The Cambridge Engineer: Bending Reality (Michaelmas 2016) by The Cambridge Engineer (Cambridge University Engineering Society Magazine)

HOLOGRAMS FLEXIBLE STRUCTURES RUNESCAPE NXT

THE CAMBRIDGE ENGINEER BENDING REALITY

Cover: Armageddon holography Calum Williams, Yunuen Montelongo & Jaime Tenorio-Pearl The aftermath of a piece of debris crash landing on a plasmonic hologram imaged using a scanning electron microscope. The hologram is composed of metallic nanostructures which optically scatter when excited by light at specific wavelengths due to plasmon oscillations and the light constructively interferes forming an image in the far field.

CONTENTS Editorial

Bending Light Advances in 3D holographic projection

Cleaning with AI New intelligent robotics research from Dyson

Centrefold Coiled light

Flexible Electronics Talking polymers and PCBs with Oxford materials’ Hazel Assender

Industrial Experience - Video Games Interview with ARM intern Alessia Nigretti

Bendable Structures Folding paper with Dr Keith Seffen

Re-engineering Runescape Talking to the developers behind Runescape NXT

New Technologies We hear from Leonardo about new GaN research

“When you are applying a lot of excess heat energy to a surface… the surface changes, and the properties change” - Prof Hazel Assender, Page 12

IN SEARCH OF THE ADVENTUROUS 4

IN SEARCH OF REMARKABLE ENGINEERS To be truly adventurous is a remarkable quality. After all, how many people are really prepared to continuously explore new ideas, opportunities and experiences? To take their thinking to places no one has ventured? At Shell, right now, we’re in search of remarkable, adventurous people to join us in our pioneering work on some of the most innovative engineering projects in the world – from the world’s largest gas-to-liquids plant at Pearl in Qatar, to Prelude, the first ever floating natural gas platform. Discover the opportunities at www.shell.com/careers

Shell is an equal opportunity employer

EDITORIAL Hello and welcome to the Michaelmas edition of the CUES Magazine: Bending Reality. The year is well underway and we’re excited to bring you this term’s selection of submissions, interviews and photography. The Cambridge Engineer - CUES’s magazine - is written for students, by students and researchers, and seeks to print a broad range of interesting engineering content. We aim for a mix of material from both inside and outside the department, and to link together multiple disciplines each issue around a central theme. The theme ‘Bending Reality’ was chosen in part because of its broad scope of interpretation: the articles gathered in this issue range from paper folding to flexible electronics, with articles, interviews and images exploring engineering, architectural and physical ideas that change our traditional perception of ‘real stuff’. Dr Calum Williams from the CAPE writes about his work bending light in 3D holographic displays, and we interview Oxford professor of materials engineering Hazel Assender about her work in flexible electronics and the applications it could have. This theme was chosen in order to explore the cutting edge of current engineering developments, both in the department and further afield. Looking at developments that are redefining what we think of as reality - changing both the digital and analogue worlds - gives an exciting insight into the weird and wild aspects of engineering. The theme of the Lent term issue will be ‘Learning’ - this covers anything from Artifical Intelligence to Structural Failure Analysis, educational outreach work or Computational Linguistics. We take submissions in the form of interviews, written pieces, photos and illustrations - to see your name in print, pitch an article to: magazine@cuengineeringsociety.org.uk Enjoy the issue; get in touch! Agnes Cameron and Sara Troyas CUES Magazine editors 2016-17

BENDING LIGHT Calum Williams - a researcher in Cambridge’s CMMPE (Centre of Molecular Materials for Photonics and Electronics) group - along with Yunuen Montelongo, Richard Bartholomew and Prof Tim Wilkinson have designed a new type of pixel element which could make three-dimensional holographic displays possible. In this article, he explains its unique switching capability, and the science behind holographic displays. Real-time dynamic holographic displays, long the realm of science fiction, could be one step closer to reality, after the development of a new type of pixel element that enables far greater control over displays at the level of individual pixels. The results are published in the journal Physica Status Solidi. As opposed to a photograph, a hologram is created when light bounces off a sheet of material with grooves in just the right places to project an image away from the surface. When looking at a hologram from within this artificially-generated light field, the viewer gets the impression that the object was directly in front of them. Currently, the development of holographic displays is limited by technology that can allow control of all the properties of light at the level of individual pixels. A hologram encodes a large amount of optical information, and a dynamic representation of a holographic image requires vast amounts of information to be modulated on a display device. A relatively large area exists in which additional functionality can be added through the patterning of nanostructures (optical antennas) to increase the capacity of pixels in order to make them suitable for holographic displays. In a typical liquid crystal on silicon display, the pixels’ electronics, or backplane, provides little optical functionality other than reflecting light. This means that a large amount of surface area is being underutilised, which could be used to store information. We have now achieved a much greater level of control over holograms through plasmonics: the study of how light interacts with metals on the nanoscale, which allows the researchers to go beyond the capability of conventional optical technologies.

