Textination Newsline

Researchers have trained an artificial intelligence to design the structure of so-called metamaterials with desired mechanical properties for a wide range of applications.

• ETH researchers have used artificial intelligence to design metamaterials that show unusual or extraordinary responses to complex loads.
• Their new AI tool deciphers the essential features of a metamaterial’s microstructure and accurately predicts its deformation behaviour.
• The tool not only finds optimal microstructures but also bypasses time-consuming engineering simulations.

Bike helmets that absorb the energy of an impact, running shoes that give you an extra boost with every step, or implants that behave just like natural bone. Metamaterials make such applications possible. Their inner structure is the result of a careful design process, following which 3D printers produce structures with optimised properties. Researchers led by Dennis Kochmann, Professor of Mechanics and Materials in the Department of Mechanical and Process Engineering at ETH Zurich, have developed novel AI tools that bypass the time-consuming and intuition-based design process of metamaterials. Instead, they predict metamaterials with extraordinary properties in a rapid and automated fashion. A novelty, their framework applies to large (so-called non-linear) loads, e.g. when a helmet absorbs major forces during an impact.

Kochmann’s team has been among the pioneers in designing small-scale cellular structures (similar to beam networks in timber-frame houses) to create metamaterials with specific or extreme properties. “For example, we design metamaterials that behave like fluids: hard to compress but easy to deform. Or metamaterials that shrink in all directions when compressed in a particular one,” explains Kochmann.

Efficient, optimal material design
The design possibilities seem endless. However, the full potential of metamaterials is far from realised, since the design process is based on experience, involving trial and error. Furthermore, small changes in the structure can give rise to huge changes in properties.

In their recent breakthrough, the researchers succeeded in using AI to systematically explore the abundant design and mechanical properties of two types of metamaterials. Their computational tools can predict optimal structures for desired deformation responses at the push of a button. Key is the use of large datasets of the deformation behaviour of real structures to train an AI model that not only reproduces data but also generates and optimises new structures. By leveraging a method known as “variational autoencoders”, the AI learns the essential features of a structure from the large set of design parameters and how they result in specific properties. It then uses this knowledge to generate a metamaterial blueprint whenever the researchers specify its desired properties and requirements.

Assembling building blocks
Li Zheng, a doctoral student in Kochmann’s group, trained an AI model using a dataset of one million structures and their simulated response. “Imagine a huge box of Lego bricks – you can arrange them in countless ways and over time learn design principles. The AI does this extremely efficiently and learns essential design features and how to assemble the building blocks of metamaterials to give them a particular softness or hardness”, says Zheng. Unlike prior approaches using a small catalogue of building blocks as the basis for design, the new method gives the AI freedom to add, remove, or move building blocks around almost arbitrarily.  Together with Sid Kumar, an assistant professor at TU Delft and a former member of Kochmann’s team, they showed in a recently published paper that the AI model can even go beyond what it has been trained to do and predict structures that are far better than anything ever generated before.

Learning from the movies
Jan-Hendrik Bastek, also doctoral student in Kochmann’s group, used a different approach to achieve something similar. He used a method originally introduced for AI-based video generation, which has become commonplace: if you type in ‘an elephant flying over Zurich’, the AI generates a realistic video of an elephant circling the Fraumünster Church. Bastek trained his AI system using 50,000 video sequences of deforming 3D-printable structures. “I can insert the trajectory of how I want the structures to deform, and the AI produces a video of the optimal structure and the complete deformation response,” explains Bastek. Most previous approaches have focused on only predicting a single image of the optimal structure. However, giving the AI videos of the entire deformation process is crucial to retain accuracy in such complex scenarios. Based on the video sequences, the AI can create blueprints for new materials, taking into account highly complex scenarios.

