A method for optimizing the production of advanced materials based on Imperial research was chosen as an example of AI for the public good.
Empire spin-out company Polaron has been shortlisted for the inaugural Manchester Prize, a government-run competition to support breakthroughs in artificial intelligence for the public good. Ten finalist teams will be awarded £100,000 to further develop their solutions, with one finalist set to receive the £1 million grand prize next spring.
It’s really exciting to be part of an ecosystem where we can interact with other great AI startups. Isaac Squires Polaron
“The Manchester Prize really raises our profile and we're really excited to be part of an ecosystem where we can interact with other great AI startups,” said Isaac Squires, Polaron's chief executive and part of the founding team of the Dyson School of Design Engineering.
In addition to the cash prize, shortlisted companies will benefit from a comprehensive support package, including funding for computing resources, investor readiness support and access to a professional network.
“The entire AI field is so dynamic and rapidly changing right now, so you'll benefit greatly from being part of a community that knows what the most exciting new tools are and how to deploy them,” says Dr. Sam Cooper, Polaron's other co-founder and the company's chief scientist.
Optimizing material production
Polaron was founded earlier this year to develop generative machine learning algorithms based on Imperial research that can accelerate the design of advanced materials for applications such as batteries and high-performance alloys.
“Many AI companies are working on materials discovery, promising to find, for example, the next generation of battery cathode materials or the next generation of solar panels,” says Dr Cooper. “Polaron is doing something different: we're optimizing products that manufacturers can already make with existing production lines and materials. It's an approach that will enable the industry to move forward more quickly.”
The traditional way to approach this type of process optimization is a combination of engineering intuition and rules of thumb about what works, but because the design space is vast, this trial-and-error approach can take years.
Polaron optimizes products that manufacturers can already produce with their existing production lines and materials.” Dr. Sam Cooper Polaron
Polaron is transforming this materials design approach by learning the relationship between a material's microstructure and how it is manufactured directly from image data. These models enable rapid exploration of potential designs, shortening design cycles from hundreds of weeks to hundreds of hours.
“We use imaging data of the microstructure of materials to train our models,” said Dr. Steve Kench, Polaron's third co-founder and chief technology officer. “This allows us to link how materials are made with how they work, allowing us to rapidly design materials that will perform best, including how fast the battery will charge, how much it will store, and how cheap it will cost to manufacture.”
The potential benefits are significant, with optimization increasing output by 10% and reducing R&D costs by more than 50%.
Agile AI
Unlike many AI companies that use large models developed elsewhere or train their own large-scale models, Polaron's models are relatively small, allowing companies to train their own bespoke models.
The data most companies already collect for quality control is enough to start these optimization workflows. Dr. Steve Kench Polaron
The models can also start working with just a moderate amount of information about the process: “The level of data collection that most companies already do for quality control purposes is enough for us to start these optimization workflows,” says Dr. Kench.
In addition to optimizing battery materials, Polaron's approach can be applied to a wide range of materials, from concrete and composites to biomaterials and alloys, and the company is currently planning pilot studies with early adopters to demonstrate the power of its approach.
AI for the public good
The Manchester Prize is organised by the UK Department of Science and Innovation and will be awarded each year for the next 10 years to an innovation that has the potential to transform lives across the UK and maintain the UK's position as a world leader in cutting-edge innovation.
“The potential for AI innovation to boost UK economic growth is huge, with the sector already employing more than 50,000 people and expected to deliver billions of dollars in benefits to the UK economy over the coming years,” Chancellor of the Exchequer Jeremy Hunt said. “That's why this funding is one of the best investments we can make.”
The prize received nearly 300 applications from innovators, academics and entrepreneurs across the UK. “Selecting the 10 finalists from such a diverse field of applications was challenging, but choosing the winner will be even harder,” said Dr Hayatun Sillem, CEO of the Royal Academy of Engineering and one of the Manchester Prize judges. “We look forward to seeing how the ideas develop over the coming months.”
