Projets novateurs réalisés

International Validation of a Cognitive Screening Tool for Pilot Risk of Accidents in Flight

General Aviation (GA) contributes billions in GDP, and supports hundreds of thousands of jobs in Canada and the United Kingdom (UK). Sustaining the industry requires experienced pilots flying safely for as long as possible. To preserve pilot engagement in GA, Carleton University’s Advanced Cognitive Engineering Laboratory has developed the CANFLY, a cognitive health screening tool for GA pilots to monitor changes in cognition associated with increased accident risk. This project expands the impact of CANFLY, by adapting and validating it for use in the UK. CANFLY will be distributed to a sample of UK GA pilots (ideally 35), to ensure the tool reflects local operations, then, adapted based on pilot feedback, and finally, distributed to a larger sample of GA UK pilots (60 for planned analyses). This collaboration strengthens CANFLY’s international application, and creates the foundation for future joint projects in GA between Canada and the UK. Through the collaboration, Carleton University will expand the reach of CANFLY beyond Canada, while Imperial College will gain access to CANFLY, assisting them to engineer a safer GA industry. The project allows Carleton and Imperial to work in tandem to achieve their research goals, with an impact that will long outlive the collaboration.

Establishing an EEG-based protocol, for in-clinic BCI, in complex behavioural needs children diagnosed with ASD

An electroencephalogram (EEG) is a method to measure brain activity. It can be used to detect atypical brain development. Research has shown that EEG can be used as a clinical tool to detect early signs of Autism Spectrum Disorder (ASD) (Bosl et al., 2018). However, measuring EEG in children diagnosed with ASD turns out to be difficult. This is due to children that are diagnosed with ASD experience anxiety or distress, agitation, and sensory sensitivities towards the EEG. This might be related to the cap being used, or, more generally, to the entire process, which is new and perhaps frightening. This limits the use of EEG to measure the brain activity of children diagnosed with ASD. Consequently, this limits the discovery of the underlying mechanisms of ASD. By untangling the process of ASD development, researchers might find a way to prevent it, solve it, or find ways to help the symptoms.
Therefore, since EEG seems a promising measure, it is important for children diagnosed with ASD to tolerate EEG measurements. This current project aims at establishing an EEG-based protocol in which children diagnosed with ASD become desensitised and will tolerate EEG more easily.

Wintersport Female Athlete Health

The main objective of this research project is to investigate the prevalence and impact of menstrual cycle-related symptoms and low energy availability-related health issues among competitive female athletes, and to examine how these factors impact training, competition, and the use of pain medications. This project contributes to a larger research initiative led by the host supervisor’s research group, which focuses on female athlete health, menstrual cycle-related performance consideration, and the early identification of the Female Athlete Triad and Relative Energy Deficiency in Sport. The project promotes ongoing international collaboration and knowledge exchange between Canadian and Austrian sport-science communities, as it will combine the University of Alberta’s expertise in female athlete and respiratory health with the University of Innsbruck’s established strength in environmental stressors and athlete health.

Approche hybride hazard-prédiction des chutes de performances des réseaux de pompage

Les réseaux de pompage sont essentiels pour la distribution d’eau, mais ils subissent des problèmes comme cavitation, fuites et obstructions, entraînant des coûts élevés et une baisse d’efficacité. La cavitation, causée par la formation et l’implosion de bulles de vapeur, endommage les pompes et réduit leur rendement. Les méthodes traditionnelles de détection, basées sur des essais ou l’écoute acoustique, sont limitées par leur coût et leur incapacité à traiter des données en temps réel. L’essor des technologies numériques et de l’IA a changé la donne : la simulation CFD et les modèles avancés prédisent la cavitation, tandis que des techniques de apprentissage automatique améliorent la détection des fuites. L’utilisation de capteurs acoustiques intelligents permet désormais une surveillance continue et une maintenance prédictive. Notre projet « Approche hybride Hazard » combine simulation numérique et IA pour anticiper cavitation, blocages et fuites, réduisant les coûts et prolongeant la durée de vie des équipements.

