Research projects

AI Hub for Productive Research & Innovation in Electronics

Our national consortium funded by UKRI comprises 20 academic institutions and over 30 industrial partners that brings to market AI-based tools to boost productivity across the entire electronics industry supply chain. APRIL is transforming the electronics industry by developing world-leading AI researchrelevant to the electronics industry, building a strong expandable network that can influence the direction of research, policy and regulation, enabling the adoption of technology into business through the creation of AI tools or translation into new start-ups and generating a strong pipeline of talent thatspans AI and electronics. The APRIL Hub will lead the way in AI research, focusing on key areas of the electronics supply chain: Material Discovery; Device Design; Circuit & System Design; Test & Verification; and Modelling. APRIL’s development of AI tools will lead to faster, cheaper, greener and overall more power-efficient electronics.

Adiabatic Capacitive Artificial Neuron

Bigger and more capable AI systems require increasingly more power to operate at full capacity, straining global energy supplies and limiting the democratisation of AI. Project ACAN uses charge recovery to implement AI systems with far higher power efficiency; aiming to lower the power consumption of such systems by a factor of 10x while remaining fully compatible with current, tried and tested CMOS manufacturing technology. The approach is particularly promising when applied on neural networks, where it leverages their parallelism. Our aim is to demonstrate how modular ACAN-type artificial neurons can be implemented and combined together into practically usable networks of virtually any shape and size. The ability to build tailor-made, ultra-low power artificial neural networks, such as project ACAN offers, is critically important in areas where both cost and power budgets are severely constrained with applications in low-power, long-lasting implants for medical use and microchips for autonomous drones as prime examples of impact. This work is funded by DSTL.

Functional-oxide Reconfigurable Technologies

The EPSRC-funded programme grant FORTE brings together world-leading expertise across the University of Edinburgh, Imperial College London and the University of Manchester with the ambition to transform ICT through delivering highly scalable, resilient, power efficient and affordable reconfigurable electronic systems. Our approach is targeted on rebalancing the UK electronics remit from a ‘more’ to ‘beyond Moore’ era via developing functional oxide-based memristor-based architectures and integrating these with CMOS for boosting conventional electronics capabilities. This will allow our society to efficiently expand the operational envelope of electronics, enabling the use of electronics in inaccessible environments as well as reusing or re-purposing electronics affordably.

Shiwei Wang is the holder of a Bayes Innovation Fellowship, awarded to academics with a compelling research translation ambition. Shiwei focuseson innovating novel chip technologies that interface with the brain. A short-term target is to change the limitation that no technologies today allow reliable neural recording and stimulation in a concurrent manner (i.e., read from and write to the brain simultaneously). This is an urgent unmet need because testing and fine-tuning of devices intended to treat or diagnose neurological conditions are difficult without knowing the brain’s immediate responses to stimuli. Shiwei’s team has tackled this challenge using a novel integrated circuit chip that can readout tiny neural signals with high fidelity even at the presence of large stimulation artifacts. The technology has been validated both in-vitro and in-vivo and through the fellowship he is translating this innovation into commercial products. The immediate impact is to allow more reliable data from pre-clinical testing, accelerating the neurological healthcare innovation.

AI MeTLLE - Memristive Technologies for Lifelong Learning Embedded AI Hardware

The Royal Academy of Engineering’s Chair in Emerging Technologies scheme aims to identify global research visionaries and provide them with long term support to lead on developing emerging technology areas with high potential to deliver economic and social benefit to the UK. In 2020 Themis Prodromakiswas awarded a Chair in Emerging Technologies for his work on Memristive Technologies for Lifelong Learning Embedded AI Hardware, using innovations in nanotechnology to create a new electronic fabric that merges memory with computing power while maintaining extreme power efficiency – alike the human brain. The research, using memristors or memory chips based on transition metal-oxides, focuses on enabling electronic systems to sense, recognise, learn and reason, with the goal of embedding artificial intelligence.

Artificial intelligence Enabled Guide for Investment Strategies

AI is a rapidly evolving and highly disruptive technology being implemented within the financial services industry, for example in dealing with an enormousvolume of financial transactions and decisions with wide-reaching implications in the real world. There is potential for catastrophic incidents if the risks associated with implementing AI in this field are not adequately managed. Project AEGIS uses generative AI to carry out trading or insuring activity in a high-quality, trustworthy and self-correcting manner, and to understand what the limits of this approach are. We are developing software that can perform important trading/insuring tasks autonomously and can be equipped with a rudimentary set of circuit-breakers. This will be accompanied by a set of benchmark tests illustrating how the software can be expected to perform under various market conditions and outlining the challenge and opportunity space in this technology. Thus, AEGIS seeks to set the foundations for what we fully expect will become an ever more elaborate system, eventually “earning the trust” to work side-by-side with a human in the financial industry, as per our vision.

This joint Energy Technologies Institute (ETI) and EPSRC fellowship award is held by Julianna Panidi whose research focus is on developing methods for improving the sustainability of solution-processed solar cells and using them as light-powering sources. Attention is given to the material selection and processing methods for the development of organic solar cells without compromising their power conversion efficiency. Another important aspect for Julianna’s focus is their stability, which will be evaluated in both indoor and outdoor settings. By combining electrical and morphological characterisation, Julianna aims to understand the main degradation mechanisms, which will then support future research developments.

Low-Power, High-Speed, Adaptable Processing-In-Sensing Capability

ProSensing aims to define a novel approach to embed intelligence locally, enabling training at the edge by developing novel in-sensing processing elements (enabling electronic and photonic control) that will be combined with tinyML technologies. The development of an in-sensor processing architecture using emerging devices (RRAMs) for image classification is proposed, with the capability of being used in various domains including light, RF, IR, and gas. This will significantly reduce the data transmission penalty, where only relevant features are transmitted to the near-sensor high-level processing and tinyML layers. Furthermore, we aim to eliminate the analogue-to-digital interfacing complexity by processing the extracted features as analogue vectors. RRAM-based Analogue Content-Addressable Memories (ACAMs) will capture pretrained analogue templates to act as a near-sensor classifier. This is a collaborative programme with the Universities of Strathclyde and Nottingham Trent.

EPSRC and MoD Centre for Doctoral Training in Sensing, Processing, and AI for Defence and Security

This University of Edinburgh led Centre for Doctoral Training Programme focuses on generation-after-next technologies for information processing in defence and security, spanning the entire range from hardware development to algorithmic AI development. SPADS is training the next generation of highly professional defence scientists, including engineers, computer scientists, and mathematicians, capable of leading developments in cutting-edge and generation-after-nexttechnologies in information and communication technology that are poised to transform not only the world of defence and security, but also broader civilian society. This will be approached by underpinning interconnected technology and sensing modalities working across multiple sensing domains, leveragingadvances in autonomy, embedded systems, and AI across both software, algorithms and hardware. Alex Serb is a Co-Director on this CDT, leading onNovel computing and Beyond-CMOS hardware.