aerosol therapy

Aerosol therapy: An ex vivo model of lungs

/0 Comments/in , , /by

A researcher in Health Engineering at Mines Saint-Étienne, Jérémie Pourchez, and his colleagues at the Saint-Étienne University Hospital, have developed an ex vivo model of lungs to help improve medical aerosol therapy devices. An advantage of this technology is that scientists can study inhalation therapy whilst limiting the amount of animal testing that they use.

 

This article is part of our dossier “When engineering helps improve healthcare“

Imagine a laboratory where, sat on top of the workbench, is a model of a human head made using 3D printing. This anatomically correct replica is connected to a pipe which mimics the trachea, and then a vacuum chamber. Inside this chamber is a pair of lungs. Periodically, a pump stimulates the natural movement of the rib cage to induce breathing and the lungs inflate and deflate. This physical model of the respiratory system might sound like science-fiction, but it is actually located in the Health Engineering Center at Mines Saint-Étienne.

Developed by a team led by Jérémie Pourchez, a specialist in inhaled particles and aerosol therapy, and as part of the ANR AMADEUS project, this ex-vivo model stimulates breathing and the illnesses associated with it, such as asthma or fibrosis. For the past two years, the team have been working on developing and validating the model, and have already created both an adult and child-sized version of the device. By offering an alternative and less expensive solution to animal testing, the device has allowed the team to study drug delivery using aerosol therapy. The ex vivo pulmonary model is also far more ethical, as it only uses organs which would otherwise be thrown away by slaughterhouses

Does this ex vivo model work the same as real lungs?

One of the main objectives for the researchers is to prove that these ex vivo models can predict what happens inside the human body accurately. To do this, they must demonstrate three things.

First, the researchers need to study the respiratory physiology of the models. By using the same indicators that are used on real patients, such as those used for measuring the respiratory parameters of an asthmatic, the team demonstrate that the model’s parameters are the same as those for a human being. After this, the team must analyze the model’s ventilation; for example, by making sure that there are no obstacles in the bronchi. To do this, the ex vivo lungs inhale krypton, a radioactive gas, which is then used as a tracer to visualize the air-flow throughout the model. Finally, the team study aerosol deposition in the respiratory tract, which involves observing where inhaled particles settle when using a nebulizer or spray. Again, this is done by using radioactive materials and nuclear medical imaging.

These results are then compared to results you would expect to see in humans, as defined by scientific literature. If the parameters match, then the model is validated. However, the pig’s lungs used in the model behave like the lungs of a healthy individual. This poses a problem for the team, as the aim of their research is to develop a model that can mimic illnesses so they can test the effectiveness of aerosol therapy treatments.

From a healthy model to an ill one

There are various pathologies that can affect someone’s breathing, and air pollution tends to increase the likelihood of contracting one of them. For example, fibrosis damages the elastic fibers that help our lungs to expand when we breathe in. This makes the lung more rigid and breathing more difficult. In order to mimic this in the ex vivo lungs, the organs are heat treated with steam to stiffen the surface of the tissue. This then changes their elasticity, and recreates the mechanical behavior of human lungs with fibrosis.

Other illnesses such as cystic fibrosis occur, amongst other things, due to the lungs secreting substances that make it difficult for air to travel through the bronchi. To recreate this, the researchers insert artificial secretions made from thickening agents, which allows the model lung to mimic these breathing difficulties.

Future versions of the model

Imitating these illnesses is an important first step. But in order to study how aerosol therapy treatments work, the researchers also need to observe how they diffuse into the bloodstream. This can be either an advantage or a disadvantage. Since the lungs are also an entry point where the drug can spread through the body, the research team installed one last tool: a pump to simulate a heartbeat. “The pump allows fluid to circulate around the model in the same way that blood circulates in lungs,” explains Jérémie Pourchez. “During a test, we can then measure the amount of inhaled drug that will diffuse into the bloodstream. We are currently validating this improved model.”

One problem that the team is now facing is the development of new systemic inhalation treatments. These are designed to treat illnesses in other organs, but are inhaled and use the lungs as an entry point into the body. “A few years ago, an insulin spray was put on the market,” says Jérémie Pourchez. Insulin, which is used to treat diabetes, needs to be regularly injected. “This would be a relief for patients suffering from the disease, as it would replace these injections with inhalations. But the drug also requires an extremely precise dose of the active ingredient, and obtaining this dose of insulin using an aerosol remains a scientific and technical challenge.”

