Podcast #226 Shoestring Digital Manufacturing

TRANSCRIPT

Ken: Good day, and welcome to Momenta's Digital Thread podcast. Today, I'm pleased to host Duncan McFarlane, Professor of Industrial Information Engineering at the University of Cambridge and lead the Shoestring Digital Manufacturing program. Duncan is a fellow of St. John's College and Head of the Distributed Information and Automation Laboratory within the Institute for Manufacturing. His work involves the development of manufacturing automation and control systems and, more broadly, using digital systems across the industrial supply chain. He had started at BHP in engineering and research roles before joining Cambridge in 1995 as a lecturer in industrial automation systems. He was appointed Professor of Service and Support Engineering in 2006 and Professor of Industrial Information Engineering in 2011. Duncan has co-led key industry initiatives, including the Auto-ID Center, the Innovative Manufacturing Research Center, the AERO ID program, and most recently, the Digital Manufacturing on a Shoestring program. Notably, he won the RAE President's Award for providing industrial engineering support to local hospitals managing the COVID-19 epidemic. He is co-founder and chairman of RedBite Solutions, an Industrial RFID and Track and Trace Solutions Company. Duncan, welcome to our Digital Thread podcast today.

[00:02:04]

Duncan: Ken, thank you very much. It's a pleasure to be here.

[00:02:06]

Ken: A pleasure to have you. You have a very impressive background, and I'm happy to say our backgrounds have woven on more than one occasion along the way, so I've long awaited to finally have a good conversation with you. Thank you for launching this Digital Manufacturing on a Shoestring program; it was the perfect topic today. We call this the Digital Thread podcast, and I always like to start with the question, what would you consider your digital thread? 

[00:02:35]

Duncan: I mean, I've been working industrially since the early 1980s, and as you said before, I began in the steel industry. Interestingly, we talk about digitalization being new, but even back in the 1980s, industry was pretty digital, perhaps in a more rudimentary way. Again, doing the hard yards, trying to deploy IT systems and new algorithms and ideas within steel plants, I segued into the aerospace and space industry, working on robust control in the late '80s with BAE Systems, for example. I then went back into the steel industry, having learned how to control machines in my Ph.D., found that there weren't any hard machine control problems in the steel industry. So I moved out of that and instead got more interested in how you would make the whole material flow through industry, particularly steelmaking. I left there in the mid-'90s and moved to Cambridge and took some of the ideas I worked on in the steel industry, which was about how—this is the early '90s—we were looking at how to improve industrial control using AI ideas and particularly trying to understand how to use AI to help manufacturing operations manage in the face of disruption. That was interesting, and AI was probably pretty rudimentary in those days. Then, in the late '90s, I got involved through a few eccentric discussions in the area of RFID, working with MIT, as you know, on something called the Auto-ID Center, and RFID radio frequency tagging was a little bit boring by the sound of it. It was just about having an electronic version of a barcode. But I did see a way to connect that to some of the AI work we've been doing, and that was interesting. We had a huge program that led to the worldwide rollout of RFID and industrial supply chains. I spent time at a company, RedBite, which went on to do all sorts of interesting things. It still runs in the information tracking and the Internet of Things space. That program, the Auto-ID center, launched the phrase 'Internet of Things' which actually- the chap that did it, a guy called Kevin Ashton, who was at Procter and Gamble- it was ahead of its time, I don't think it kicked off till the late 2000s.

From there, as you said in your introduction, I took many of the ideas we worked on in manufacturing and worked in services. It's very interesting just to understand the engineering role in services. I got caught up in construction for a while, and I've done a lot of work in logistics and, more recently, in healthcare. The interesting thing is that most of those domains are just slightly more customer-focused than manufacturing often is. Then, in the mid-2010s, Industry 4.0 started to arise. Industry 4.0, to begin with, I was a little bit grumpy about it because I felt like it was corralling together many things that we'd all worked on for several years. Ultimately, I'm very grateful because it took many advanced notions and the research we'd worked on and put them into an industrial context. Some of the challenges in Industry 4.0 for small companies led us to the Shoestring program, which tries to do simple things for small companies with digital solutions. That's my thread. I don't know if it's a thread, a weave, or a jumble. Anyway, that's it.

