Results Industrial Fellowship 2021-2 Call

The FNR is pleased to communicate that 6 of 9 Industrial Fellowship projects have been selected for funding in the 2021-2 Call, representing an FNR commitment of 1.27 MEUR. 

The aim of the Industrial Fellowships programme is to foster the cooperation between Luxembourg based companies active in R&D and public research institutions in Luxembourg and/or abroad. The scheme awards PhD and Postdoc grants to researchers who carry out their PhD and/or postdoc training in collaboration with a company in Luxembourg. The scheme is open to all scientific domains, and do all researchers, regardless of their nationality. Collaborating companies must have a presence in Luxembourg.

Go to Industrial Fellowships programme page

Funded Industrial Fellowships projects

PhD – Life Sciences, Biology & Medicine

Applicant

Hélène De Franco

Project title

Personalized Food Response Models (PERFORM)

Host institution

Luxembourg Institute of Health (LIH)

Collaborating company

Abstract

The human gut contains trillions of microorganisms, the majority of which are commensal bacteria. These bacteria feed on fibers from our food and produce molecules that are useful to us, which makes fiber consumption beneficial to health. Today most of us are deficient in the intake of dietary fibers, which correlates with a multiplication of non-communicable digestive tract-related diseases (such as colon cancer, diabetes and inflammatory bowel disease), suggesting a link between nutrition, the microbiota and health. As a matter of fact, studies suggest that fibers are such an important source of energy for gut bacteria, that if they aren’t provided with adequate sustenance, these bacteria can turn against their human host. Since bacterial composition of the gut microbiota is specific to each individual and several different types of fibers exist, it makes sense to us to investigate what effect does which fiber have on which specific bacteria, and what repercussion to expect on the human host. This led to the development of PERFORM, a project carried out by the Luxembourg Institute of Health and Nium, to better understand individualized responses to foods. To do so, we will analyze data generated by the ongoing LUXFICO study, in which forty people will each be fed successively diets high and low in fibers. We will create computerized models for each participant, matching their physiological characteristics, and for each diet. First, we will monitor every change in participants health status and gut microbiota composition throughout the dietary interventions. This will allow us to better understand the impact of the presence or absence of specific dietary components on health. Then, incorporating these responses into the participants computerized alter ego will enable us to design new tools to predict people’s very own response to foods, based on their personal characteristics. This project will contribute to the development of an innovative approach to personalized nutrition by first aiming to expand our understanding of the effects of the gut microbiota on health, but also by the development of a commercial application, making use of this newly acquired scientific knowledge to benefit society.

Applicant

Svonko Galasso

Project title

Machine LEarning-based Marker-lEss gait analysis system for clinical assessment of humaN moTiOn (MEMENTO)

Host institution

LUNEX University

Collaborating institution

Rehazenter – Centre Nat. de Rééducation Fonctionnelle; University of Cassino and Southern Lazio

Gait Analysis (GAn) is a standard diagnostic laboratory procedure to quantitatively assess and analyse the human body motion. GAn has a crucial function in the healthcare system to provide timely interventions and keep track of recovery in rehabilitation. GAn is used to assess quality of life, health status, and physical function. The motion data acquired by the GAn can also be applied to assess risk of dementia, risk of falling, and even risk of early mortality. However, for the healthcare professionals involved in treatment and rehabilitation (physicians/doctors, physiotherapists, nurses, care givers) GAn in clinical practice still remains a subjective process of visual examination…why? Because current commercially available solutions for gait analysis (infra-red cameras, inertial sensors, sensorised platforms, etc.) do not meet the needs of the healthcare professionals. The healthcare professionals have very limited access to expensive equipment and a lack of time for data collection and analysis of the numerous parameters extracted from the motion data. The MEMENTO project outcomes will contemplate the development of an innovative software tool able to provide gait analysis and an automatic learning process of the gait data via Machine Learning approaches. The motion data will be extracted from processing gait images, acquired via a commercial camera system. The MEMENTO novel GAn system will provide a solution for the actual unmet need of clinicians in obtaining data to help clinical decision making with fast setting and simple to use solution.

PhD – Materials, Physics & Engineering

Applicant

Florent Mauret

Project title

An Advanced Simulation Model for Transition Towards Enhanced Blast Furnaces with High Hydrogen Load and Low CO2 Emissions (BBFH2)

Host company

Paul Wurth

Collaborating company

Åbo Akademi University – Process and Systems Engineering

Abstract

The steelmaking industry is one of the largest CO2 emitters, accounting for around 7% of the global human-induced CO2 emissions. With the growing concern of global warming, the industry is coming under increased pressure to reduce its environmental impact. The Blast furnace (BF), used to reduce iron ore into liquid metallic iron, is emitting the majority of the CO2 from the whole steelmaking industry. The BF is therefore highly requested to minimize its CO2 emissions and transition towards the so-called “Enhanced BF”. Paul Wurth S.A.’s research project aims to develop an advanced BF simulation solution to evaluate a variety of CO2 emission mitigation technologies. Due to the extreme conditions inside a BF, experiments are challenging. Simulation is therefore the most promising way to improve the Blast Furnace process due to its ability to rapidly and safely test a wide range of operational parameters. Coke, a derivate of coal, is the main reducing agent of the BF. However, the chemical reactions transforming iron ore into liquid metallic iron using coke are emitting large quantities of CO2. A promising way to mitigate the BF environmental impact is to replace coke with hydrogen in the furnace since hydrogen used as a reducing agent only produces water as a by-product. However, the impact of H2 and H2O on the furnace limits the extent to which hydrogen can replace coke. Using the simulation model developed in this project, a wide range of operational parameters will be tested in order to find a feasible BF process yielding minimum CO2 emissions in case of a high hydrogen load. Furthermore, the simulation tool developed will be influential to improve the BF by supporting many other innovative projects, thus further helping with the reduction of CO2 emissions. In particular, the possibility to use Artificial Intelligence (AI) in a BF will be greatly enhanced. In fact, the fast model developed could be used to generate the tremendous amount of data needed for AI applications. Considering all the objectives of the project, it is clear that the developed model will be crucial to enable the steel industry to reduce its global carbon footprint in the world.

