an EMAT probe in 3-dimensional space.

Fig3. Schematically illustrates the components of a wheeled scanner system for encoding/localizing an EMAT probe in 3-dimensional space.

Fig4. Schematically illustrates the components of an omnidirectional scanner system for encoding/localizing an EMAT probe in 3-dimensional space.

Fig5. Schematically illustrates the EMAT probe and stereo camera assembly when used to measure the remnant wall thickness of a conductive/magnetic material.

Fig6. Schematically illustrates the reflection of ultrasonic vibrations from smooth surface.

Fig7. Schematically illustrates the reflection of ultrasonic vibrations from a rough surface.

Fig8. Schematically illustrates a system for encoding pose of an EMAT probe when used to screen a conductive/magnetic structure for defects using ultrasonic vibrations.

Fig9. Schematically illustrates a system for encoding pose of an EMAT probe when used to screen a defective region of a conductive/magnetic structure using ultrasonic vibrations.

Fig10. Shows a flow diagram schematically illustrating the use of an EMAT probe and a stereo camera to determine the pose of system at the point when the ultrasonic vibrations are obtained.

Fig11. Shows an example in which the system for encoding/localizing an EMAT probe is used to obtain a 3-dimensional thickness map of a flat component with varying thickness.

Fig 12. Shows an example in which the system for encoding/localizing an EMAT probe is used to obtain a 3-dimensional thickness map of a curved component with a constant thickness.

DETAILED DESCRIPTION OF DRAWINGS

Fig1. Schematically illustrates a system 2 for simultaneous NDE measurement and localization for obtaining inspection scans of components. NDE measurements can be obtained with various types of probes including but not limited to EMATs to perform inspections of a component. Depending on the nature of the inspection, the NDE probe 4 can be reconfigured to obtain the desired measurement such as, but not limited to, measuring the remnant thickness of a component using ultrasonic vibrations 6 or screening a component for defects using ultrasonic guided waves. A data acquisition unit 8 is used to obtain the NDE measurements. The same or separate data acquisition unit 8 is used to obtain the pose data produced from the stereo camera 10. The NDE data and the pose data are then merged based on the corresponding time of acquisition to obtain the position of the NDE probe in 3D space at the point in time when the measurement is obtained. The merged data is then transmitted, via a wired connection or a wireless connection, to a receiving machine 12 such as, but not limited to, a computer, mobile device, tablet, or a robot; this data may then be displayed using a graphic interface device 14 such as, but not limited to, a monitor or tablet screen.

Fig2. Illustrates the components of a scanner device, excluding the acquisition system, used for simultaneous NDE measurement and localization for obtaining inspection scans of components. The device consists of an NDE probe 16 such as, but not limited to, an EMAT sensor which can generate ultrasonic vibrations 18 in a conductive material 20 which are then used to obtain measurements such as, but not limited to, the remnant thickness of the conductive material 20. A stereo camera 22 is used to obtain the pose of the device using SLAM by analyzing the images obtained from the incorporated cameras; the stereo camera and the NDE probe are physically connected and assembled using a rigidly attachment 24.

Fig3. Illustrates the components of a wheeled scanner device, excluding the acquisition system, used for simultaneous NDE measurement and localization for obtaining inspection scans of components. The device consists of an NDE probe 16 such as, but not limited to, an EMAT sensor which can generate ultrasonic vibrations 18 in a conductive material 20 which are then used to obtain measurements such as, but not limited to, the remnant thickness of the conductive material 20. A stereo camera 22 is used to obtain the pose of the device using SLAM by analyzing the images obtained from the incorporated cameras; the stereo camera and the NDE probe are physically connected and assembled using a rigidly attachment 24. The assembly is then mounted on a chassis 26 which consists of at least 2 wheels 28 in order to minimize the friction between the device and the conductive component 20 which could help with the manipulation of the device over surface of the component 20 with an arbitrary geometry.

Fig4. Illustrates the components of an omnidirectional scanner device, excluding the acquisition system, used for simultaneous NDE measurement and localization for obtaining inspection scans of components. The device consists of an NDE probe 16 such as, but not limited to, an EMAT sensor which can generate ultrasonic vibrations 18 in a conductive component 20 which are then used to obtain measurements such as, but not limited to, the remnant thickness of the conductive material 20. A stereo camera 22 is used to obtain the pose of the device using SLAM by analyzing the images obtained from the incorporated cameras; the stereo camera and the NDE probe are physically connected and assembled using a rigidly attachment 24. The assembly is then mounted on a chassis 26 which consists of at least 2 omnidirectional wheels 30 such as, but not limited to, ball transfer units, in order to minimize the friction between the device and the conductive component 20 which could help with the manipulation of the device over surface of the component 20 with an arbitrary geometry.