“A hologram encodes a large amount of optical information”

Normally, devices which use plasmonic optical antennas are passive, meaning that their optical properties cannot be switched post-fabrication, which is essential for real-world applications. Through integration with liquid crystals, in the form of typical pixel architecture, the researchers were able to actively switch which hologram is excited and therefore the output image selected. Optical nanoantennas produce a strong interaction with light according to their geometry. Furthermore, it is possible to modulate this interaction with the aid of liquid crystals. The work highlights the opportunity for utilising the plasmonic properties of optical antennas to enable multi-functional pixel elements for next generation holographic displays. Scaling up these pixels would mean a display would have the ability to encode switchable amplitude, wavelength and polarisation information, a stark contrast to conventional pixel technology.

Deforming light: Yunuen Montelongo, Calum Williams and Jaime Tenorio-Pearl Broad diffraction-angle holograms projected onto a sphere. Nano-antenna holograms can produce switchable images in the far-field. Switching reconstructions can be achieved be rotating the orientation of the electric field.

CLEANING WITH AI Dyson gives us an overview of the development of the Dyson 360 Eye™ robot, an intelligent vacuum cleaner that uses a blend of computer vision and systems engineering to change the realities of everyday tasks. AI and machine learning are advancing into everyday life. In the future, the development of artificial intelligence (AI) will accelerate beyond anything we have previously imagined. It will offer limitless possibilities – changing our experiences, transforming every area of life and redefining how we interact with technology. Yet AI is not just future fantasy. It is here and now, gaining momentum through advances in machine learning, neural networks and big data. These are exciting times and the UK is right at the heart of it. Britain is home to some of the most innovative AI companies and will see a dramatic technology shift over the next few years. There is a lot to learn. But at Dyson we’ve already learnt a lot. Recently we launched the Dyson 360 Eye™ robot vacuum. It’s been a long journey that has “Sixteen thrust us into the realm of AI, where we are now making significant moves. Dyson’s AI beginnings

years may seem a long time to work on one piece of technology”

Vacuuming used to be a case of pulling the machine out of the cupboard, dragging it around the room, lugging it up the stairs... all this before the cable gets stuck and the plug pops out of the socket. Non-stop hassle. Thanks to the advances we have made in robotics, it no longer has to be that way. When I joined Dyson in 1998, we set about making our first robot vacuum cleaner. On our original launch date in 2001, we knew we had a machine that was smart for its time – battery powered, minimal human interaction, the leader in the market and a genuine labour saving device. But we also knew there would soon be a better solution – it just wasn’t good enough. So at the eleventh hour, we pulled it. Now, after 16 more years of intensive research and development, we finally have that truly cutting-edge piece of technology: the Dyson 360 Eye™ robot. Last week we launched in the UK, and the US and Canada are still to come. This is after great success in Japan where James introduced it back in 2014.

The journey Sixteen years may seem a long time to work on one piece of technology. But it was important to understand the scope of present and future technology, and to make sure we developed the very best concept. A machine that could achieve much more than simply bumping around a room, picking up the odd piece of dirt. A machine that would revolutionise the way homes are cleaned. As we progressed, we recruited from more and more fields – system test engineers, hardware engineers, firmware engineers, algorithm design and software engineers. The list goes on. Each of them adds a different element to the total picture. With every achievement, new possibilities drove us forward. And today, the endless potential of AI is a major driving force for Dyson’s future.

