Digital twins are computer-based models of an object or even a human organ, and the technology behind them is advancing at a rapid pace. EPFL will host a series of talks and seminars on this topic at its Digital Twin Days on 15 and 16 November. An ideal opportunity to learn more about the work EPFL researchers are doing in this pioneering field.
In the end, it’s all about making copies. Digital twins replicate an object or a human organ and work in parallel with – and just like – the original. In reality, they’re a combination of data, algorithms and artificial intelligence, and they exist in a virtual world, like an avatar. Today digital twins are used in just about every industry: manufacturing, healthcare, architecture, transportation, communication systems and more. “Digital twins are computer models of things that exist in the real world,” says Prof. Frédéric Kaplan, the head of EPFL’s Laboratory for Digital Humanities. “The term can also refer to models of more abstract processes, such as a production schedule. So you can think of digital twins as both data-generated models and computer simulations.” Prof. Dimitris Kyritsis, who heads EPFL’s ICT for Sustainable Manufacturing Group, agrees: “Digital twins are generated with modeling and simulation software, and they let us study and measure physical mechanisms and test the effects of different decisions in a virtual world. In addition, thanks to IoT technology, the computer models – which could even be holograms – are permanently connected to their real-world twins, enabling us to better understand the systems we’re using and adjust how they operate.”
Several EPFL labs are working together through the Center for Intelligent Systems to develop digital-twin technology and applications. They’re tackling issues such as how to make cities more energy-efficient, how to design personalized treatments for patients, and how to evaluate the plausibility of different hypotheses about the past. The hope is that digital twins will allow scientists to find responses to these challenges and many others. However, it’s clear that implementing this kind of technology will require a broad range of skills and know-how.
A cross-disciplinary effort
Prof. David Atienza, the head of EPFL’s Embedded Systems Laboratory, and Prof. Adrian Ionescu, the head of EPFL’s Nanoelectronic Devices Laboratory – both within the School of Engineering – are working together on digital-twin technology. Prof. Ionescu is developing sensors that can collect data on patients, such as heart rate, muscle movements and even brain activity through electroencephalogram scans, as well as on ambient conditions such as air moisture levels, temperature and pressure. Meanwhile, Prof. Atienza is looking at what kinds of sensors are needed for different applications and where they should be placed on a human body or a machine. His lab collects data from the sensors and feeds them into AI programs. “Our role is to compile all the data and create a complete image of the item being studied, so that we can then build a digital twin,” he says.
Models of the entire human body
Prof. Ionescu and Prof. Atienza believe that when it comes to healthcare, the goal of digital twins should be to create real-time models of the full human body. They are also convinced that this technology will be available within the next few decades, despite the advanced skills needed and exacting standards that must be met. “The most complicated kind of digital twin is undoubtedly that of the human body,” says Atienza. “It involves different systems, all of which communicate with each other, and phenomena occurring on different scales. And the more accurate we want the digital twin to be, the longer it takes to develop. There are still a number of development steps and versions that we need to work through.”
Applications in manufacturing and city planning
EPFL researchers are also exploring the use of digital twins in manufacturing. For example, engineers at Prof. Kyritsis’ lab are developing digital twins that can monitor and control production lines in real time. “This technology could be particularly useful for hazardous production processes, such as in the more dangerous areas of steel mills and power plants,” he says.
Next year, EPFL, in association with other organizations, will build its first digital twin of a city as part of the Blue City project. The aim will be to create a digital twin of Lausanne that models all its various systems and interactions: from microbes and waste to energy, water, materials, people, vehicles and more. This should help city officials meet their sustainability targets and improve the quality of life in Lausanne. “In the fields of architecture and urban planning, digital twins are data-based replicas of buildings or entire cities,” says Prof. Jeffrey Huang, the head of EPFL’s Media and Design Laboratory, which straddles the School of Computer and Communication Science (IC) and the School of Architecture, Civil and Environmental Engineering (ENAC). “Yet unlike conventional 3D models, which depict the visible aspects of these systems – such as the shapes of buildings, the materials used and other physical elements – digital twins also provide insight into the invisible aspects like noise, pollution and dynamic flows. This additional information is extremely helpful for architects, urban planners and city residents.”
Digital twins for surveillance?
Digital twins are poised to enter our daily lives. But before that can happen, more research is needed – not just on the science and engineering, but also on the human aspects. “We still don’t have the right skills,” says Prof. Kyritsis. “We need to train or reorient experts and bring them up to date with the latest developments, because digital twins require abilities at the cutting edge of technology.” What’s more, digital twins are data driven, and much of that data will come from individuals. Prof. Huang notes, “It’s easy to imagine a bleak future where digital twins are used for surveillance purposes and managed by a handful of corporations or governments. To prevent that from happening, universities need to be involved in the development work, and the ethical implications need to be examined from all angles with regard to how digital twins are designed, so that we can ensure openness, transparency and equal access.”
In light of the considerable advancements in digital-twin technology, we may one day see a parallel world where digital twins interact with each other in the virtual space. For more on that prospect, stop by EPFL’s Digital Twin Days.
Digital twins actually date back to Apollo 13
“Some scientists point to the Apollo 13 accident 50 years ago as the first real use of digital twins,” says Prof. Frédéric Kaplan, the head of EPFL’s Laboratory for Digital Humanities. “When the explosion occurred, NASA engineers on the ground had to troubleshoot the problem and come up with a solution – even though the vessel was hundreds of thousands of kilometers away. The three astronauts on board were trapped inside the vessel and couldn’t visually inspect the damage caused by the explosion.
“The NASA engineers were able to run simulations of the main spacecraft components using networked computers. For instance, four computers were used to simulate the command module and three more to simulate the lunar module. These computers were synchronized with data transmitted from the spacecraft – which meant that it was crucial to maintain data transmission between the vessel and mission control. In short, the simulations met all the criteria for digital twins.
“And it was thanks in large part to the simulations performed on the connected computers –the digital twins – that the engineers and astronauts were able to work together, pinpoint the problem and ultimately save the mission. What’s particularly impressive is that the simulation system wasn’t actually a single digital twin but a collection of digital twins – each one modeling a specific component – that communicated with each other.”