Fig5. Schematically illustrates the NDE probe 16 such as, but not limited to, an EMAT probe and stereo camera 22 assembly when used to measure the remnant wall thickness of a conductive/magnetic material 20. In order to obtain the remnant wall thickness of a component, a pulse of input 32 is transmitted to the NDE probe 16 which results in the generation of ultrasonic vibrations 18 on the proximal surface 34 of the component, the ultrasonic vibrations then travel through the wall of the component 20 and reflect from the distal surface 36 of the component 20 following the reflecting path 38 back to the proximal surface 34 of the component 20; the reflected ultrasonic vibrations 18 then form the current pulse of output 40 for which the arrival time is detected using various signal processing techniques such as, but not limited to, time-of-flight calculations. Simultaneously, the pose of the NDE probe is obtained with the attached stereo camera 22 using the images acquired from its cameras along with SLAM algorithms running directly on the processor onboard the stereo camera 22 or on a processor located on a separate acquisition system 8.

Fig6. Schematically illustrates the reflection of a pulse of ultrasonic vibrations 18 from a smooth distal surface 36. In this case, as the distal surface 36 is smooth (at least at a scale relative to the wavelength of the ultrasonic vibrations), then the reflection will be uniform, and the reflected ultrasonic waves will suffer no unpredictable changes in phase, undesirable dispersion, or other changes in their waveform.

Fig7. Schematically illustrates the reflection of a pulse of ultrasonic vibrations 18 from a rough distal surface 36. The rough distal surface 36 gives rise to rough surface scattering in which reflections from the peaks and troughs in the roughened surface interfere with each other and produce a reflected pulse of ultrasonic vibrations which are subject to changes in phase, dispersion and other changes in its wave form that vary considerably with the backwall shape. These changes in the waveform of the reflected pulse of ultrasonic vibrations that result from unpredictable changes in back wall shape make it difficult to accurately identify a time of arrival of that reflected pulse of ultrasonic vibrations.

Fig8. Schematically illustrates a system for encoding pose of an NDE probe 16 such as, but not limited to, an EMAT sensor when used to screen a conductive/magnetic component 20 for defects using ultrasonic guided waves 42. In contrast to the ultrasonic vibrations 18 which travel perpendicular to the surface of the component 20 and are used to determine the remnant thickness of a conductive component 20, ultrasonic guided waves 42 are ultrasonic vibrations that travel parallel to the surface of the component 20 which allow for the fast screening of a large area of the component 20 at once from a single NDE probe 16 location. In some but not all cases, the ultrasonic guided waves 42 are generated symmetrically on both sides of the NDE probe 16, the waves travel parallel together but in opposite directions; this allows for the areas of the component 20 located on either side of the NDE probe 16 to be inspected simultaneously. A stereo camera 22 is used to obtain the pose of the NDE probe 16 using SLAM by analyzing the images obtained from the incorporated cameras; the stereo camera and the NDE probe are physically connected and assembled using a rigidly attachment. Similar to the procedures explained above, the ultrasonic data from the NDE probe 16 can then be merged with the pose data from the stereo camera 22 in order to generate a 3-dimensional scan of the component 20.

Fig9. Schematically illustrates a system for encoding pose of an NDE probe 16 such as, but not limited to, an EMAT sensor when used to screen a defective region of a conductive/magnetic component 20 using ultrasonic guided waves 42. Similar to the explanation above, the ultrasonic guided waves 42 generated by the NDE probe 16 travel parallel to the surface of the component 20; in a case where a defect such as, but not limited to, a crack is present, the ultrasonic guided waves 42 reflect from the defect and travels back to the NDE probe 16. This reflection appears as a pulse in the signal acquired from the NDE probe 16 which can then be used to identify the presence and location of the defect relative to the NDE probe 16. Similar to above, the ultrasonic data from the NDE probe 16 can then be merged with the pose data from the stereo camera 22 in order to generate a 3-dimensional scan of the component 20; here, the presence and absolute location of the defect is obtained and displayed by the scan.

Fig10. Shows a flow diagram schematically illustrating the use of an NDE probe 16 and a stereo camera 22 to determine the pose of system at the point when the ultrasonic vibrations 18 are obtained with one example embodiment of the present technique. At step 44 and 46, the processes for the ultrasound and pose acquisition are inactive until the user starts the data acquisition. At step 48 the desired input signal to the NDE probe 16 is generated and passed on to step 50 to be amplified to the desired level to be appropriate for the NDE probe 16. Ultrasonic vibrations 18 are created and then received by the NDE 16 and transmitted to step 52 to be amplified to the desired level. The amplified signal is then passed on to step 54 to be sampled and recorded. The recorded signal is then passed to step 56 to be analyzed using various signal processing techniques such as, but not limited to, signal filtering, cross-correlation and time-of-flight calculations. Step 56 also determines whether a valid ultrasonic data is acquired. Simultaneous to the ultrasonic data acquisition steps 44, 48, 50, 52, 54 and 56, the pose data acquisition procedure starts from step 46. The command is then passed to the stereo camera 22. The data from the step 22 are then fed into step 58 to be analyzed. Then the pose data is calculated at step 60. In the scenario where it has been determined at step 62 that a valid ultrasonic data is obtained, this data is merged with the pose data at step 64 and the resulting output is generated at step 66.

