Our partnership with the University of Cambridge

Arm supports academic research and recognizes the importance of maintaining strong links between academia and industry for continued research innovation. As part of this mission, Arm Research collaborations have established seven Centres of Excellence (CoE), broadening research opportunities, and strengthening links through a range of activities, including sponsoring students completing their PhDs. Part of a blog series, we look at the inspirational work each of our Centres are undertaking, fueling research success and collaboration between academia and industry.

Arm has a long history of collaboration with the University of Cambridge. Today there are multiple active research projects with staff and students in the Department of Computer Science and Technology (CST) and the Department of Engineering. These range from novel hardware design and related software and formal models, to programming frameworks and ML for applications such as robotics. One of our largest collaborations is with the CHERI team in CST, where we are working together to develop the Morello prototype board, with funding from the UK Government. At the other end of the spectrum, we provide support for PhD students in both departments. This enables us to engage with some of the brightest students that Cambridge has, in areas that are of direct interest to us. Our partnership does not stop there; find out how one email to Arm helped to spark a wave of innovation at the University of Cambridge, with Dr. Timothy Jones and his students, in our blog post.

Currently, we are supporting four PhD students undertaking work ranging from robotics, to binary convolutional neural networks. Take a look at their research projects, and what they hope to achieve.

Kieran Gilday | Reconfigurable modules for rapid design iteration of soft robot applications

Arm supervisor: Vasileios Laganakos

Soft robotics is a rapidly developing field, and compliant robots offer significant advantages when cooperating with humans and their adaptability in real world environments. The design of soft robots is problematic in their passive dynamic behaviors as formal manufacturing techniques have not matured. Kieran’s research aims to develop a framework programming his own passive dynamic behaviors into soft robotic modules. This allows a reduction in actuation and control complexity, without sacrificing the complex adaptive behaviors observed in natural systems, such as the human hand. Kieran is working on his PhD in the bio-inspired robotics lab, has a background in electrical and information engineering, and experience in rapid prototyping.

Hugo Hadfield | Machine Learning for motion control using geometric algebra

Arm supervisor: Sahan Gamage

Hugo’s project looks specifically at clifford/geometric algebra (GA), a mathematical system subsuming quaternion, dual quaternions, Plϋcker coordinates, screw theory, conformal transformations and several other useful mathematical engineering techniques. His research explores the properties of these algebras, as well as bringing the mathematical and computational machinery of GA to bear on very applied engineering problems. Such as 3D reconstruction from images or simulation and control of robot dynamics. Hugo hopes to make an impact on academia by explicitly exposing the geometry that underpins many problems, and in doing so, show some of the parallels between distinct fields of mathematical engineering. Before his undergraduate studies, Hugo was a technology scholar with Cambridge Consultants – we look forward to seeing how his industry and academic experience will develop this project!

“Hugo is tackling complex Robot planning algorithms and software performance in Arm platforms in collaboration with Rene de Jong in Arm Research. The task involves evaluating various planning software suites, forward/inverse kinematic algorithms, various planning and approximation algorithms and defining appropriate “complex” problems and cost functions. He is also developing various missing APIs along the way to fulfilling this. Arm Research and Arm will gain a complete and well researched evaluation framework that can be used internally for future Arm based hardware systems from his work.”

Sahan Gamage – Arm Research, Staff Research Engineer

Aida Miralaei | Energy-efficient in-memory approach for binary convolutional Neural Networks

Arm supervisor: Shidhartha Das

Aida’s PhD proposal was inspired by her work on Big Data projects at Digikala, an online retailer in Iran. Her project plans to run Binary Convolutional Neural Networks (BCNNs) with significant energy efficiency by moving computations out of the processor, and into the main memory. This will allow Aida to skip a major energy bottleneck in running CNNs. Aida’s first step was to implement a processing-in-memory (PIM) module for calculating the inference computations of BCNNs in DDR4 DRAM memory. Aida aims to create a design suitable for a wider range of BCNN architectures so that is it not limited by the design parameters of specific CNNS.

Ben Simner | Armv8-A system-level semantics

Arm supervisor: Gustavo Petri

Some features within architecture may not clearly define the envelope of allowed software-visible behavior. Operating systems and hypervisors make use of more esoteric parts of the architectural interface than other programs do or can, including cache maintenance, page table management, interrupt control, exception handling, and much more. Ben’s project aims to give these parts of the architecture clear and precise definitions, clarifying the architecture for both hardware and software development, enabling creation of testing and verification tools for both, and improving the quality of hardware implementations and systems software. Ben began his PhD in 2018 at the University of Cambridge in the Programming Languages and Semantics group, and previously obtained his master’s at the University of York. Some of Ben’s work has already been published, so we are intrigued to see what else will come of his work.

“This project attempts to formalize the system semantics of computer architectures (instruction fetching, address translation, etc.). This is a foundational piece of work towards the verification and safety of security-critical software. The benefits to Research, and Arm in general, are a definitive formal model of these architectural mechanisms, which have so far not been mathematically specified.”

Gustavo Petri – Arm Research, staff research engineer

Engaging in cutting-edge research

“Having one of the world’s leading academic institutions located in close proximity to our Cambridge office provides multiple opportunities to engage with cutting-edge research and some of the top researchers in their fields. From the deep engagements on the individual projects previously highlighted, to participation in a wide range of project advisory boards and technical discussions across topics as diverse as e-health, novel battery substrates, and swarm behavior, this is a relationship that brings a wide range of benefits to both of us, and we look forward to developing it further in future.”

Andrea Kells – Research ecosystem director

Arm Research Collaborations

Our Centres of Excellence are just one way that we collaborate with academia and industry. Find out more about our Collaborations team and how we are helping shape the industry, and read our recent blogs highlighting the work we do. In terms of tools and support, we can also provide research access to IP and technical support. This allows our partners and future collaborators the chance to work with and understand our vision at Arm, helping us build the future of technology.

This post is part of a wider series, allowing you to get to know our Centres of Excellence. Click the following links to find out more:

University of Southampton: An Arm Research Centre of Excellence