Projets novateurs réalisés

«Shallow Water Extensions to Complex Fluid Dynamics»

The realism and interactivity of large water bodies—such as oceans and rivers—are essential for creating immersive and dynamic open-world environments. With the rapid advancement in GPU computing power, physical simulation techniques are now a feasible solution for significantly enhancing the realism of water systems in real-time applications. Shallow water simulations, for example, provide efficient 2.5D methods for modeling water flow propagation in both a believable and computationally cost-effective manner. However, these techniques are primarily limited to the simulation of shallow fluid heightmaps. This project seeks to build on existing research to enhance the visual fidelity of such simulations, focusing on creating more engaging and lifelike water effects for interactive applications such as video games and simulators. We will explore advanced effects, including crashing waves, foam, splashes, and more realistic solid/fluid interactions. Through a comprehensive review of both real-time and offline graphics literature, we will identify innovative techniques to improve the visual quality and performance of these simulations. We will then implement, refine, and extend these approaches, pushing the boundaries of what is currently possible for interactive water simulation.

Optimizing STAR-RIS Deployment for Next-Generation Wireless Networks

This research project focuses on improving wireless communication systems by studying and optimizing new technologies called Reconfigurable Intelligent Surfaces (RIS). These surfaces can control how wireless signals travel, helping expand coverage and improve connection quality in urban and rural areas. A special type of RIS, known as STAR-RIS (Simultaneously Transmitting and Reflecting), allows signals to be sent in all directions at once, which can greatly improve network performance.
The intern will work closely with LATYS Intelligence Inc. to develop models and simulations that show how to best place and configure these smart surfaces in real-world environments. The goal is to make wireless networks—such as 5G and future 6G systems—more reliable, energy-efficient, and cost-effective. LATYS will use the results to improve their products and bring better wireless solutions to the market. This project also supports the training of a highly skilled graduate student, helping to build Canada’s future in advanced wireless technologies.

Training Generative Tabular Foundation Models

Layer 6 is a machine learning research and engineering company owned by The Toronto-Dominion Bank. Most of TD’s transactional data is stored in relational database management systems in the form of tabular datasets, structured as rows and columns. While Large Language Models (LLMs) have revolutionized the handling of unstructured data (such as text, audio, and images) they still lag behind in processing tabular data. Recently, Layer 6 developed a novel tabular foundation model called TabDPT. To advance research in this emerging area, Layer 6 is collaborating with the team at Polytechnique Montréal to explore some of their ideas aimed at enhancing both the training processes and understanding of performance characteristics of these models. Models like TabDPT are designed to enable the reuse of pre-trained models across projects and use in-context learning, offering both potential economic advantages and gains in accuracy over traditional methods. These improvements can translate into performance boosts in critical downstream tasks containing regression and prediction for applications such as fraud detection.

Investigating the finite temperature behaviour of polarons through the Momentum Average method and the Generalized Green’s function Cluster Explansion method.

The description of solid state systems form an intractable many-body problem as every particle is pushed and pulled due to Coulomb’s law by the other particles. It is almost impossible to directly describe every particle in a macroscopic system (with 10^24 particles). A clever solution involves simplifying the many-body problem into a problem of hypothetical non-interacting or weakly interacting ‘quasiparticles’ .

One such quasiparticle is the phonon, which represents the quantised vibrations of the lattices sites. When electrons propagate through the lattice, it interacts with phonons. The composite system (i.e the electron dressed by a cloud of phonons) has particle-like properties and are known as ‘polarons’. This leads to important consequences in electrical and thermodynamic properties of materials. However, describing the behaviour of polarons has remained one of the most challenging problems in quantum matter. Furthermore, at non-zero temperatures, there is an arbitrary number of thermal phonons in the system which further complicates the study of polarons.

Using a method known as the MA approximation and GGCE, we seek to advance our present understanding of polarons at finite temperature. Outcomes from this study have broad ranging consequences from understanding decoherence in solid-state quantum computing hardware to engineering the materials of tomorrow.