Fig11. Shows an example in which the system 2 for encoding/localizing an NDE probe is used to obtain a 3-dimensional thickness map of a flat component 68 with varying thickness. In this specific example, a block 68 made out of conductive material with a nominal thickness of 30 mm is used; the block have two machined regions where the remnant thickness of the block is reduced down to 20 mm 70 and 25 mm 72. The NDE probe 16 such as, but not limited to, an EMAT probe and stereo camera 22 assembly is manipulated over the proximal surface 34 of the component 68 in order to generate a 3-D C-Scan of the block in real time as shown in the colour 3D plot 74.

Fig 12. Shows an example in which the system 2 for encoding/localizing an EMAT probe is used to obtain a 3-dimensional thickness map of a curved component 76 with a constant thickness. In this specific example, a curved component (section of a pipe) 76 made out of conducive material with a nominal and constant thickness of 6 mm is used. The NDE probe 16 such as, but not limited to, an EMAT probe and stereo camera 22 assembly is manipulated over the proximal surface 34 of the component 76 in order to generate a 3-D C-Scan of the block in real time as shown in the colour 3D plot 78.

REFERENCES

[1] H. Durrant-Whyte and T. Bailey, "Simultaneous localization and mapping: part I," in IEEE Robotics & Automation Magazine, vol. 13, no. 2, pp. 99-110, June 2006, doi: 10.1109/MRA.2006.1638022.

[2] T. Bailey and H. Durrant-Whyte, "Simultaneous localization and mapping (SLAM): part II," in IEEE Robotics & Automation Magazine, vol. 13, no. 3, pp. 108-117, Sept. 2006, doi: 10.1109/MRA.2006.1678144.

[3] A. J. Davison, I. D. Reid, N. D. Molton and 0. Stasse, "MonoSLAM: Real-Time Single Camera SLAM," in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 29, no. 6, pp. 1052-1067, June 2007, doi: 10.1109/TPAMI.2007.1049.

[4] R. Mur-Artal, J. M. M. Montiel and J. D. Tardas, "ORB-SLAM: A Versatile and Accurate Monocular SLAM System," in IEEE Transactions on Robotics, vol. 31, no. 5, pp. 11471163, Oct. 2015, doi: 10.1109/TRO.2015.2463671.

Claims

CLAIMS1. Apparatus for nondestructively testing a test component, the apparatus comprising: an ultrasonic transducer, the ultrasonic transducer comprising a transducer housing; one or more cameras attached to the transducer housing, the one or more cameras being oriented relative to an active face of the ultrasonic transducer; and a computation unit that is operably adapted to receive ultrasonic signals from the ultrasonic transducer and camera signals from the one or more cameras, wherein the ultrasonic and camera signals are generated by the ultrasonic transducer and the one or more cameras, respectively, in a time synchronized manner.
2. Apparatus as claimed in claim 1, wherein the one or more cameras are oriented substantially orthogonally to said active face of the ultrasonic transducer.
3. Apparatus as claimed in claim 1 or 2, wherein said transducer is an EMAT transducer that is adapted to emit linearly polarized shear waves with one or more known polarization directions.
4. Apparatus as claimed in claim 1, 2 or 3, wherein said computation unit is adapted to compute a location and pose of the apparatus or the ultrasonic transducer relative to the environment using simultaneous localization and mapping (SLAM) algorithms and to localize a source location of the ultrasonic signals and relative position.
5. Apparatus as claimed in any one of claims 1 to 4, wherein two or more wheels or spherical balls, optionally of diameter R>3mm, are attached to said transducer housing so that the transducer housing is mounted centrally between the wheels or balls, thereby orientating the active face of the ultrasonic transducer, optionally substantially perpendicularly, to a radii of said wheels and at an optionally variable offset (for example <2,3,5 or 10 mm) from a surface of a test component that the wheel or balls are in contact with during operation.
6. Apparatus as claimed in claim 5, wherein a remote controllable motor is attached to each wheel so that the apparatus can be driven across said surface.
7. Apparatus as claimed in any one of claims 1 to 6, wherein the computation unit is adapted to combine the signals generated by the one or more cameras and the ultrasonic transducer and received therefrom, to output a position of an external surface as well as an internal surface of the test component and the location and orientation of any internal defects that may be present in the test component.
8. Apparatus as claimed in any one of claims 1 to 7, wherein said ultrasonic transducer is adapted to operate in a frequency range between 20kHz and 500kHz
9. Apparatus as claimed in any one or claims 1 to 7, wherein said ultrasonic transducer operates in a frequency range between 1-5MHz.
10. Apparatus as claimed in any one of claims 1 to 9, wherein the computation unit is adapted to only produce a successful output measurement point if the following conditions are fulfilled: a) 3D coordinates and a pose of a location of the ultrasonic transducer are available, and, b) the ultrasonic transducer has recorded a signal from the test component that has a signal to noise ratio (SNR) that exceeds 6dB, optionally that exceeds 12dB, optionally that exceeds 20dB.