“The endless potential of AI is a major driving force”

What’s next? When I started in Dyson, we were unknown, especially in robotics. We are now, I’d say, one of the biggest centres for robotic innovation in the UK. We know we’re not perfect. We are always looking for more. However, in the Dyson way, we enjoy the challenges and take encouragement from failure. Back in 2001, we were merely on the cusp of discovering the need for a robot with vision. Since then we’ve built one that uses complex mathematics, probability theory, geometry and trigonometry to map and navigate a room. It sees a bandwidth of light that extends beyond that of the human eye. It is incredibly effective. In essence, we’ve transformed what was an ordinary, frustrating household item into an all-seeing, constantly calculating, autonomously moving vacuum cleaner – one that knows where it is, where it’s been and where it’s yet to clean. So as Dyson puts more investment and more resources into AI, can you even start to imagine what might be round the corner?

Coiled Light: Alexander Macfaden Laser light is launched into a singlemode optical fibre. The fibre only supports one very spatially confined propagation channel; the light which is not in this channel is â&#x20AC;&#x2DC;shed,â&#x20AC;&#x2122; radiating outwards from the coils.

FLEXIBLE ELECTRONICS As we continue to integrate electronics into our everyday lives, the need for electronics to be flexible - in a literal sense - is ever increasing. CUES magazine spoke to Prof. Hazel Assender of Oxford University Department of Materials about the potential for super-cheap, printable devices. There are two main reasons why flexible electronics is a significant research area right now. The first is mass production: techniques such as roll-to-roll printing (as opposed to current electronic fabrication techniques) will allow mass production of electronic products to be cheaper than ever before. The second big driver for flexible electronics is perhaps more obvious: it bends. From applications in clothing to medical sensing,the ability to change the shape of a circuit to fit user needs opens up a huge scope in terms of manufacturing wearable, portable and adaptable devices. Hazel’s work seeks to bring these two properties together. Researching vacuum deposition and roll-to-roll printing techniques, she is currently exploring a number of different problems, including patterning methods for depositing complex printed circuits at high speeds.

“It’s the same ideas, but with completely different materials, made in completely different ways”

A key example application for this work is smart, disposable packaging. A sandwich that can not only tell you it’s sell by date, but the places it has been stored, the temperature it has been kept at, and change price with freshness might seem an unnecessary expense, but could become trivial using information encoded into a simple flexible electronic label. This same technology could be used to turn sticking plasters into monitoring devices - mechanically flexible and disposable, this could provide a much cleaner, cheaper alternative to the kind of medical equipment currently used to care for elderly patients. Bending a thin polymer PCB

Much of Hazel’s more abstract research is focused on polymer surface morphology - a crucial factor in determining the mechanics and quality of a printed circuit. “When you are applying a lot of excess heat energy to a surface… the surface changes, and the properties change”. Right now, the main challenge is with the idea of doing mass-market electronics not with silicon. The organic semiconductors used have a naturally lower electron mobility; “We’re not about to make a fully-flexible computer”, and one major challenge is making the manufactured transistors completely reproducible on a large scale. “It’s the same ideas, but with completely different materials, and made in completely different ways” says Hazel “So, it’s difficult. But it also means that there are a lot of opportunities.”

ARM GAME DESIGN CUES magazine speaks to Alessia Nigretti, a Computer Science student who completed an internship at Cambridge-based hardware giant ARM this Summer. What was it about ARM that made you want to apply for them? As a 20 year old Software Intern and first year student of a Computer Science course, I did not quite know what to expect from an internship in a big and wellknown company like ARM. I joined the tech environment and started getting involved in study and work experience within the Informatics and Engineering field only about two years ago, after moving to the UK – I am originally from Italy – and I have never felt the need to highlight that even if I am a woman I can still be a valid computer scientist.

“ ‘If there is anything on which you feel conﬁdent working, I will not let you work on it’ ”

What kind of team do you work in?

At ARM, I work in the Demo/Gaming team and I love every moment of it. We work on developing demonstration software in virtual reality to showcase the latest ARM technology, and we also get a chance to play around with the HTC Vive and the PlayStation 4 that we have at the office. The team welcomed me and immediately involved in the tasks that they were working on. Not without some difficulties of course, but that is what makes it entertaining! During the first week, my line manager presented the following two months to me saying “If there is anything on which you feel confident working, I will not let you work on it”. I am the kind of person that loves getting out of her comfort zone as a way to learn and grow up, so that single sentence gave me the input and the enthusiasm I needed. What has your experience at ARM been like so far? After less than one month into my internship here at ARM I realised how this company knows how a welcoming and entertaining work environment is part of what makes a job enjoyable. The internationality of the company makes the environment diverse and interesting: there is always something new to learn, either about work or colleagues, which is what makes employees grow as people as well as engineers.