As well as being easier to use, an advantage of inhaling a treatment is how quickly the active ingredient enters into the bloodstream. “That’s why people who are trying to quit smoking find that electronic cigarettes work better than patches at satisfying nicotine cravings”, says the researcher. The dose of nicotine inhaled is deposited in the lungs and is diffused directly into the blood. “It also led me to study electronic cigarette-type devices and evaluate whether they can be used to deliver different drugs by aerosol,” explains Jérémie Pourchez.

By modifying certain technical aspects, these devices could become aerosol therapy tools, and be used in the future to treat certain lung diseases. This is one of the ongoing projects of the Saint-Étienne team. However, there is still substantial research that needs to be done into the device’s potential toxicity and its effectiveness depending on the inhaled drugs being tested. Rather than just being used as a tool for quitting smoking, the electronic cigarette could one day become the newest technology in medical aerosol treatment.

Finally, this model will also provide an answer to another important issue surrounding lung transplants. When an organ is donated, it is up to the biomedicine agency to decide whether this donation can be given to the transplant teams. But this urgent decision may be based on factors that are sometimes not sufficient enough to be able to assess the real quality of the organ. “For example, to assess the quality of a donor’s lung,” says Jérémie Pourchez, “we refer to important data such as: smoking, age, or the donor’s known illneses. Therefore, our experimental device which makes lungs breathe ex vivo, can be used as a tool to more accurately assess the quality of the lungs when they are received by the transplant team. Then, based on the tests performed on the organ that is going to be transplanted, we can determine whether it is safe to perform the operation.”

Photographie d'un implant osseux à base de phosphate de calcium conçus par l'équipe de David Marchat.

Bone implants to stimulate bone regeneration

/0 Comments/in , , /by

Mines Saint-Étienne’s Centre for Biomedical and Healthcare Engineering (CIS) seeks to improve healthcare through innovations in engineering. David Marchat, a materials researcher at CIS, is working on developing calcium phosphate-based biomaterials. Due to their ability to interact with living organisms, these bone implants can help regenerate bones.

 

This article is part of our dossier “When engineering helps improve healthcare“

New-generation bone implants is one of David Marchat’s areas of research. The Mines Saint-Étienne chemist is developing a calcium phosphate-based biomaterial which can interact with living organisms and stimulate bone generation. Although calcium phosphate-based bone substitutes have been used in health systems for decades, their action has remained limited.

“Our research focuses on the need for a bio-instructive implant, meaning one that is able to tell cells how to rebuild the bone and facilitate its vascularization.” In practice, this includes two major aspects: working on the chemical composition of calcium phosphate, and on the architecture of the implant. This is important, since existing implants are only able to regenerate small bone defects (less than 1 cm3).

When the bone is damaged too badly, it is not able to rebuild itself. A structural bone graft is then required, either from a donor (allograft), or from the patient (autograft). If the graft comes from a donor — primarily from bone banks — there is a risk that the remaining proteins cause an inflammatory response, infection or rejection. This is not the case if the graft is taken directly from another part of the patient’s body, but there is a limited quantity of harvestable bone structure. Furthermore, this involves two successive operations for the patient and the partial or complete loss of a bone.

Synthetic materials such as calcium phosphate-based biomaterials avoid these constraints. Moreover, since calcium phosphates form the mineral part of the bone, they are generally well-tolerated (i.e., non-toxic). The bone implants developed by the team meet a number of needs. The shape is personalized so as to correspond to the nature of the patient’s bone defect. This close contact between the bone margin and the implant therefore facilitates the migration of fluids, tissues and cells in the implant while also facilitating  their regeneration. The overall architecture, which ranges from the marco-scale (greater than 100 micrometers) to the nano-scale (less than 1 micrometer), is designed to “guide” this regeneration.

In addition, the composition of the calcium phosphate powder is optimized to provide chemical elements that contribute to bone formation. “We had to invent new tools and processes, in order to synthesize calcium-phosphate based powders, analyze them and then make customized bone implants,” explains David Marchat.