Structural Barriers Faced by Women Entrepreneurs Impacting Their Ability to Start and Scale up Their Businesses, Identifying solutions and Best Practice Models

This project explores the challenges women entrepreneurs face when starting and growing their businesses, focusing on structural barriers such as access to funding, networking opportunities, and business support. By identifying practical solutions and best practices, the research will provide valuable insights to help women succeed in entrepreneurship. Expertise Hub Cooperative (EHC) will benefit from this study by gaining evidence-based recommendations that align with its mission to address systemic barriers and promote equity in the labor market. The findings can support EHC’s initiatives, helping them design better programs and policies to assist women entrepreneurs, particularly immigrant professionals, in overcoming these challenges

Development of highly flame-retardant polyurethane composites for pipe liners

Polyurethane is a flammable polymer that can release toxic gases when exposed to fire, restricting their application in areas such as pipe liners. This project aims to develop fire-proof polyurethane composites pipe liners with a long cycle time and good stability in harsh environments.
Given the increasing demand for advanced materials in industry, we will conduct a systematic feasibility study of different additives (fillers and flame retardants) that will reduce the fire hazard for polymeric materials pipe liners and at the same time, more environmentally friendly. We will use sophisticated laboratory tests to evaluate the feasibility of different formulations of materials. One of the key measurements for fire-resistant of materials is limiting oxygen index (LOI). LOI measures the quantity of oxygen needed burn the material at ambient temperature. We will find materials that do not burn at ambient conditions. We will scale up newly discovered products and fabricate at Rosenxt’s facility in Calgary for use in short pipes.

This project aligns with Canada’s focus on industrial safety, environmental protection, and sustainable materials development. The development of flame-resistant polyurethane coatings will enable industries to enhance fire safety, reduce maintenance costs, and extend material lifespan while meeting stringent regulatory standards. By integrating these innovations into real applications, this collaboration between University of Calgary and Rosenxt will make the company a leader in polyurethane coatings technology and protective solutions, and also benefit Canada’ economy.

Digital Twin Technology in Urban Planning: Housing, Biodiversity, and Infrastructure – The case of Dieppe & Moncton (NB)

Rapid urban growth challenges cities with rising populations, infrastructure demands, and environmental sustainability. Traditional urban planning, based on outdated models and fragmented data, struggles to address these complexities. Digital Twin (DT) technology offers a solution by creating dynamic virtual models of real-world environments using real-time data. By integrating sources like geographic information systems (GIS), sensors, and demographic databases, DT enables data-driven decision-making.

This project examines how DT can enhance urban planning in Dieppe and Moncton (New Brunswick), focusing on housing densification, mobility, urban development, and student housing. The goal is to develop strategies for building resilient cities that accommodate population growth while maintaining high living standards and environmental sustainability. Through DT and AI-driven approaches, the research will provide insights to help the cities plan for future growth, improve traffic management, and conserve biodiversity.

Development of Hybrid Flow Diverting Stent

Our primary mandate at API is to advance basic research and innovation to commercialization by providing access to world-class industry expertise, services, and infrastructure. Our activities focuses on engaging and supporting drug discovery and development initiatives, ensuring compliance with regulatory standards and driving innovation and commercialization through collaborative research and clinical studies. The project aims to research the mechanisms of action in how flow diverting brain stents function and also further the development of a first-of-its-kind flow diverting brain stent, made of both polymer and metal components – leading to new areas of potential design and development pathways for improved products. The anticipated social and economic benefits come from enhanced patient outcomes from effectively treating brain aneurysms.

Improving tropical tree biodiversity mapping from drone imagery to support the scientific study and conservation of tropical forests