BENDING STRUCTURES CUES Magazine talks to Dr Keith Seﬀen - lecturer in the department and expert in structural mechanics. Along with Dr Simon Guest, Keith runs the Advanced Structures Group, which explores solutions in novel lightweight, deployable and morphable structures. For the past few years Keith has been investigating these ideas using the structural properties revealed by origami (paper-folding) and its cousin kiragami (folding and cutting). Surrounded by labs full of heavy failure-testing machinery and giant beams, Dr Keith Seffen’s set-up seems surprisingly small and low-tech. Pulling out a mug, and a plastic wallet full of paper circles, he proceeds to demonstrate a failure analysis experiment that he’s recently been working on. Placing the paper on top of the mug, and pressing the centre with a pen, he points out 2 cone sections that form in the paper. Governing the shape of everything from folding curtains to the edges of leaves - as well as critical failure zones - are these intersecting cones. The d-cone - or “deficit-cone” forms a larger well around the peak of the e-cone “excess cone”. No matter what size of circle, or how many times the experiment is repeated, these cones are always about the same size: what Keith refers to as the “fundamental motifs” of the deformation.

“You can’t always change the material, but you can give it the right geometry”

This kind of research is critical in failure analysis. “You can’t always change the material, but you can give the material the right geometry”: if you can map how particular structures buckle and deform, you can redesign them to avoid these modes of failure. Cone failure is a mode observed in buckling, particularly in cylindrical structures, and these simple observations reveal a lot about the failure mechanism. Keith’s research is focused on investigating the fundamental motifs associated with how surfaces change shape. Though there have been investigations into structural analysis using

Samples of kiragami experiments:

origami as a tool for a while, kiragami is a relatively recent addition. By allowing material to be cut and added, the number of parameters that may be altered is vastly increased: and with it, the scope for new discoveries. Taking a pair of scissors, he bisects the cone to form a peaked “p-cone“: demonstrating another change in shape, and another family of behaviours.

“When you pull and bend the cut paper, you see something you haven’t seen before.”

One of the most exciting applications of kiragami is the ability to completely change the properties of something which, unstressed, may lie as a flat surface, This kind of research is key in looking at deployable, rollable structures. Different patterns of cutting can completely change the behaviour under strain, and morph new landscapes, and, in the case of the cones, highlight preventable modes of failure. “When you pull and bend the cut paper, you see something you haven’t seen before.” In addition to modelling macroscopic structures, kiragami could well be providing solutions to problems on the micro- and nano-scale. Currently being used in research in the Centre for NanoPhotonics, kiragami has been used to create conductive structures called ‘micro-accordions’. By cutting patterns of slits into copper tracks on bendable surfaces, the micro accordion tracks can be bent and stretched to degrees that would break ordinary wiring, whilst maintaining ~70% of the conductivity. Moreover, this stretching is entirely elastic: the micro accordion track returns to the original position without the absorption of energy. The applications of bendable structural analysis - and thus, the systems that could be modelled using kiragami techniques - are far reaching. “Any application involving deployable structures, flexible electronics - anything where you can bend and pull a surface around”. These kinds of systems are increasingly in demand, and operating to increasingly high quality constraints - structural analysis is key in making these products mechanically resilient and long-lasting. forming new surface morphologies

RE-ENGINEERING CUES sponsor Jagex’s Matthew Burnett (Head of Core Technology) and David Gilham (Lead Graphics Developer) talk us through their next big development. Throughout its 15 years of fantastical worlds and creatures, quests and adventures, and mighty battles against demons and gods, RuneScape has always had technical innovation at its heart. We were founded on a technical advantage and we have always experimented, which has left us with many systems (some good, some… ‘unique’). Like any other business we know that you have to keep moving forward if you want to stay in the game - pun intended. In our case however, we have been – and still can sometimes be – too bespoke to incorporate other technologies.