[00:06:13]

Ken: It sounds like a complex fabric, as in a textile, due to the multitude of threads interwoven within. However, I must chuckle at your use of the term "grumpy"; it brings to mind the notion of standing on the shoulders of giants. In a sense, Industry 4.0 couldn't have emerged without the pioneering groundwork laid by those before us to construct its infrastructure. As you mentioned, we're building upon that foundation with digital manufacturing or the shoestring program. It's intriguing how we continually develop these structures over time.

One aspect of your career that I truly admire is its traditional aspect. We tend to categorize practitioners/operators and academics separately, but in your case, you've excelled in both realms. You were already a seasoned practitioner before embarking on an academic path. I'm curious—what ultimately motivated you to pursue teaching?

[00:07:10]

Duncan: I should say that I began at BHP when I was 16 and stayed with them for about 14 years. By the time I was 30, I had left. This experience greatly impacted me and has certainly stayed with me. One of the challenges I faced there was getting to know the steel industry well. Despite BHP also having interests in minerals and petroleum, much of the R&D was focused on steel at the time. Additionally, it's worth noting that BHP, now BHP Billiton, no longer has a significant presence in the steel industry. However, I gained extensive experience in that particular industry.

What inspired me to begin teaching wasn't a desire to teach. In fact, I was drawn to the role at Cambridge because it offered me the opportunity to work with various companies in different sectors and understand the diverse challenges they faced. My journey at Cambridge began after a round of golf with my old supervisor, who was part of a different engineering department. He mentioned, "Well, look. They're interested in recruiting you for this manufacturing group." Curious, I asked, "What do they do exactly?" He explained, "They frequently travel around the country in minibusses, visiting factories." To me, that sounded fantastic, so I eagerly accepted the offer.

I also saw an opportunity to challenge conventional approaches to engineering teaching. The manufacturing part of the engineering department was already driving a much more practical orientation into how students were educated. Whether through student projects, exposure to industry, or inviting industrial speakers, we provided hands-on activities in our automation lab to build industrial systems. This approach appealed to me, as it addressed many of the challenges I faced as an undergraduate.

This inspired me to become more involved in teaching, aiming to encourage more students to gain practical industry experience.

[00:09:21]

Ken: You are a professor of Industrial Information Engineering at Cambridge University. Can you tell us a bit about what attracted you to this field? But more importantly, what is Industrial Information Engineering versus some of the same more common traditional industrial or information sciences?

[00:09:40]

Duncan: This is an interesting question. When I got this chair, I sat down with the head of the department, and in Cambridge, it's left to you to decide what the title of your chair is. She said, "What do you want it to be?" I'm never good at these things, and I said, "I don't know. Automation Engineering? Information Engineering?" She said, "Well, we've got professors of Information Engineering already." They do lots of stuff with matrices and math and do clever analysis, and it didn't feel that was me. It might have been me 15 years earlier, but the word 'industrial' was missing.

I said, "I'm not sure if I'm too worried about the rest, but I want the word 'industrial' in my title." So, very much the focus was on how I bring information or automation engineering developments into industrial use. I had to do interviews for chartered engineering status in recent years. In one of the interviews, the IET panel in the UK said, "We'd been looking at you. You're not a conventional sort of academic, are you?" I didn't quite know how to take that, but in the end, I decided it was positive. I guess what has happened to me is that my industrial experiences significantly influenced the nature of the way I did R&D. I mentioned that when I did my Ph.D., I worked on some very interesting algorithms for robust control satellites. Going back into the steel industry, I was told there's only one hard problem in the company. But there were a lot of other complex problems that ought to be looked at. I also talked to a colleague in Australia who worked on radar, and he said, "Actually, these algorithms you're working on, they're really good. But honestly, they're less than 5% of the solution." I got interested in what that other 95% was. I built my research group around that, making it problem- and solution-driven. If you go to the website now, it has changed in 30 years, but it's very much focused on bridging the gap between academic developments and industrial needs.

I have this thing called the "dial hamburger." Dial is my research group, Distributed Information and Automation Lab. The hamburger, the top of the bun, is industrial practice, and the bottom of the bun is academic development. The meat in the burger is that interfacing piece. We work very hard at being good at that interfacing piece in drawing problems and issues down from industry, then moving solutions back up once we've achieved something that matches industrial needs. The other thing I've done within my group is make sure that every project involves working with a company, and that's been pretty consistent over 30 years. That forces me, my research team, and my students to be able to not only work on clever solutions but to keep them simple and articulate in terms of the needs a company might have, and that's been a really rewarding thing to have achieved because I think every student that I've had that's been- sometimes forced, mostly willingly- with our company, has gained significant life skills, as well as research skills from that process. Over these 30 years, three dominant issues remain as the core thrust of the research group: smarter automation, making better use of information, and managing in the face of disruption and change. None of those look like they are going away anytime soon. Perhaps the last thing to say on this is- as I said, I began looking at AI and its role in developing solutions for the industry. The shift's occurred over the last couple of years; it's still looking at intelligence and AI, and these sorts of ideas, perhaps less so within the solution we deliver, but within the approach we use to build the solutions, has been a rather interesting spin for me.