A new architecture

When  bone tissue is weakened a process is initiated to regenerate it, in which two types of cells act together. The first type of cells is responsible for breaking down the damaged tissue in order to recover elements that can be used by the second type of cells to rebuild the tissue. This second group of cells weave a fabric of collagen fibers – the extracellular matrix – which they then mineralize by precipitating calcium phosphate crystals. In order for the implant to effectively contribute to this bone remodeling process, the cells, blood vessels, and more generally, the new tissues, must be able to colonize it and eventually replace it.

The bone implants are modeled on several architectural levels, with pores of various sizes to promote bone regeneration and  vascular penetration. At the macro-scale level, the smallest pores  (less than 150 micrometers) confine the cells responsible for bone regeneration and stimulate their activity. Meanwhile, the largest pores (greater than 500 micrometers) allow for greater colonization by bone cells and blood vessels. “The combination of macropores of various sizes which make it possible to increase permeability and confinement,” explains David Marchat, “is essential to new bone regeneration strategies.”

To obtain the desired architecture, the researchers have developed a method based on pouring a calcium phosphate suspension — a liquid mixture containing calcium phosphate powder, water and  chemical stabilizers — into a wax mold, made using 3D printing. This “impregnated” mold is then dried and heat-treated at different temperatures in order to eliminate the wax mold and consolidate the implant.

“Another question that inevitably arises is,” adds the chemist, “for a given application, how long does the structure have to remain in the body so that the bone has time to regenerate itself,” adds the chemist. If the implant deteriorates too slowly, it will block bone formation. But if, on the other hand, it deteriorates too quickly, it will not be able to serve as a scaffold for bone formation. “It’s hard to estimate the right balance between the two.”

An ambitious engineering project for the testing phase

There are currently two ways to carry out biological assessments of these bone implants, which are used at different stages of testing. Standard in vitro cultures have the advantage of allowing for direct observation through a microscope but do not reproduce real conditions inside the body. In vivo experimentation with implantation inside an animal recreates these conditions more closely and provides what is referred to as a “physiological” environment, although animal and human physiology are different, and it is difficult to access the information. These testing stages are crucial, but researchers would like to move away from animal testing.

This is precisely the aim of an ambitious project: developing a 3D bioreactor with human cells to mimic human physiology. Such a structure would provide conditions equivalent to the human body and allow for direct observations, thereby making animal testing unnecessary for majority of validation phases for medical devices or medications. This project calls for expertise in fluid mechanics and a greater understanding of the human body and its workings. Other research topics in healthcare medical engineering have a similar aim. One such example are Organ-On-Chips, microfluidic chips that act as artificial cell cultures to simulate the inner workings and physiological activity of human organs.

nose

An “electronic nose” analyzes people’s breath to help sniff out diseases

/0 Comments/in , , /by

In partnership with IMT Atlantique, a team of researchers at IMT Lille Douai have developed a device which can measure the level of ammonia in someone’s breath. The aim of the artificial nose is to use this device to create a personalized follow-up care for patients affected by chronic kidney disease.  Eventually, the machine could even allow doctors to detect the disease in undiagnosed people.

 

This article is part of our dossier “When engineering helps improve healthcare“

In the human body, the kidneys’ main role is to remove toxins which are carried in the blood. However, when a person suffers from chronic kidney disease, this filtration function no longer works to the same standard. In France, the disease affects around 5.7 million people and can range from causing a degree of impairment, to terminal. Around 76,000 people are terminally affected in France, which means that their kidneys cannot filter their blood at all.

For these patients, the only options are to either wait for a transplant or face extensive treatment which hugely affects their daily life. It is therefore essential for doctors to be able to detect this silent and progressive disease early enough to slow down the effects. At the moment, doctors use blood or urine tests to identify the disease. But, to make it easier to diagnose, scientists are exploring another route: breath analysis. Studying the substances in the air we breathe out can provide valuable information about a person’s health.

Ammonia: a key element

In order to make this possible, two teams of researchers from IMT Lille Douai and IMT Atlantique have been working in partnership with the nephrology department at the Lille University Hospital. For the past three years, they have been developing a compact and turnkey device for doctors. The device is an ‘electronic nose’, a system made up of several sensors that can measure the specific concentration of a substance in someone’s breath.