“One of the reasons why RuneScape became so popular... is its accessibility”

For instance, it was clear that our previous Java-based game client wasn’t appropriate to start taking advantage of more modern hardware, and so we needed a radical change. Additionally, we had been seeing a decline in browser-based play of RuneScape for quite a while, especially as Java applets go out of fashion as a way of delivering web content and support is removed from browsers. This allowed us to put our priority on a new downloadable client. The process of creating new client engine, called NXT, began with an innovation team armed with a remit to investigate the best ways for us to move forward. In a short time, it became clear we were getting somewhere with the performance control that C++ can provide, and the demo we were working on graduated into a project. Over a year of hard effort later, in October 2015, we had something to show at our annual fan convention, RuneFest, and over a thousand people went hands on with the new client before we finally launched NXT in April this year.

RUNESCAPE With NXT we’re much less CPU bound than with the previous Java-based client. There are many reasons for this, but mainly we’ve significantly reduced draw calls, which have a big CPU hit especially if drivers are poorly implemented. Our dynamic geometry batching system is partly responsible for this, along with our new innovative dynamic hybrid occlusion culling system, meaning we’re submitting much less to the GPU per world area than in Java.

“We have always experimented”

We get much more bang for buck on the GPU as we’ve significantly reduced overdraw through better sorting, and our batching significantly reduces GPU context switching. Unlike the previous client, the new C++-based NXT provides easier access to all the latest rendering technologies, which we’re now able to take advantage of with new architecture, lighting and post-processing pipelines. One of the reasons why RuneScape became popular, and has remained so for 15 years, is its accessibility. Players don’t require a high-end gaming PC to experience the many thousands of hours of gameplay. It turns out that the aims of giving the best top end performance and reaching the lowest common denominator are not that far away. While we added features for the best graphics cards (increasing the draw distance, better antialiasing, dynamic shadows) we also lowered the memory and power required for many of the existing capabilities. The result is a client that is designed to cover at least 12 years of hardware, if not a little more. We’re already well into the development of the next round of graphical upgrades for NXT, which will in future include normal maps, larger textures, point light shadows, particle lighting, physically-based shading, improved volumetrics and further improved global illumination. We’ve also got plans to add support for the Vulkan rendering API to reduce driver CPU overhead even further!

NEW TECHNOLOGIES Merv Haynes, head of the Microwave Engineering department at Leonardo Airborne & Space Systems, gives an overview of the company’s GaN capability. The increased availability of GaN technology in recent years has enabled higher-power amplifiers and components, though high costs and performance at microwave frequencies has kept this largely in the research domain. Solutions to these barriers are now coming to fruition and products using this technology are emerging.

“GaN transistors operate at much higher temperatures and voltages”

Gallium nitride (GaN) is a semiconductor commonly used in light-emitting diodes. Its wide band gap of 3.4 eV makes it particularly useful for optoelectronic, high-power and high-frequency devices. As GaN transistors operate at much higher temperatures and far higher voltages than gallium arsenide (GaAs) transistors, they make ideal power amplifiers at microwave frequencies. GaN technology at Leonardo started 20 years ago with active involvement in the “MIGHT” programme, spawning hybrid circuit designs using simple coplanar transistor cells. Later work focussed on MMIC (Monolithic Microwave Integrated Circuit) implementation, allowing microwave operation to be optimised. The MAGNUS programme was a European Defence Agency project to develop a new generation of integrated circuits and sub-systems based on stable, industrial power GaN technology. The aims of MAGNUS were to demonstrate the use of GaN HEMTs (high-electron-mobility transistors) in switching and high-dynamic range. The large-signal transmit performance of the MMIC was encouraging, with an output power of 10W across most of the frequency band being an excellent result. Recent work has also been to investigate wideband hybrid solutions using offthe-shelf GaN power bars. The process used offers the possibility of providing driver amplifiers integrated with the input network, reducing parts, assembly and interconnects to deliver a highly-integrated solution.

Processes are becoming more available to designers for custom MMIC design, along with discrete transistors and modules in GaN. Hybrid assemblies - the starting point for Leonardo’s initial designs - are now flourishing due to the recent availability of new GaN transistors, and it will be interesting to see where these technologies go next.

MAIN SPONSORS

ASSOCIATE SPONSORS