[00:14:01]

Ken: I was going to ask you about what changed during your tenure. But I think you did a great job of already describing that. Interestingly, the starting point for you was AI, and now here we are again post-AI winter. Suddenly, it's become an extremely hot topic again, especially relative to the optimization of industrial processes. Perhaps you can go back for a moment. You mentioned something about what really is different is the approach. Maybe you can say a few more words on that because I'm kind of curious about that. 

[00:14:36]

Duncan: Particularly with our work with small manufacturers as part of this Shoestring activity that you mentioned earlier, we realized that what we need to deliver for a small company with very limited resources and skills in the digital area is a solution that is as simple and straightforward to understand, deploy, use, and maintain. On average, small companies' needs don't immediately lend themselves to AI. I believe AI will have a future role once companies become more digitally enabled, but our key focus right now is to deploy these solutions as easily as possible and assist organizations in their deployment. The AI opportunity for us lies not in simplifying or enhancing the solution itself but in streamlining the solution process, such as acquiring, customizing, and building it using AI in those stages. I think this approach aligns closely with the generative AI approaches that have been evolving in the last couple of years, which serve more as advisory tools than what I call closed-loop AI tools. I also want to mention another fundamental shift for me, aside from the internet, which is IT's evolution from being a service function within a business to a strategic imperative. In the 2000s, many companies outsourced their IT, but today, IT, information, and AI are integral to business strategy. For me, this shift, enabled by the internet, has been a major turning point in my career.

[00:16:44]

Ken: It's an interesting thought. IT was always described as I came up with in both manufacturing and IT: IT was 'keep the lights on.' That was the major focus there. Of course, you may get an OT side; it was all about efficiency, productivity, safety- everything that goes with manufacturing processes, and now all of a sudden, we changed it to digital. It has become a core competency, and a competitive one at that, driving many companies. We interview a lot of Chief Digital Officers, and it's interesting to see how they're changing a business. But it's not necessarily just off-the-shelf technologies; it's the application of catalytic technologies, if you will, toward business transformation change and business models that go with it. You're absolutely right. Hadn't quite thought about that metamorphosis where we suddenly saw it so critical that we weren't outsourcing it anymore, but it certainly has happened along the way. Let's go ahead and drill down on Digital Manufacturing on a Shoestring. This is what really attracted me to this conversation. You're looking at how low-cost, readily available digital technologies can be implemented to support growth and productivity in small and medium-sized enterprises, or SMEs, as we like to call them. What inspired you to create this initiative?

[00:18:00]

Duncan: Well, I mentioned one of the inspirations was this grumpiness about Industry 4.0. Actually, the unfair grumpiness- but I also went to many Industry 4.0 seminars, workshops, and events in the mid-2010s. I found myself at various events, talking to people from small companies, and the overwhelming message that I got at that time was that some of the pretty shiny, exciting new options coming through in Industry 4.0 had limited resonance with small companies. The comments were along the lines of, "Gosh, I'm sure I'll never have enough money to get involved, work with those sorts of systems," or "We don't have the digital capability in-house to engage with this type of development." The limitations that small companies were seeing were one of the issues. I also did some digging because I thought, well, small companies- how many are there anyway? Maybe not that many. The astonishing thing for me at the time was finding out that in the UK alone, about 200,000 small manufacturers were operating, and they had an average number of employees of around two to three. In my sense, most of them are sitting at the bottom of what we call the digital pyramid. The digital pyramid has the big players at the top, another set of medium-sized businesses underneath, and everyone else underneath. Those organizations were a little bit disenfranchised by these developments around Industry 4.0. That was a key factor.