In the case of chronic kidney disease, the substance being measured is ammonia. Ammonia is mainly produced by intestinal bacteria and is supposed to be filtered from the body by the kidneys. Previous studies have established a concentration threshold for levels of ammonia in a person’s breath, which doctors can then use to determine the likelihood that a patient has chronic kidney disease.

Ammonia offers some resistance

But how can you measure this compound using a portable device? The scientists used a series of sensors which react in the presence of ammonia, as Caroline Duc, team member and researcher at IMT Lille Douai, explains. “The sensors are made from two electrodes with a sensitive surface placed on top of them. The resistance of this surface varies depending on the amount of ammonia present”. When used, the device’s ability to resist electrical current increases when ammonia is present and returns to its initial state when the ammonia disappears. This therefore allows scientists to measure the level of the molecule in a patient’s breath.

Additionally, each sensor in the artificial nose has a unique composition. As Caroline Duc points out, “it is very complicated to use a material whose resistance varies when one type of gas is present”. This is why scientists decided to increase the accuracy of the analysis by combining several different sensors which have different responses to ammonia.

During the tests, the electronic nose was periodically exposed to a person’s breath. This resulted in an increase in the resistance when ammonia was present, and a decrease when the sensors were no longer exposed to the exhaled air.  Several factors were then measured and analyzed using statistical processing algorithms.

These algorithms rely on tools such as machine learning. The only difference with this case was that here, these tools were applied using small amounts of data and  supervised learning to categorize the different types of breath. In other words, the algorithms were taught using a dataset that had already classified breath which belongs to a healthy, ill or ‘uncertain’ individual. New profiles were then passed through this algorithm, so that they could be classed into these three categories.

Breath: a complex compound

To date, the first prototype developed is around 15cm in length and contains between 10 and 13 sensors.  The device was tested in a laboratory using artificial breath, which allowed the teams to verify that it could distinguish a healthy individual from someone who was ill, using the criteria defined in up-to-date scientific literature.  Then, experiments were carried out in clinics with patients suffering from chronic kidney disease. The idea was to measure the concentration of ammonia in their breath, before and after dialysis. The results demonstrated that there was a reduction in ammonia after the treatment.

However, they also highlighted the limitations of using a single marker. Before dialysis, some patients had levels of ammonia that were similar to those of a healthy person. Measuring the amount of ammonia in a person’s breath did not give a reliable diagnosis for chronic kidney disease.  This has caused scientists to launch a new study in order to identify other biomarkers characteristic of the disease.

More generally, this reflects the difficulty of conducting a reliable breath analysis, which can include both the air we breathe in as well as out. “Initially, I think clinical breath analysis will be developed for personalized patient follow-up care, rather than for diagnosis”, says Caroline Duc. For example, by measuring how a patient’s ammonia levels change in response to medication, doctors will be able to monitor the effectiveness of a treatment and then adapt it based on the results.

What is the future of follow-up care for other illnesses?

Researchers at IMT Lille Douai will continue to work on improving the electronic nose. At the moment, patients’ breath is collected and sealed in an airtight bag and is then analyzed in a laboratory. Consequently, the team’s aim is to develop a functional and completely autonomous prototype which would give doctors real-time results. However, this raises several new issues, such as the study of fluids, controlling the speed that the air is exhaled, etc. As well as this, to improve their data analysis, Caroline Duc and her colleagues have started a partnership with researchers who specialize in data handling at Télécom SudParis.

Moreover, the team is involved in a European-wide project which aims to identify biomarkers for lung cancer, and then create a multi-sensor system that is specifically designed to detect these substances. IMT Lille Douai’s expertise will be especially useful for this second objective.

This electronic nose, capable of sniffing out illnesses such as chronic kidney disease, is therefore still in the early stages of development, and needs a lot of work before it can really be used by doctors and their patients. But doctors are waiting with bated breath; in several years’ time it could be a breathtaking medical innovation!

