How did Omni-NDE get started?
It started with an industrial computed tomography (CT) conference in Wels, Austria. I was helping develop pre-clinical photon-counting CT. My career had been in biomedical imaging, but I started looking at what other markets might be attractive. Wels is the only conference focused on industrial CT. That’s where I got the idea to use photon-counting CT to solve industrial problems. There are great applications for CT in industry, but it has tended to lag behind biomedical applications. This was great because there was development work to do.
I started to develop photon-counting CT for industrial applications, which led to the idea behind Omni-NDE. It's common in medical imaging that you either see structure or function — but rarely does a single scan or modality give both. In medicine, it’s common to take the strengths of a technology that's good at structural information and combine it with one that's good at functional information. That gave me the idea of overlaying, like CT and ultrasound (UT) — CT gives good structural information about the part, its details, construction, and different materials, while UT gives good functional information — delaminations and disbonds.
If you saw only the outside of this pressure vessel, how would you know what's on the inside? And if you just saw the ultrasound images, you see an anomaly, but where? There's no other reference point from which to work. If you look at this image versus the shearography and thermography images — you can see the outline of the vessel, but where is the defect on the bottle? They put tape on the parts because it's hard to accurately co-locate functional data with structural data. And we confirmed the defect with robotic x-ray CT shown on the right.
But we also wanted something portable. In the industrial world, aircraft wings and rocket fairings don’t fit through a CT donut. That meant robotics, and that’s how Omni-NDE started.
Aren’t robotic NDT systems already being used?
Yes, but almost all robotic UT systems today use water squirters to achieve the coupling for transmitting sound waves efficiently. Conventional $3+ million systems can scan relatively large objects like engine nacelles, but there is still a size limit. Also, the robot movement relies on a CAD model of the part. The technician has to build tolerances because the CAD model never exactly matches the actual part. Those tolerances decrease resolution. It also makes the system inflexible because it takes weeks of work and a specialized technician to develop the routine and optimize the parameters.
Collaborative robots
Cobots—collaborative robots—have advantages over industrial robots. They are smaller, more portable, less expensive, and more accurate. Their torque sensors provide collision detection and allow cobots to be operated around humans. They also enable increased accuracy. This is key for scanning with the precision needed for arbitrary 3D paths without CAD files and overlaying multiple data sets.
How does overlaying data work?
The idea is to know where the part is relative to the robot heads. The robots know where they are in space. However, we must also know where the part is in 3D space. So, we build a very accurate surface map as a skeleton onto which we can merge the data from multiple scans. So, we match the scan data in relation to the part and do that on a part-by-part basis, but it takes hours, not weeks. We can also do this when there is no CAD file.
We’ve spent several years developing this technology, talking to manufacturers and NDT experts. Everyone agrees that the CAD model never matches the part perfectly. That is one reason we’ve taken this approach, and it also sets up a true NDT digital twin — which is the missing piece in the digital thread for each part that companies are developing. They have the design files and the manufacturing data but cannot overlay NDT data onto and through the part. Imagine if you could start providing location-accurate data and correlate it with the sensors you use in processing and simulations.
Multi-modal
We realized that thermography would also be suitable, as it helps scan large areas quickly to identify potential defect areas. Then, UT or CT can be used on a much smaller area to get the details needed. Even though my background is CT, at Omni-NDE we don’t have a stake in any one modality. We want the best solution for the customer. We can trial all of these NDT techniques with side-by-side evaluations. I don't know many companies who do that. There aren’t even many labs with multiple large systems — versus our single cobot-based system — that will say, “Send us your part, and we'll try everything we have and tell you what works best.” We think that’s the best approach — to test before investing in Voidsy thermography, Xarion laser ultrasound, or whatever. We can save companies the time of researching and sorting the options by showing proof of why one particular solution or combination of solutions is ideally suited to that specific problem or part.
Thus, our systems are modular. Thermography is the solution today, but what if you get a new problematic part, and the best solution is Xarion or CT? With our system, you're not throwing away the initial investment — just adding a new module — like a Swiss Army Knife. And that new modality will have the same interface and robots. You pay for one add-on instead of a whole new system, which requires additional floor space, foundation, etc. So, when you pick a modality, that’s not a final answer. You also don’t have to figure out how to integrate each new tool using some new user interface or robots.
What have you seen using multiple modalities?
One customer sent us a part for a space application with a very odd, complex shape. When we did the scans, we could see a delamination, precisely locate it, and go through that with the customer. And they’re asking, ‘Where exactly is the delamination?’ I pulled up the data overlay onto a 3D map of the part, and they could see it clearly. We can discuss the problem and not spend time trying to describe where exactly the defect is. Having the accuracy to measure the delamination's size and location further eliminates confusion.
We’re also seeing where the different modalities work and where they don’t. We’re scanning a ceramic matrix composite (CMC) part, which is a material where the part starts as a porous body before it’s infiltrated and sintered into a high-temperature 3D-shaped composite. Xarion works well to a certain depth, but these materials can be almost impossible to scan. So we could scan it with micro-CT and see what the customer was looking for.
Contactless
We like to focus on enabling contactless NDE. Not only do aerospace and defense parts have very tight specifications, but drying out a part after squirting it with water or submerging it in a tank takes time and additional steps. If you’re Airbus and making 60 or 75 aircraft per month, saving time on every part could make a real difference in efficiency and cost.
Solving hard problems
Companies come to us because they have problems that existing methods can't solve. When you can't solve a problem, you need better sensors or methods. We provide both.
We're pushing the boundaries of technology to solve complex problems. Sometimes, you need better sensors, like Xarion, which has better spatial resolution than current air-coupled UT. It also allows a significant amount of post-scan analysis that other forms of UT can’t do.
Omni-NDE was started to do something very new and different—apply medical imaging technology and the latest sensors with cobots to enable systems that are portable, modular, multi-modal, contactless, and in line with how digital manufacturing is developing, including AI-assisted scanning and digital twins. We’re excited about how our technology is already being used and how it will continue to grow.