A second thing was a growing understanding- and working with my students helped this a lot- that through very low-cost systems and components becoming available, open-source software became more professional in operation and development. More generally, there is a shift in balance in developments in digital systems outside compared to inside the industry, so looking outside the industry for some digital tools might have been a good pathway. Then, finally, I realized that our next generation of students, apprentices, trainees, etc., will all come forward with great hands-on digital skills, but possibly not the digital skills we expect of industrial engineers at the moment. In fact, I had a student who came to me one day and said, "I want to build a smart fridge for my project." I said, "That's a six-week project. Are you kidding?" He said, "Well, I think I could do it. I want to do something else in the project as well. But I think I could do it in two to three weeks. I just need a GoPro camera and a Raspberry Pi temperature sensor and get some AWS access; we should be able to do it."

Anyway, he had a go in his six-week project. He didn't quite get there but spent the rest of the summer fixing it. But I was shocked by the speed that he was able to put this together and thought, surely, we must be able to do better than that for industrial companies we work with. Shoestring kind of spun out of those different notions into how we can do something very low-cost for small manufacturers and keep it simple so they can just get started using digital technologies.

[00:21:22]

Ken: How is this Shoestring program organized? How do the small to medium enterprises participate in it?

[00:21:29]

Duncan: Shoestring, as you mentioned earlier, focuses on low-cost solutions for small and medium-sized enterprises. We really aim to provide companies with an end-to-end approach where we not only assist them in adopting a solution but also engage in specifying, designing, deploying, and maintaining the solution. Our goal is to introduce a simple digital solution to as many SMEs as possible and help them progress on the digital pyramid, as I mentioned earlier.

During our discussions with companies, which is crucial for us, we noted that although there are numerous small companies, many of them have similar needs, often quite straightforward. For instance, one of the most requested solutions is a job state tracking system, which answers the question, "Where is my job in the factory at the moment?" This type of system is commonplace in larger companies, but many small companies lack such a simple tracking mechanism. That's precisely the kind of solution Shoestring provides.

We initially began as a program within a UK-funded research project by the Engineering and Physical Science Research Council. This initiative allowed us to develop a design approach, tools, and solutions and conduct pilot projects. Based on our success, we received funding to establish a Shoestring business unit in 2022. As a not-for-profit organization, we oversee and develop the approach, manage deployment through regional contracts in the UK and overseas, engage in research programs, direct sales, and increasingly collaborate with third-party providers. Our aim is not to deploy hundreds of thousands of solutions ourselves but to empower others to do so.

[00:23:39]

Ken: What have been some of the notable results you've seen in another way to think about it? What has surprised you the most about the program's impact?

[00:23:49]

Duncan: It's interesting. I've probably had around 50 research grants in my career, possibly even more, and you begin each of them thinking, "This is going to be the big one," and sometimes they are, sometimes they're not. One surprise was the level of engagement and enthusiasm from the companies involved in the research project right from the start. About one and a half years into the project, the group of companies came to see me - and keep in mind, we were in the middle of COVID during the initial phase of the work - and they said, "You need to have a plan for what happens when the research project ends." I reminded them that it still had a year and a half to go, but they insisted on having a plan. The big surprise was that this was what the companies wanted from the beginning. We found that we were pushing on open doors whenever we spoke to people.

The program was initially intended to conduct a couple of pilots, but we ended up doing 20 pilots, most of which were carried out during COVID lockdowns and deployed over the phone with enthusiasts in the companies. Almost all of them were successful, with people not only pleased with the results but also with being actively involved in achieving them.

I think there's also a side benefit that has been a bit of a surprise: the impact on digital skills and training, whether it's for end users or organizations involved in deployments. To date, we've deployed about 100 Shoestring or Shoestring-based solutions as we continue to refine our tools, streamline our approaches, and become more proficient. We focus heavily on our modular, reusable design approach. Interestingly, we've developed about eight modules that encompass basic technologies, and we use them repeatedly, much more than we initially anticipated.

Another surprise was that after we started the program, several companies I regularly work with - Boeing, Exxon, Amazon, Unilever - all approached us and said, "We don't get it; why can't we join this project? We want to be involved." I explained that it was intended for small companies and low-cost solutions, but they remained interested. Consequently, we've since undertaken quite a bit of work with large companies, exploring different approaches and how they can utilize low-cost solutions. We also received unexpected inquiries from around the world, including one during COVID from a gentleman in Perth, Western Australia, named Andrew Duffy, who expressed interest in starting a Shoestring program in Western Australia and asked for our support. This interest spread to Egypt, New Zealand, and South Africa, among other places. It's been an incredibly exciting five or six years since we launched the program, and it shows no signs of slowing down.