Big benefits for bike helmets and shoe soles
The researchers have made available their AI tools to the metamaterials community. This will hopefully lead to the design of many new and unusual materials. The tools are opening new avenues for the development of protective equipment such as bicycle helmets and for further applications of metamaterials from medical engineering to soft robotics. Even shoe soles can be designed to absorb shocks better when running or to provide a forward boost when stepping down. Will AI completely replace the manual engineering design of materials? “No,” laughs Kochmann. “Used well, AI can be a highly efficient and diligent assistant, but it must be given the right instructions and the right training – and that requires scientific principles and engineering knowhow.”

• For sports and physiotherapy alike

Researchers at ETH Zurich, Empa and EPFL are developing a 3D-printed insole with integrated sensors that allows the pressure of the sole to be measured in the shoe and thus during any activity. This helps athletes or patients to determine performance and therapy progress.

In elite sports, fractions of a second sometimes make the difference between victory and defeat. To optimize their performance, athletes use custom-made insoles. But people with musculoskeletal pain also turn to insoles to combat their discomfort.

Before specialists can accurately fit such insoles, they must first create a pressure profile of the feet. To this end, athletes or patients have to walk barefoot over pressure-sensitive mats, where they leave their individual footprints. Based on this pressure profile, orthopaedists then create customised insoles by hand. The problem with this approach is that optimisations and adjustments take time. Another disadvantage is that the pressure-sensitive mats allow measurements only in a confined space, but not during workouts or outdoor activities.

Now an invention by a research team from ETH Zurich, Empa and EPFL could greatly improve things. The researchers used 3D printing to produce a customised insole with integrated pressure sensors that can measure the pressure on the sole of the foot directly in the shoe during various activities.

“You can tell from the pressure patterns detected whether someone is walking, running, climbing stairs, or even carrying a heavy load on their back – in which case the pressure shifts more to the heel,” explains co-project leader Gilberto Siqueira, Senior Assistant at Empa and at ETH Complex Materials Laboratory. This makes tedious mat tests a thing of the past. The invention was recently featured in the journal Scientific Reports.

One device, multiple inks
These insoles aren’t just easy to use, they’re also easy to make. They are produced in just one step – including the integrated sensors and conductors – using a single 3D printer, called an extruder.

For printing, the researchers use various inks developed specifically for this application. As the basis for the insole, the materials scientists use a mixture of silicone and cellulose nanoparticles.
Next, they print the conductors on this first layer using a conductive ink containing silver. They then print the sensors on the conductors in individual places using ink that contains carbon black. The sensors aren’t distributed at random: they are placed exactly where the foot sole pressure is greatest. To protect the sensors and conductors, the researchers coat them with another layer of silicone.

An initial difficulty was to achieve good adhesion between the different material layers. The researchers resolved this by treating the surface of the silicone layers with hot plasma.
As sensors for measuring normal and shear forces, they use piezo components, which convert mechanical pressure into electrical signals. In addition, the researchers have built an interface into the sole for reading out the generated data.

Running data soon to be read out wirelessly
Tests showed the researchers that the additively manufactured insole works well. “So with data analysis, we can actually identify different activities based on which sensors responded and how strong that response was,” Siqueira says.

At the moment, Siqueira and his colleagues still need a cable connection to read out the data; to this end, they have installed a contact on the side of the insole. One of the next development steps, he says, will be to create a wireless connection. “However, reading out the data hasn’t been the main focus of our work so far.”

In the future, 3D-printed insoles with integrated sensors could be used by athletes or in physiotherapy, for example to measure training or therapy progress. Based on such measurement data, training plans can then be adjusted and permanent shoe insoles with different hard and soft zones can be produced using 3D printing.

Although Siqueira believes there is strong market potential for their product, especially in elite sports, his team hasn’t yet taken any steps towards commercialisation.

Researchers from Empa, ETH Zurich and EPFL were involved in the development of the insole. EPFL researcher Danick Briand coordinated the project, and his group supplied the sensors, while the ETH and Empa researchers developed the inks and the printing platform. Also involved in the project were the Lausanne University Hospital (CHUV) and orthopaedics company Numo. The project was funded by the ETH Domain’s Advanced Manufacturing Strategic Focus Areas programme.