Article written (in French) for I’MTech by Bastien Conteras

Mendeleev

Mendeleev: The history of a table

/0 Comments/in , /by

2019 marks the 150th anniversary of the periodic table of elements. To celebrate this anniversary, the Mines ParisTech Library and Mineralogy Museum have teamed up to create the exhibition Before Mendeleev: Genesis of a Table, on view until 31 January 2020. The exhibition traces the contributions of the scientists who preceded Mendeleev and led him to present the periodic table of elements, which has since served as a reference for all scientists and students.

 

To celebrate the 150th anniversary of the periodic table of elements, Mines ParisTech is presenting the exhibition Before Mendeleev: Genesis of a Table until 31 January 2020. Visitors have the opportunity to discover the scientists who contributed to formulating this classification and to developing knowledge over the years. Amélie Dessens is a curator at the library and head of Mines ParisTech’s heritage collections and Sarah Hijmans is a PhD student at the Université de Paris’ SPHère laboratory. They created the exhibition in collaboration with Didier Nectoux, curator at the Mineralogy Museum, to showcase and share the rich cultural collections of the school’s library and museum. “It’s this type of exhibition, along with school partnerships,” says Didier Nectoux, “that allows us to keep this heritage alive outside of the school.” He adds, “this rich heritage must be preserved and shared. And these collections are still essential today. The transformations of the 21st century are driving us to study new possibilities to find alternatives, and we need documentation, archives, in order to know which avenues have already been studied and abandoned, and the reasons why.”

The exhibition, which is presented in chronological order, starts on the doorstep of the Library with the beginnings of the study of elements: alchemy. “The alchemists were not just interested in turning lead into gold,” explains Amélie Dessens. “Beyond the esoteric sense with which alchemy is often related today, it was also – and more importantly – the beginning of chemistry and of identifying the elements that are presented here.” In display cases, eight minerals accompany the works. The first seven elements identified, and bismuth, the earliest written record of which dates from 1558 by the German scholar Georg Agricola. However, it was already well-known in European mining centers prior to this date. This also demonstrates the importance of accompanying discoveries with publication, which is crucial to situating knowledge in time.

A long road to developing the table

From Bergen to Lavoisier, Döbereiner to Newland, a series of display cases present the various steps of the advances, decisions and research that shaped the study and classification of the elements. First, there was Lavoisier, who brought about a true chemical revolution by introducing a scientific method to prove his theories, proposing the first classification of the “33 simple substances,” and working with Berthollet to develop a chemical nomenclature, which made it possible for everyone to use the same names for the elements. The second major turning point came in the 1860s, when scientists realized that elements could have similar chemical properties based on their atomic weight. They thus started to classify them based on these criteria and proposed potential classification formats, which are presented in the exhibition through diagrams, notes and publications.

For example, there was Alexandre-Émile Béguyer de Chancourtois, geologist, mineralogist and professor at the school of Mines de Paris, who made a significant contribution in 1862. He was the first to demonstrate the principle of periodicity through a spiral-shaped classification: the telluric screw. “Mendeleev was not the first to demonstrate periodicity, or to indicate where the missing elements should be placed in the table,” explains exhibition curator Amélie Dessens, “but unlike the others, he dared to predict the properties of the missing elements.” Dmitri Mendeleev published his table in 1869. When gallium was discovered in 1875, confirming his predictions, the news spread throughout the scientific community. It was at this point that Mendeleev’s classification would make its mark in history and earn its place in our textbooks.

applis mobiles, mobile apps

Do mobile apps for kids respect privacy rights?

/0 Comments/in , , , /by

The number of mobile applications for children is rapidly increasing. An entire market segment is taking shape to reach this target audience. Just like adults, the personal data issue applies to these younger audiences. Grazia Cecere, a researcher in the economics of privacy at Institut Mines-Télécom Business School, has studied the risk of infringing on children’s privacy rights. In this interview, she shares the findings from her research.

 

Why specifically study mobile applications for children?

Grazia Cecere: A report from the NGO Common Sense reveals that 98% of children under the age of 8 in the United States use a mobile device. They spend an average of 48 minutes per day on the device. That is huge, and digital stakeholders have understood this. They have developed a market specifically for kids. As a continuation of my research on the economics of privacy, I asked myself how the concept of personal data protection applied to this market. Several years ago, along with international researchers, I launched a project dedicated to these issues. The project was also launched thanks to funding from Vincent Lefrere’s thesis within the framework of the Futur & Ruptures program.