[00:26:44]

Ken: Look, it's a very commendable program. It's interesting how, in some sense, we saw the reshoring initiatives, de-globalization initiative, some might call it- that are all of a sudden creating much more demand for digital capabilities onshore, local, if you will. Before the call, we talked about local optima versus global optimist. I think it certainly comes to mind, and it sounds like you have been doing a great job already scaling. I'm curious. If somebody wants to get involved, how did they find out more?

[00:27:18]

Duncan: Well, digital shoestring.net is probably the first point of contact for information, and as I mentioned, we now directly sell solutions from that website. However, what we've been striving to achieve, which ties into your comment about scaling, Ken, is the recognition that if we want to deploy on the scale we're aiming for - thousands and tens of thousands - it won't be our team alone accomplishing that. We're focused on establishing regional capabilities for deploying Shoestring solutions and identifying individuals or organizations within those regions who can facilitate this.

Another approach through which many organizations participate is by operating as regional clusters. For instance, we're currently running a program in Cornwall, UK, where a local engineering organization is conducting deployments in collaboration with the Australian team based in Cambridge. Similarly, in the Western Australian initiative, there's an engineering company partnering with us. Our aim to involve more people can be achieved through direct sales, regional programs, and, increasingly, through the development of relationships with third-party developers.

[00:28:29]

Ken: It's excellent work. I must say, it's interesting to see how organically you continue to grow it. Again, very commendable. As we wrap up, I'm always curious how you maintain your edge as a leader, particularly in digital manufacturing. Any recommendations you'd like to highlight for our listening audience?

[00:28:49]

Duncan: That's an intriguing question. If I've managed to maintain an edge, it's likely not through conventional sources. By that, I mean if you rely solely on what's prevalent in the mainstream press or even in typical academic literature, you tend to follow a linear path of development. Personally, I'm drawn to unconventional avenues, which is why I appreciate the Shoestring approach. It's an area that hasn't received much research attention because the focus often lies on enhancing functionalities rather than reducing costs. I draw inspiration from unexpected sources. While the company itself isn't unusual, there's often an underestimation of the innate capabilities of individuals within organizations. Not just researchers but operators on the shop floor can provide invaluable insights. Back in my days in the steel industry, I discovered that I could glean a lot of information by simply talking to the person operating the equipment I was trying to automate.

Furthermore, I indulge in unconventional hobbies like writing poetry, drawing, and solving cryptic crosswords, which serve as sources of inspiration. Poetry, for instance, offers a succinct way of expressing ideas, while cryptic crosswords provide a self-checking mechanism, which I find fascinating. My interest in minimalism and simplicity has also played a role in my approach.

In my research, I've delved into the history of automation and found it intriguing how ingenious automation existed before the advent of digital systems. Early innovations like water clocks and steam engine governors demonstrate this. This leads me to ponder the possibility of exploring non-digital technologies in the future.

Lastly, I'm currently exploring the philosophy of logic through a book I'm reading. While AI is a hot topic, I believe there's still much to uncover in this field. Perhaps a deeper understanding of logic and probability is necessary before fully harnessing the potential of AI. This might peg me as an unconventional academic, but that's where I draw my inspiration from.

[00:31:40]

Ken: Wow, I love the power of open-ended questions. We start with the digital thread and end with this one because you never know the answers you'll get. I think the last- probably two or three minutes so describe who you are as a person. I'm debating- I'd love to do another podcast. Is it crypto crosswords or the teaser you put out there on AI? We'll know which one it's going to be, but what a great conversation, Duncan. Thank you so much for sharing this time and these insights with us today.

[00:32:12]

Duncan: It's been really enjoyable. This has been a great chat, Ken. I'd love to do another one.

 [00:32:17]

Ken: Oh, believe me, we'll put a point on the next one around AI because I think- we're just as you say, right at the beginning of that revolution. This has been Duncan McFarlane, Professor of Industrial Information Engineering at the University of Cambridge and lead of Shoestring Digital Manufacturing. Thank you for listening, and please join us for the next episode of our Digital Thread podcast series. We wish you a momentous day. You've been listening to the Momenta Digital Thread podcast series. We hope you've enjoyed the discussion, and as always, we welcome your comments and suggestions. Please check our website at momenta.one for archived versions of podcasts, as well as resources to help with your digital industry journey. Thank you for listening.

[The End]