Duratec has partnered with CSIRO and ANU to take part in the former’s Next Generation Graduates Program. The business is sponsoring PhD student and newest team member, Ben Burton, who is looking at ways to harness AI in the inspection of large-scale assets, such as wharves and bridges.

Duratec’s engineering division maintains strong connections with universities and industry specialists across Australia in an effort to share knowledge and stay up-to-date with the latest technological advancements. Moreover, Duratec is committed to giving back to academia and the scientific community by assisting in the development of tomorrow’s engineers.

The CSIRO’s Next Generation Graduates Program is a cohort-based, industry-driven, multi-disciplinary training program designed to equip students with skill sets that will assist them in making breakthrough innovations in the fields of AI and other emerging technologies.

The program prepares students for the problem-based environments they will experience upon entering their respective fields and facilitates their collaboration with researchers and industry professionals. It is hoped that this, in turn, will help build a competitive and capable workforce that will drive the growth of the Australian tech sector.

CSIRO + ANU + Duratec

As part of the CSIRO’s Next Generation Graduates Program, Duratec has partnered with the Australian National University (ANU) to sponsor student Ben Burton, who is currently undertaking a PhD in computer vision. Like other types of artificial intelligence (AI), computer vision seeks to perform and automate tasks that replicate human capabilities. inspec spoke to Duratec’s newest employee to find out more about his work in AI-enabled technology.

Talking AI with Ben Burton

inspec: Hi, Ben, and thanks for agreeing to share with us a bit about your work in AI and computer vision. Could you tell us about yourself and your studies?

Ben: Sure, so my name’s Ben Burton and I’m a PhD student at ANU. I’m currently working with Duratec, ANU and CSIRO to look at ways we can use advanced AI and computer vision to inspect large-scale assets, such as wharves, bridges and multi-storey buildings.

inspec: Could you tell us about the technology you developed – RustSEG?

Ben: RustSEG was our first-generation AI model and it works by processing images of an asset to identify areas of corrosion. We capture the images via drones and feed them into the AI model, which tells us where the defects are located.

At the moment, we’re looking at corrosion, but we hope to be able to use this model to identify cracks, spalling from rebar, concrete discoloration and other common defects. The defects are currently identified using the images, however, eventually, we hope to be able to fit the drones with additional sensors and detect degradation that is not visible to the human eye.

inspec: There must be so many benefits from that method – increased safety, for example.

Ben: The safety benefits are huge, especially when inspecting water towers or chimney stacks. We can easily fly a drone up and around such assets so that no one has to get into a harness and work at heights.

And it’s the same for confined spaces. We have really small drones, which you can fly in pipelines or areas where people wouldn’t normally fit. Another example is the work Duratec is doing at Cockatoo Island (Sydney Harbour Federation Trust project), where engineers need to be up around high cranes or over water. Drones eliminate these safety risks.

The other benefit is consistency. On large assets, it is common for multiple engineers to undertake the inspection and while one engineer might see a defect in a certain light, another might view it differently. AI is really consistent – the same defect has the same level of severity every single time.

Finally, AI never gets tired. You can feed through thousands of photos, it can run overnight and you can scan large assets with increased consistency. That allows the engineers to focus on more complex tasks, such as repair methodologies, rather than defect identification.

inspec: What do you think the future of AI looks like in the context of asset management?

Ben: There are some really cool things you can do with AI in this space. As well as identifying defects, we hope to eventually be able to predict potential degradation before it even occurs. Structures are now being built with more sensors, which means they can collect lots of data. By feeding this into AI models, we’ll be able to recognise a potential defect and take action before it becomes a problem.

Another potential capability is simulation – being able to predict how a defect might develop over years. This could determine if and when to intervene, as in, do we address this now? Or is it unlikely to affect the health of the structure, thereby allowing us to wait?

inspec: Would this take into consideration potential environmental factors and allow us to forecast how a structure will age?

Ben: That’s the Holy Grail of what we could eventually do with this AI technology – apply specific details. For example, a structure might be built on the North Australia coast. Seeing as we have a lot of information about those conditions, we would recommend a maintenance plan that would be quite different for a structure located in the middle of Australia.

inspec: Could you tell us more about the Next Generation Graduates Program?

Ben: CSIRO (Commonwealth Scientific and Industrial Research Organisation) runs the program and has partnered with ANU and Duratec to focus on research, which has industry applications. I started some of this research when I was an undergraduate, however, we didn’t have enough data to create high-quality AI models. We eventually spoke to Duratec, who had thousands of photos of damaged buildings, as well as the associated reports and explanations of repair methodologies. It ended up being a perfect match. Duratec was already using drones before these kinds of AI models were introduced and continues to accelerate its work in this area. Our focus is to find ways to take the big 3D models we’re building and feed them into these AIs.

inspec: What are you passionate about and why?

Ben: I’ve always been passionate about how technology can change things for the better. And I think the really cool thing about AI is that it unlocks prospects that we never thought were possible. So we might have been collecting data for one point of view but suddenly now, with AI, we find that data is really important in discovering something else. I’ve really enjoyed seeing people discover AI applications, such as ChatGPT, for day-to-day purposes and how they can speed up certain processes. But I also like how the work we’re doing in AI can affect much larger systems and free up human beings to make more critical decisions that will never be able to be made by AI.

I think the way that technology can change people’s lives and change the way we go about doing things that we’ve been doing for ages is really interesting.

A lot of these AI applications are still just research but bodies like CSIRO and businesses like Duratec are saying, let’s see how we can use this technology to create positive change. Let’s move it from the research stage and apply it to actual structures. That’s what’s really cool about this partnership.

About Ben Burton

Ben Burton is currently completing a PhD in computer vision at the Australian National University (ANU), Sydney, and has been sponsored by Duratec to take part in CSIRO’s Next Generations Graduates Program. Ben’s focus is AI-enabled advanced materials technology and how it can be used in the inspection of large scale assets, such as wharves, bridges and multi-storey buildings.

In 2022, Ben completed a Bachelor of Engineering (Honours) and submitted a thesis titled RustSEG: Automated Segmentation of Corrosion Using Deep Learning. Developed by Ben during his Honours
year, RustSEG is a corrosion semantic segmentation algorithm, which can take most images and use deep learning to determine if the image contains corrosion.

Duratec’s team of engineers is thrilled to have Ben on board and eagerly awaits his findings in the use of AI to detect defects in concrete-and-steel structures, and the role this technology could play in the future.