Do platforms consider children’s personal data differently than that of adults?

GC: First of all, children have a special status within the GDPR in Europe (General Data Protection Regulation). In the United States, specific legislation exists: COPPA (Children’s Online Privacy Protection Act). The FTC (Federal Trade Commission) handles all privacy issues related to users of digital services and pays close attention to children’s rights. As far as the platforms are concerned, Google Play and App Store both have Family and Children categories for children’s applications. Both Google and Apple have expressed their intention to separate these applications from those designed for adults or teens and ensure better privacy protection for the apps in these categories. In order for an app to be included in one of these categories, the developer must certify that it adheres to certain rules.

Is this really the case? Do apps in children’s categories respect privacy rights more than other applications?

GC: We conducted research to answer that question. We collected data from Google Play on over 10,000 mobile applications for children, both within and outside the category. Some apps choose not to certify and instead use keywords to target children. We check if the app collects telephone numbers, location, usage data, and whether they access other information on the telephone. We then compare the different apps. Our results showed that, on average, the applications in the children’s category collect fewer personal data and respect users’ privacy more than those targeting the same audience outside the category. We can therefore conclude that, on average, the platforms’ categories specifically dedicated to children reduce the collection of data. On the other hand, our study also showed that a substantial portion of the apps in these categories collect sensitive data.

Do all developers play by the rules when it comes to protecting children’s personal data?

GC: App markets ask developers to provide their location. Based on this geographical data, we searched to see whether an application’s country of origin influenced its degree of respect for users’ privacy. We demonstrated that if the developer is located in a country with strong personal data regulations—such as the EU, the United States and Canada—it generally respects user privacy more than a developer based in a country with weak regulation. In addition, developers who choose not to provide their location are generally those who collect more sensitive data.

Are these results surprising?

GC: In a sense, yes, because we expected the app market to play a role in respecting personal data. These results raise the question of the extra-territorial scope of the GDPR, for example. In theory, whether an application is developed in France or in India, if it is marked in Europe, it must respect the GDPR. However, our results show that among countries with a weak regulation, the weight of the legislation in the destination market is not enough to change the developers’ local practices. I must emphasize that offering an app to all countries is extremely easy—it is even encouraged by the platforms, which makes it even more important to pay special attention to this issue.

What does this mean for children’s privacy rights?

GC: The developers are the owners of the data. Once personal data is collected by the app, it is sent to the developer’s servers, generally in the country where they are located. The fact that foreign developers pay less attention to protecting users’ privacy means that the processing of this data is probably also less respectful of this principle.

 

robots

Robots on their best behavior in the factory of the future

/0 Comments/in , , /by

A shorter version of this article was published in the monthly magazine Acteurs du franco-allemand, as part of an editorial partnership.

[divider style=”normal” top=”20″ bottom=”20″]

Robots must learn to communicate better if they want to earn their spot in the factory of the future. This will be a necessary step in ensuring the autonomy and flexibility of production systems. This issue is the focus of the German-French Academy for the Industry of the Future’s SCHEIF project. More specifically, researchers must choose appropriate forms of communication technology and determine how to best organize the transmission of information in a complex environment.

 

“The industry system is monolithic for robots. They are static, and specialized for a single task, but it is impossible for us to change their specialization based on the environment.” This observation was the starting point for the SCHEIF[1] project. SCHEIF, conducted in the framework of the German-French Academy for the Industry of the Future, seeks to allow robots to adapt more easily to function changes. To achieve this, the project brings together researchers from EURECOM, the Technical University of Munich (TUM) and IMT Atlantique. The researchers’ goal is “to create a ‘plug and play’ robot that can be deployed anywhere, easily understand its environment, and quickly interact with humans and other robots,” explains Jérôme Härri, a communications researcher with EURECOM participating in this project.

The robots’ communication capacities are particularly critical in achieving this goal. In order to adapt, they must be able to effectively obtain information.  The machines must also be able to communicate their actions to other agents—both humans and robots—in order to integrate into their environment without disruption. Without these aspects, there can be no coordination and therefore no flexibility.

