The ethos at the Distributed Algorithms Centre for Doctoral Training resonates with me. It is all about solving real world problems.
Throughout my time as a PhD student, there have been countless opportunities for me to learn; not only within my own specialist domain, but from experts in adjacent fields of research within the group. Seeing how data science can be utilised to tackle complex, multi-domain, real-world problems is honestly (and sadly, depending on your viewpoint) what gets me up in the morning. The impact these data driven solutions can provide is second to none, and being able to unlock that for others is something I genuinely love doing.
Around a year into the PhD, a good friend of mine and I started collaborating to explore potential intersections between our specialised fields, aiming to address a real-world issue. With his knowledge of digital transformation within the advanced manufacturing sector, and my knowledge of data science – we started to scope out what problems there were in his day-to-day that data science could potentially be used to solve. After our fair share of idea generating, exploring and then flattening, we landed on a big problem we thought that we could solve. Emerging Data Technologies Ltd was born.
Through experience and literature review, we understood that the majority of all dimensional and geometrical quality failures within precision manufacturing can be attributed to a defined set of risk factors. We thought it was not unreasonable to deploy machine learning methods to predict these risk factors ahead of time, allowing machine operators to act before the part being created, turns to (expensive!) scrap.
After some initial testing, we put in an application to Innovate UK – the UK’s national innovation agency. We were lucky enough to have our application accepted under the Transformative Technologies scheme giving us the resources to build what we had envisaged only a few months prior. Not only did the grant allow us to build our initial proof of concept, it allowed us to bring another expert from within the DA CDT into the team, accelerating the research and development process.
Now operating as a team of three, we are able to aggregate our views to work towards the common goal of enhancing the precision manufacturing sector. The impact of the technology is best described in the following example:
“Imagine an SME manufacturer with a turnover of circa £10m and a scrap rate of 3%, which is not uncommon. With the adoption of a technology that can identify risk factors contributing up to, say, three quarters of these quality errors ahead of time (2.25% of the turnover), the manufacturer could look to make a saving that equates to £225,000 per annum from direct intervention alone.
The cost savings do not only come from directly reducing scrappage caused by manufacturing errors. Reduction in corrective costs, both in terms of labour and energy, can also be realised from the reduction in rework. Aside from the obvious reduction in machining power consumption, resources within the manufacturers’ operations are also freed up, increasing throughput and overall efficacy of the business.”
Whilst building the technology is important, we realised that there is a requirement for further market validation. With this in mind, we set out to host a stand on Innovation Alley at Digital Manufacturing Week at the NEC in Birmingham. Here, the one-minute pitch practice regularly facilitated by the DA CDT proved invaluable. Being able to deliver the idea succinctly to a variety of potential stakeholders allowed us to gain the feedback we set out to acquire – the response from attendees was overwhelmingly positive.
Since then, we have continued to develop the business on a commercial front by having a successful application to the Future Club at Sci-Tech Daresbury, giving us access to a bustling community of high-tech innovators and tailored support. We have also continued to develop the technology by exploring other machine learning methods that can offer different benefits to those utilised in the original project. The University of Liverpool Enterprise Fund has also provided us with funding to purchase dedicated hardware to accelerate the training and testing of machine learning models.
It’s important to note that this journey has had its share of challenges, which, looking back, are humorously disastrous. In a nearly poetic turn of events, the night before receiving our acceptance email from Innovate UK, the two of us co-founders found ourselves on stage in front of approximately 100 people. At that moment, we blanked out, forgetting our entire presentation, and essentially, as the saying goes, 'crashed and burned'.
With all of this said, the value of being part of such a fantastic group (the DA CDT) cannot be overstated. The relentless opportunities to learn, academic writing practice and the collaborative cohort experience has allowed me to develop as an individual and realise the potential of applying data science techniques to the real world, delivering impact and value.
Back to: Centre for Doctoral Training in Distributed Algorithms