Rubisco AI is a startup that aims at monitoring forests worldwide to study their biodiversity using high resolution drop imagery. One of their core applications is to monitor tropical rainforest where tree identification and mapping is currently a very hard task: a single local biologist expert can only identify ~20 individual trees per day due to the difficulty of the task. By contrast, Rubisco AI’s technology could speed this up by at least 1000 x. Current field-based tropical tree inventories are a bottleneck to the overall understanding of tropical rainforests and their use for local communities, and how to protect forests from human activity and climate change. This project will leverage the momentum that both the industrial and academic partners have obtained by being part of the winning team in the XPRIZE Rainforest competition this past year. We developed a pipeline to go from high resolution drone imagery of forests to different forms of annotation, from bounding boxes of trees to segmentation maps of different species. One of the key issues we have encountered is that it is extremely challenging to obtain ground truth labels for less known tree species. A performant tropical rainforest tree segmentation and species identification AI pipeline would have a large number of applications to improve monitoring to help mitigate deforestation, biodiversity loss and climate change in those crucial ecosystems that hold most of the Earth’s terrestrial carbon and biodiversity. This will help to improve tree species distribution models, forests, biodiversity mapping, new species discovery, improved quantification of above-ground biomass/carbon stock and better atmospheric carbon flux predictions. It would also empower local communities as they could quickly and reliably monitor the forest ecosystems they manage and care for. Overall, this would be highly relevant to Rubisco AI and our overall canopy tree mapping solution using drone imagery and AI.

Objective Correlation Method Between Laboratory Direct Shear Tests and Field Measurements for the Evaluation of Joint Strength Parameters

Rock slopes and underground excavations are important aspects of mining, civil engineering and energy projects. The shear strength, or resistance to sliding, of rock joints (e.g., cracks and fractures) is the dominant component for stability of underground excavations and large rock slopes. Laboratory testing of these features is challenging, expensive and difficult to interpret. Emerging technologies, such as laser scanning, AI photo analysis and related technologies provide an opportunity to advance our capabilities to more accurately determine the strength and roughness of rock joints more objectively (less judgement-based), economically (computer aided engineering) and more accurately. This project aims to explore the combined use of some of these data analysis methods to develop a novel joint shear strength determination method to evaluate the statistically representative strength of samples observed during site investigation programs, such as core drilling and downhole drill hole wall imaging.

Investigation on the Enhancement of Design and Manufacturing Processes through the Integration of Large Language Models and Vision Models

In the world of design and manufacturing, there is always a need to supply certain goods (parts, assemblies, materials, etc.) from a manufacturer to a buyer. In most cases, there is a wide variety of procurement options, and choosing the right supply is a complicated process that requires taking multiple factors into account. The optimal supply is difficult to achieve.
Axya makes collaboration between buyers and suppliers easier, faster, and more transparent. Its software optimizes the procurement process by offering simple technological solutions to facilitate and organize the sourcing process.
However, Axya’s optimization process involves the manual processing of engineering documentation, such as engineering drawings and manufacturing procedures. Machine learning techniques, namely Vision Transformers (ViTs) and Large Language Models (LLMs), could facilitate this process by automatically interpreting and extracting knowledge from this documentation.
Thus, the main goal of this project is to develop an adaptation technique for ViTs and LLMs that will equip them with the ability to interpret and extract domain-specific knowledge. The adapted models will speed up Axya’s engineering documentation processing workflow.

L’intervention par la nature et l’aventure en contexte Inuit : Regard sur l’implantation du programme Tuttuit et sur les mécanismes en oeuvre.

Depuis 2014, Nurrait | Jeunes Karibus (NJK) œuvre à titre d’OBNL pour promouvoir le développement personnel et social des jeunes du Nunavik par l’entreprise de programmes d’intervention par la nature et l’aventure (INA). Dans le cadre d’un de leur programme, nommé Tuttuit, NJK a pour objectifs l’identification des forces et intérêts des jeunes, l’acquisition de compétences issues des savoir-être et savoir-faire, la création d’un sentiment d’appartenance, la démonstration d’un progrès au niveau de l’autonomie personnelle et l’exploration du milieu professionnel. Dans un contexte interculturel et dans une perspective exploratoire et descriptive, le projet de recherche mettra en lumière le vécu des personnes participant au programme Tuttuit et permettra de décoder les mécanismes et les ingrédients actifs qui permettent à ce programme d’être un succès. Comme les initiatives en INA en contexte autochtone sont peu nombreuses, notre compréhension d’un tel programme nous permettrait donc de mieux comprendre comment s’articule cette modalité d’intervention en contexte Inuit.