This is precisely one of the major challenges of the SCHEIF project, since the industrial environment imposes numerous constraints on machine communications. They must be fast in the event of an emergency, and flexible enough to prioritize information based on its importance for production chain safety and effectiveness. They must also be reliable, given the sensitivity of the information transmitted. The machines must also be able to communicate over the distances of large factories, not just a few meters. They must combine speed, transmission range, adaptability and security.

Solving the technology puzzle

“The solution cannot be found in a single technology,” Jérôme Härri emphasizes. Sensor technology, for example, like Sigfox and LoRa, which are dedicated to connected objects, have high reliability and a long range, but cannot directly communicate with each other. “There must be a supervisor in charge of the interface, but if it breaks down, it becomes problematic, and this affects the robustness criterion for the communications,” the researcher adds. “Furthermore, this data generally returns to the operator of the network base stations, and the industrialist must subscribe to a service in order to obtain it.”

On the other hand, 4G provides the reliability and range, but not necessarily the speed and adaptability needed for the industry of the future. As for 5G, it provides the required speed and offers the possibility of proprietary systems. This would free industrialists from the need to go through an operator. However, its reliability in an industrial context is still under specification.

Faced with this puzzle, two main approaches emerge. The first is based on increasing the interoperability and speed of sensor technology. The second is based on expanding 5G to meet industrial needs, particularly by providing it with features similar to those of sensor technologies.  The researchers chose this second option. “We are improving 5G protocols by examining how to allocate the network’s resources in order to increase reliability and flexibility,” says Jérôme Härri.

To achieve this, the teams of French and German researchers can draw on extensive experience in vehicular communication, which uses 4G and 5G networks to solve transport and mobility issues. The cellular technology used for vehicles has the advantage of featuring a cooperative scheduling specification. This information system feature decides who should communicate a message and at what time. A cooperative scheduler is essential for fleets of vehicles on a highway, just like fleets of robots used in a factory. It ensures that all robots follow the same rules of priority. For example, thanks to the scheduler, information that is urgent for one robot is also urgent for the others, and all the machines can react to free the network from traffic and prioritize this information. “One of our current tasks is to develop a cooperative scheduler for 5G adapted to robots in an industrial context,” explains Jérôme Härri.

Deep learning for added flexibility

Although the machines can rely on a scheduler to know when to communicate, they still must know which rules to follow. The goal of the scheduler is to bring order to the network, to prevent network saturation, for example, and collisions between data packets. However, it cannot determine whether or not to authorize a communication solely by taking communication channel load into account. This approach would mean blindly communicating information: a message would be sent when space is available, without any knowledge of what the other robots will do. Yet in critical networks, the goal is to plan for the medium term in order to guarantee reliability and reaction times. When robots move, the environment changes. It must therefore be possible to predict whether all the robots will start suddenly communicating in a few seconds, or if there will be very few messages.

Deep learning is the tool of choice for teaching networks and machines how to anticipate these types of circumstances. “We let them learn how several moving objects communicate by using mobility datasets. They will then be able to recognize similar situations in their actual use and will know the consequences that can arise in terms of channel quality, or number of messages sent,” the researcher explains. “It is sometimes difficult to ensure learning datasets will match the actual situations the network will face in the future. We must therefore add additional learning on the fly during use. Each decision taken is analyzed. System decisions therefore improve over time.”

The initial results on this use of deep learning to optimize the network have been published by the teams from EURECOM and Technical University of Munich. The researchers have succeeded in organizing communication between autonomous mobile agents in order to prevent the collision of the transmitted data packets. “More importantly, we were able to accomplish this without each robot being notified of whether the others would communicate,” Jérôme Härri adds. “We succeeded in allowing one agent to anticipate when the others will communicate based solely on behavior that preceded communication in the past.”

The researchers intend to pursue their efforts by increasing the complexity of their experiments to make them more like actual situations that occur in industrial contexts. The more agents, the more the behavior becomes erratic and difficult to predict. The challenge is therefore to enable cooperative learning. This would be a further step towards fully autonomous industrial environments.

[1] SCHEIF is an acronym for Smart Cyber-physical Environments for Industry of the Future.