Projects

Completed

Basic and strategic applied research

PHYSICAL SCIENCES AND MATHEMATICS (Metallurgy and Materials)
PHYSICAL SCIENCES AND MATHEMATICS (Computer Science and Informatics)

Not Cross-cutting

Ian Kinloch
University of Manchester

Standard

EPSRC

01 October 2025

01 April 2026

6 months

£299,122

Unknown

North West

Artificial Intelligence and Robotics

Ian Kinloch, University of Manchester

Jacqui Cole, University of Cambridge
Thomas McDonald, University of Manchester
Aron Walsh, Imperial College London

The National Materials Innovation Strategy has highlighted developing a digital thread through materials discovery, manufacturing, in-service performance and (re)use – aka Materials 4.0 - as the highest priority to realise accelerated scientific and economic benefit across the UK materials supply chain.  This short proposal will produce three case studies to underpin future programmes and investment for delivering Materials 4.0, empowered by AI, to ensure innovative global leadership for the UK in this vital area. AI is key tool in delivering Material 4.0. The widespread adoption of emergent detailed data capture and control capability, alongside AI tools and techniques, will revolutionise the pace and impact of materials’ discovery, optimisation and automation across many industries from national infrastructure to defence. To achieve this, the AI needs include relevant data access (historic and future), applying materials-informed machine-learning and language models, to forecast emerging materials innovation. This is underpinned by automated data capture, storage and analytics that include the AI-driven characterisation of materials and properties, and identify structure-property relationships that can be exploited. A UK transition to Materials 4.0, extracting the transformational power of AI, must be cohesive for maximum impact, facilitating the integration of digital tools and data development across diverse industries and sectors. The Henry Royce Institute (Royce), the EPSRC’s national materials institute, aims to work with the materials community to establish the framework for delivering the necessary skills, best practices and infrastructure. This  includes consideration of a national capability (e.g. a UK Materials Informatic Centre) to drive cross-sector change and draws on inspiration from other sciences (e.g. European Bioinformatics Institute) and countries.  Our initial focus will be on automating materials data capture, storage and analytics and integrating this into an exemplar ‘design-to-device’ supply-chain for data-driven materials discovery and performance. This includes the exploitation of open-source materials datasets and materials-domain-specific language models that Royce has collated for the materials community via its new Digital Materials Foundry (launched 16 May 2025: https://www.royce.ac.uk/programmes/digital-materials-foundry/). This aligns with §1.2 of the UK government’s AI Opportunities Action Plan. We will exploit materials-informed machine-learning and big-data analytics for materials and property prediction. This will be done through incorporating AI enabled surrogate model approaches to bridge numerical and analytical models that cover multiple length scales up to component/device level; this includes their interface into manufacturing needs where our centre will ultimately connect into wider programmes and challenge-led activities such as Made Smarter. Royce has commissioned work to develop a framework and implementation strategy for a national, connected framework that will accelerate the widespread adoption and integration of Materials 4.0 across the UK’s materials innovation ecosystem and industrial supply chain.  This proposal will provide three underpinning case studies to understand the architecture needed for Materials 4.0.  These cases will focus on (1) AI driven automation of data collection and analysis for facilities, (2) development of FAIR data infrastructure for autonomous energy materials discovery, and (3) optimising a factory production line’s efficiency using AI analysis of historical data

14/01/26