Alzheimer's and other dementias are among the most widespread diseases today. Diagnosis is complex and can often only be established with certainty late in the course of the disease. A team of Empa researchers, together with clinical partners, is now developing a new diagnostic tool that can detect the first signs of neurodegenerative changes using a sensor belt.

Forgetfulness and confusion can be signs of a currently incurable ailment: Alzheimer's disease. It is the most common form of dementia that currently affect around 50 million people worldwide. It mainly afflicts older people. The fact that this number will increase sharply in the future is therefore also related to the general increase in life expectancy.

If dementia is suspected, neuropsychological examinations, laboratory tests and demanding procedures in the hospital are required. However, the first neurodegenerative changes in the brain occur decades before a reduced cognitive ability becomes apparent. Currently, these can only be detected by expensive or invasive procedures. These methods are thus not suitable for extensive early screenings on a larger scale. Empa researchers are working with partners from the Cantonal Hospital and the Geriatric Clinic in St. Gallen on a non-invasive diagnostic method that detects the early processes of dementia.

Signs in the unconscious
For the new method, the researchers Patrick Eggenberger and Simon Annaheim from Empa's Biomimetic Membranes and Textiles lab in St. Gallen relied on a sensor belt that has already been used successfully for ECG measurements and has now been equipped with sensors for other relevant parameters such as body temperature and gait pattern. This is because long before memory starts to deteriorate in dementia, subtle changes appear in the brain, which are expressed through unconscious bodily reactions.

These changes can only be recorded precisely when measurements are taken over a longer period of time, though. "It should be possible to integrate the long-term measurements into everyday life," explains Simon Annaheim. Skin-friendly and comfortable monitoring systems are essential for measurements that are suitable for everyday use. The diagnostic belt is therefore based on flexible sensors with electrically conductive or light-conducting fibers as well as sensors for motion and temperature measurement.

To enable such long-term measurements to be used for monitoring neurocognitive health, the researchers are integrating the collected data into in-house developed mathematical models. The goal: an early warning system that can estimate the progression of cognitive impairment. Another advantage is that the data measurements can be integrated into telemonitoring solutions and can thus improve patient care in their familiar environment.

Suspicious monotony
The human body is able to keep its temperature constant in a range of 1 degree Celsius. The values naturally oscillate in the course of the day. This daily rhythm changes with age and is conspicuous in neurodegenerative diseases such as dementia or Parkinson's disease. In Alzheimer's patients, for example, the core body temperature is elevated by up to 0.2 degrees Celsius. At the same time, the spikes in daily temperature fluctuations are dampened.

In a study, the researchers have now been able to show that altered skin temperature readings measured with the sensor belt actually provide an indication of the cognitive performance of test subjects – and can do so well before dementia develops. The test subjects in the study included healthy people with or without mild brain impairment. This mild cognitive impairment (MCI) does not represent a disability in everyday life, but it is considered a possible precursor to Alzheimer's disease. The subjects took part in long-term measurements and neuropsychological tests. It was found that a lower body temperature, which fluctuated more throughout the day, was linked to a better cognitive performance. In individuals with MCI, body temperature varied less and was slightly elevated overall.

The heartbeat is also subject to natural variations that show how our nervous system adapts to sudden challenges. The small silence between two heartbeats, about one second in duration, has great significance for our health: If this pause always remains the same, our nervous system is not at its best.

A study by researchers from ETH Zurich determined that poorer measurements in older, healthy people can be improved within six months through cognitive-motor dance training. In these "exergames," the test subjects imitated sequences of steps from a video. In contrast, participants who instead only trained in straight lines on a treadmill, but also trained their memory, benefited less.

"The point is to intervene early with appropriate training as soon as the first negative signs can be measured," says Patrick Eggenberger. "With our sensor system, any improvements in cognitive performance can be tracked through movement-based forms of therapy." Studies with long-term monitoring will now be used to clarify how the sensor measurements can be used to predict the progression of mild brain disorders.

Further information
Dr. Simon Annaheim
Biomimetic Membranes and Textiles   
Phone +41 58 765 77 68
Simon.Annaheim@empa.ch