Applicant

Mohammadhossein Norouzian

Project title

Developing and Online Monitoring of Laser Welding Between Hard metal and Steel Based on Artificial Neural Network Feedback (Bright)

Host institution

University of Luxembourg

Collaborating company

CERATIZIT Luxembourg S.a.r.l

Abstract

Industry constantly demands brand spanking new, eco-friendly, revolutionary design, and low cost but high quality for a non-stop development system. It is common for cutting tool manufacturers to contend with this principle that they need to evolve into innovative technologies to develop and control the quality of their products to overcome the competition in the market. Advanced design and accurate quality control of cutting tools have always been crucial for manufacturers due to their rapid development and intense competition in the industry market. Hence, the role of material science in understanding the properties of materials and utilizing this knowledge in complex production strategies for manufacturing these products is undeniable. Welding, which is directly related to materials science and manufacturing engineering, is an essential part of production strategies that have progressed rapidly and nowadays are used in producing tools. However, the welding of tools has complexities and difficulties due to using different materials. Dissimilar welding of Hard Metal (HM) as the active part and Steel (St) as the carrier or base of cutting tools is the most challenging part of the production, mainly due to their mismatched thermal expansion coefficient. They require developed and intelligent welding systems to work with few resources and create precise quality control without mistake to promise high-quality products. Project BRIGHT seeks to provide Ceratizit Luxembourg S.a.r.l as the industrial partner, with production technology know-how related to welding challenging-to-join material combinations like HM to St. The challenge originates from uneven cooling shrinking behavior of sintered cemented carbides as the HM that creates heat cracks which deteriorate mechanical strength of the joints. This project plans to build a defined spatial and temporal temperature profile using Vertical Cavity Laser (VCSEL), enabling crack-free, thus more stable joints. In addition, online monitoring of the welding process by signals of sensors based on Artificial Neural Networks (ANN) feedback provides opportunities to predict the joint quality, reducing the time and costs of offline analysis.

PhD – Domain ICT & Space

Applicant

Maxime Gautier Louis Hubert Delisle

Project title

Design of a Capturing, Absorbing, Securing system for active space Debris removal (CASED)

Host institution

University of Luxembourg

Collaborating company

Spacety

Abstract

After more than 60 years of space activities, the number of in-orbit objects hardly decreased, especially in Low-Earth Orbit (LEO). The statistics of ESA shows an amount of 330 million debris objects not bigger than 1 cm. The recent mega-constellations (for example with Starlink of SpaceX) and Earth data collection fleets, makes rocket launches more frequent, setting many more satellites in orbit. The Kessler syndrome stipulates that the number of objects in orbit will grow exponentially over the years. As a result, LEO is on the way to become an unsafe, crowdy orbit, as collisions will occur more often. One of the solution to tackle this issue, is to use Active Debris Removal (ADR) systems. The state-of-the-art (SotA) on capturing systems for ADR is mostly focusing on specific technologies such as harpoons or tethered nets, targeting one specific cooperative debris at a time. Assessing these new technologies can require substantial facilities, not often accessible. Because of that, the majority of them reach a Technology Readiness Level (TRL) of 4 (TRL-4). To that extent, CASED proposes to design and develop a foldable CubeSat-application capturing mechanism for a defined range of uncooperative targets. A first ‘soft capture’, followed by a ‘hard capture’, will ensure that the debris is securely attached to the servicer, without dragging it in their residual movement, thanks to the flexibility of the system. In addition, an adhesive material will help the debris not to slip away during the ‘soft capture’ phase. Within this collaboration between Spacety Luxembourg and SnT, the design, the prototyping and the verification & validation of the novel SotA capturing system for ADR, CASED, will be conducted, and assessed in SnT experimental facilities for high-fidelity simulations of orbital scenarios for ADR.

Applicant

Nesryne Mejri

Project title

UNsupervised multi-type explainable deepFAKE detection (UNFAKE)

Host institution

University of Luxembourg

Industry partner

POST Luxembourg

Abstract

The recent emergence of artificial intelligence (AI) generated visual content, known as deepfakes, has given rise to substantial concerns about the potential misuse of this technology. These AI techniques allow the generation of various types of facial manipulations, some of which are nowadays normalised due to social media platforms. As a result,
deepfakes can have several creative applications in entertainment, education and movie post-production. Nevertheless, their abuse can cascade from fraud, impersonation and psychological harm at the individual level to the erosion of trust in digital media at the societal level. Within this context, many efforts have been put in detecting forged videos, for example, to detect whether someone’s face has been replaced. However, these approaches remain limited as they only identify one type of deepfake at a time. Therefore, the goal of UNFAKE is to develop a novel AI-based deepfake detection solution capable of producing an explainable prediction regardless of the encountered facial manipulation type. The proposed solution will benefit from recent computer vision and AI advances and will only require using non-manipulated videos.

Furthermore, UNFAKE also aims to develop a solution that complies with real-life constraints such as portability across devices with different specifications. All of the research and developments conducted within this project will be done in close collaboration with POST, based in Luxembourg.

RELATED NEWS

Results 2021-2 BRIDGES Call

Results 2021-1 BRIDGES Call

Results 2021-1 Industrial Fellowships Call

This site uses cookies. By continuing to use this site, you agree to the use of cookies for analytics purposes. Find out more in our Privacy Statement