Légitimité et acceptabilité des initiatives plurilatérales à l’Organisation mondiale du commerce

En raison de profondes divergences d’intérêts entre ses 166 membres et d’un système institutionnel peu flexible reposant sur le consensus, l’Organisation mondiale du commerce n’a pas su apporter de réponses concrètes aux nouveaux enjeux de la mondialisation économique. Face à cette impasse, de nombreux pays ont choisi de signer des accords plurilatéraux, conclus entre un groupe volontaire de membres de l’OMC, offrant une alternative plus pragmatique que les négociations multilatérales. Or, ces initiatives plurilatérales suscitent des contestations de la part de certains pays qui doutent de leur légalité ou qui remettent en cause leur légitimité.

Le projet de recherche vise précisément à éclairer ces controverses. Dans un premier temps, il s’agira d’analyser la légalité des initiatives plurilatérales et des accords qui en découlent en mobilisant les sources pertinentes du droit de l’OMC, le droit international coutumier et les principes généraux du droit international public. Dans un second temps, l’étude s’attachera à la question de la légitimité. Cette partie de l’analyse portera sur les déclarations officielles, les communications des États et des organisations, ainsi que sur des entretiens avec les négociateurs.

In-situ Carbon Dioxide Reduction Reaction in Bipolar Membrane Electrochemical Systems

Carbon dioxide (CO2) in air is commonly captured using alkaline electrolytes such as potassium hydroxide (KOH), which react with CO2 to form potassium bicarbonate (KHCO3). To convert the captured CO2 into useful chemical feedstocks or fuels, the absorbed solution is typically heated to regenerate pure CO2, which is then converted through chemical or electrochemical processes such as the CO2 reduction reaction (CO2RR). These thermal regeneration and separation steps are among the most energy-intensive and costly parts of the overall process. In this project, we aim to bypass these steps by directly converting KHCO3 into value-added chemicals, such as ethylene. This is achieved using a bipolar membrane (BPM), which supplies protons (H+) to locally acidify the electrolyte, thereby releasing CO2 in-situ from KHCO3. The generated CO2 is immediately converted to ethylene on the catalyst surface. The objective of this project is to enhance catalyst efficiency and optimize operating conditions to maximize ethylene production.

Experimental Validation: SINDy-Based Dynamic Identification of GFM/GFL Converters from PMU Data

This project develops a data-driven method to identify the internal dynamics of grid-forming (GFM) and grid-following (GFL) converters in renewable-rich power systems using only output-side PMU measurements. Small, continuous perturbations (e.g., 0.01 p.u. in active/reactive power setpoints) are injected during normal operation to safely excite the system, and SINDy is applied to the recorded voltage, current, and frequency data to obtain reduced-order models that capture the converters’ voltage–frequency behavior. From these models, we estimate key control parameters, such as PLL PI gains, P-f and Q-V droop coefficients, and virtual inertia and damping, and then design robust controllers that improve stability under high RES penetration. The full workflow has been implemented and validated in Simulink at McGill University, and the next phase will experimentally validate the approach using the double Power Hardware-in-the-Loop setup at Karlsruhe Institute of Technology (KIT). The project strengthens McGill’s capability in data-driven converter modeling and control and leverages KIT’s advanced laboratory infrastructure, fostering a long-term research collaboration between the two institutions.

An ensemble machine learning framework for streamflow data reconstruction

This project will develop a new framework to reconstruct missing streamflow data using advanced machine learning techniques. Reliable streamflow records are essential for flood forecasting, drought monitoring, and water resource planning, but many stations have missing or incomplete data. The proposed approach will combine traditional statistical methods with modern single-learner and ensemble machine learning models to estimate missing values more accurately across North American river basins. The collaboration between the University of Saskatchewan and UNAM will strengthen expertise in statistical hydrology and deliver practical tools that contribute to efforts toward improving water management and water security for both countries.

Understanding the effects of Early-Life Adversity on Microglia in the Developing Brain

Early-life adversity (ELA) is one of the most consistent and robust predictors of poor mental health outcomes across the lifespan. These are developmental periods characterized by high neural plasticity, in which the brain is shaped by experience. Thus, ELA can provoke detrimental, long-lasting changes in the brain and it is critical to understand the mechanisms of how ELA changes the developing brain that remain unclear. Thus, we propose to use the well-established Limited Bedding and Nesting (LBN) model, to perform bulk-RNA seq in the hypothalamus; a stress-sensitive brain region. Previous literature showed that ELA-caused developmental changes in the brain are mediated by microglia, the immune cells of the brain. Therefore, we will isolate microglia from the paraventricular nucleus of the hypothalamus of P8 mice from ELA or control conditions, to look for changes in gene expression that could provide potential therapeutic targets. This collaboration will capitalize in Dr. Ciernia’s expertise in RNA-seq and bioinformatics, as well as Dr. Bolton’s expertise in ELA and brain development to produce data that will be used for future grant writing to bring funding for future research.

