BAE Systems was one of the industry challenge owners for the Made Smarter Technology Accelerator (MSTA) programme. They were looking for a solution to a unique inspection challenge: to automatically analyse digital X-ray images with a high level of confidence. How could an AI system be trained when there were very few examples available?
Through the MSTA, Machine Intelligence (an innovative startup with unique machine learning software), took on the challenge, and worked with BAE Systems to develop a prototype that would solve the problem.
The project plan was to develop a proof-of-concept demonstrator that put Machine Intelligence’s technology to work within BAE Systems’ inspection process, and which could lead to its broader application for other inspection types.
The Machine Intelligence team had already been able to train AI models using only a handful of examples. Their technology generates computer programs that can be read by humans, so that they can understand how AI makes decisions, and give them the confidence to use the system for visual inspections.
The BAE Systems project was highly relevant to Machine Intelligence’s existing technology, and with the additional funding that the MSTA provided, they could use this opportunity for further development. This would also be their first time working directly with the end customer, rather than through intermediary partners, and would also prove to be a crash-course in dealing with major suppliers.
While Digital Catapult provided workshops, mentoring and technical expertise, BAE Systems committed significant resources to guide the project, produce data and provide feedback to help make the project a success.
Because the MSTA ran within a set timescale, Machine Intelligence selected development routes that would not be dependent on single points of failure, such as not being able to get the exact right example data quickly. With everyone recognising these risks from the start, there were hardly any delays, as everyone worked together to find alternative work paths.
Through the MSTA, Machine Intelligence has been able to update its core AI technology, and now has a product to market. BAE Systems now has a clear technology demonstrator to make processes much more efficient and deliver a better product to customers.
Machine Intelligence found that the mentorship aspect of the programme was extremely useful – and at a level that would not usually be accessible or affordable for a small business.
The team has also gained a new network of contacts, within and beyond BAE Systems. This includes other finalist innovators, with whom they are already planning collaboration on future projects.
Simon Harding, Director and Technical Lead, Machine Intelligence, said:
"The MSTA gave us an exciting opportunity to engage with a world-leading company and further develop our AI technology.
The programme exceeded our expectations, as previous interactions with funding bodies have been a bit more hands-off. The level of involvement was just right to help us as a smaller business working with a top-tier company. We are now well-placed to continue to grow and innovate in the field of AI-powered visual inspection."
Yves Conan, Software Engineer, Machine Intelligence, said:
"Adding new capabilities to our machine learning system, and delivering a complete new user interface were challenging tasks, but through the MSTA we got to try new platforms and technologies that made our software production ready in a few short months."
Susan Harding, Director, Machine Intelligence, said:
"The business mentoring helped us become more competitive, and has significantly influenced how we run the company."
The challenges. Every day brings something new to learn or a new problem to solve. I formed the company to take my academic research into the “real” world – so in many ways I am working on my passion project.
It’s been a great experience.
BAE Systems has recognised Machine Intelligence’s technology as something with great potential within its manufacturing and they are excited by the possibilities. BAE Systems and Machine Intelligence is on a learning curve trying to work out how and where we can best fit Machine Intelligence’s solution into its processes. It has been a real team effort on both sides.
Advanced digital technologies change everything. Machine learning will impact all aspects of manufacturing. In this programme we are ‘just’ looking at inspection, but the ability of a machine to either augment or replace a person at every stage of manufacturing will dramatically improve the cost and quality of the produced items, and free up people to do other important jobs.
I’m excited that the solution has many end applications. Machine learning, when applied to imagery, has limitless opportunities – many outside of manufacturing.
Some of these can make a real difference to people – whether it is making the products they buy safer or more efficient, or when these techniques are applied to medical imagery to help with spotting and diagnosing disease.
Machine learning is transforming what we can do, and here at Machine Intelligence we look forward to providing the tools.
That giant companies can be excited to work with micro-businesses. Our conversations with BAE Systems have revealed that there is real enthusiasm around leveraging the skills and technology that exist outside the world of tech giants and multi-nationals.
What are you most excited about heading into the minimum viable product phase of the programme?
The prototype phase gave BAE Systems and Machine Intelligence a taste of what is possible. The minimum viable product phase will allow us create something amazing that not only works for BAE Systems, but moves Machine Intelligence’s technology and solution forward.
Machine Intelligence Ltd develops computer vision and machine learning techniques for industry and academia. It offers expertise in systems integration, software development and AI.
Website: www.machineintelligence.co.uk
LinkedIn: Machine Intelligence Ltd
The Made Smarter Technology Accelerator (MSTA) was designed to create opportunities for collaborative change that would deliver impact for industry. The programme enabled four innovative startups to produce a minimal viable product to address a specific industry challenge. As well as support from Digital Catapult, tech sponsors and the industry challenge owners, the programme provided funding of £100,000 per project.
Part of the national Made Smarter movement, the MSTA was delivered by Digital Catapult and funded by UKRI through the Industrial Strategy Challenge fund.
Don’t miss out on Digital Catapult’s updates – with regular news, insights and opportunities in advanced digital technologies.
If you have any additional questions about the Made Smarter Innovation Sustainability Programme, please email madesmartersustainability@digicatapult.org.uk
Register to find out how Made Smarter can support you on your digital journey.