System-of-Systems Solution with Quantum Simulation for Monitoring, Reporting, Validation, and Verification of GHGs/Carbon Cycle in Biodiversity Ecosystems for SEPLS to Strengthen Sustainable Frameworks Development

Title :System-of-Systems Solution with Quantum Simulation for Monitoring, Reporting, Validation, and Verification of GHGs/Carbon Cycle in Biodiversity Ecosystems for SEPLS to Strengthen Sustainable

Overview:
This project aims to develop a System-of-Systems solution using quantum simulation to monitor, report, validate, and verify greenhouse gases (CO2, CH4, N2O) and the carbon cycle in biodiversity ecosystems, particularly in SEPLS (Socio-Ecological Production Landscapes and Seascapes). By integrating satellite data, field sampling, AI, and quantum-based spectral analysis, the project will create accurate regional carbon-sequestration models, maps of soil and water carbon, and identify carbon-sink hotspots. Participating institutions will benefit by gaining access to cutting-edge tools and data for ecosystem carbon monitoring, enhancing research capacity, and strengthening collaborations for sustainable ecosystem management and climate action.

Academic supervisor: Cynthia Goh and Yen-Hsun Su
Intern:Ke-Hsin Chen

Mitigation des impacts opérationnels des erreurs de contextualisation dans une scène routière

E-Smart commercialise une solution innovante qui offre aux gestionnaires un moyen automatisé d’assurer la conformité aux limites de vitesse, sans qu’ils aient à surveiller les camionneurs. Une interface placée à la commande d’accélération contrôle activement la vitesse du camion en temps réel, en se basant sur la cartographie des limites de vitesse. Toutefois, lorsque la localisation GPS est imprécise ou que les vitesses sont absentes de la carte ou erronées, le système produit des erreurs. Ceci entraine un manque de confiance envers l’outil. En effet, les systèmes de vision artificielle basés sur la détection d’objets, la classification et la segmentation sémantique monoculaires, ne permettent pas de comprendre ce qui se passe dans une scène, comme les positions relatives d’un objet par rapport à l’autre ou leur orientation respective [1] [2]. Ces limitations découlent des défis inhérents à l’estimation de la profondeur et à la compréhension de la scène avec des images monoculaires [3]. Ces systèmes sont encore moins capables de comprendre la sémantique (signification des relations entre le sujet et les objets) d’une scène. Donc pour améliorer le produit et offrir du coup l’ensemble des fonctionnalités de vidéo-télématique classique aux gestionnaires de flotte, E-Smart souhaite développer une nouvelle génération de produit avec une caméra et des réseaux de neurones profonds. Ce système devra faire la détection d’objets et de conditions sur le réseau routier (ex. piétons et cyclistes, neige), afin de bonifier les informations disponibles, en plus d’accroître la précision globale de la solution.

Agrilo and Agrilo-VR: Field Validation and Real-World Deployment of AI-Driven Soil and Nutrient Sensing Systems in South Africa

Healthy soil is crucial for global food security, yet farmers face declining fertility, nutrient depletion, and rising fertilizer costs. These challenges are particularly pressing for smallholder farmers in South Africa, where timely soil information is limited. This project connects researchers from Canada and South Africa to advance Agrilo, a real-time soil nutrient sensing system, and Agrilo-VR, an immersive virtual reality platform for environmental education.
Agrilo’s portable, low-cost sensors measure essential soil indicators—including nitrate, phosphate, potassium, pH, boron, sulfur, CEC, NOM, magnesium, manganese, iron, and calcium—with near-laboratory accuracy. These data help farmers optimize fertilizer application, reduce nutrient runoff, and improve crop performance. Field validation will take place across South Africa’s diverse agricultural landscapes, allowing for assessment under real farming conditions.
A student intern will travel to South Africa to support field validation, collect soil datasets, and assist with the deployment and testing of Agrilo-VR in schools, training centres, and conservation programs. Agrilo-VR will transform real soil and ecological data into interactive learning environments that make environmental science accessible and engaging.
By integrating sensing technology, AI-driven analytics, and VR-based education, this project promotes climate-smart agriculture, strengthens international collaboration, and expands access to sustainable soil management tools across both partner regions.