Projets novateurs réalisés

A Lane-level Navigation System with HSLIG Fusion Scheme for Smart Vehicles

As autonomous technology evolves, smart vehicles emerge as a pivotal development for transforming public transportation and minimizing transport costs. Key to their broad acceptance is the resolution of safety, stability, and compatibility issues. This project proposes a cutting-edge solution through a High-Definition (HD) map-aided, multi-sensor fusion approach, aiming at precise lane-level positioning and navigation in dense urban canyons. The objective is to create a resilient sensor fusion scheme that seamlessly merges the onboard sensors with environmental contextual information from High-Definition (HD) maps, achieving lane-level navigation even in urban canyon environments. The expected output for the industry partner will be a prototype of a navigation system that can be deployed on the partner organization’s smart vehicles, facilitating reliable lane-level navigation in typical urban canyon environments.

Integrating graph-based data management into materials acceleration platforms

This research project aims to significantly improve the way data are managed in a specific self-driving laboratory in the AUTODIAL group of Prof. Hattrick-Simpers at the University of Toronto, focusing on discovering new materials that are resistant to corrosion. This class of labs, known as Self-driving labs (SDL) or Materials Acceleration Platforms (MAPs), use advanced technologies, such as AI and automated experiments. The project introduces a new system for organizing and analyzing data using a new approach called a graph database, which is better at handling complex and interconnected information. This upgrade will also involve the use of sophisticated language-processing technologies to better understand and utilize the data collected. The goal is to make the labs more efficient and effective, reducing the time and cost of the experiments and simulations. The project will also compare how these labs communicate in an in-operable fashion with similar labs across Canada and Germany to identify common challenges and solutions, ultimately aiming to accelerate the development of new materials.

2D material band-gap prediction by machine learning

Two-dimensional (2D) semiconductor materials are materials with thickness on the atomic scale that provide unique properties compared to their 3D counterparts. One important property of semiconductors is their band gap, which dictates how the semiconductor material will behave. However, manufacturing and testing 2D semiconductors can be costly and difficult, so the ability to predict the band gap of 2D semiconductors would be very useful to allow researchers to focus efforts on testing materials that will yield desired band gaps. This project aims to train a machine learning algorithm on data from 2D and 3D semiconductors to predict the band gap of new 2D semiconductor materials. This machine learning approach has the potential to be more accurate and quicker to compute than current quantum-mechanics based computations. This project will benefit both institutions by developing the application of machine learning in materials engineering, as well as providing an algorithm that can be used by researchers and engineers to design new 2D semiconductor materials.

Principes de logiciel de nouvelle génération d’assistance à la mise au point de procédés industriels

Dans un cadre éducatif, les procédés industriels sont rarement accessibles pour effectuer tests. Le recours à des outils de simulation de modèles dynamiques de ces procédés est souvent une solution pour permettre l’étude de ces procédés. Le réalisme de ces simulations est étroitement lié à la précision des modèles utilisés. Le présent projet vise à développer des modèles mathématiques des principales unités d’opération utilisées dans le domaine du génie des procédés afin d’inclure un nouveau volet dans le logiciel de simulation Automation Studio, propriété de notre partenaire Famic Technologie Cette bibliothèque sera d’une grande importance dans le milieu éducationnel et au niveau de la formation des ingénieurs et permettra d’attirer des nouveaux clients intéressés par cette nouvelle bibliothèque.

Systemic and local role of complement component 3 (C3) in an experimental model of autoimmunity

Building Trust in AI-Generated Content: Innovative Strategies for Quality and Integrity Verification

In an era where AI can write articles, create reports, and even craft stories, ensuring this content is accurate, free from errors, and trustworthy is crucial. Our project, “Building Trust in AI-Generated Content: Innovative Strategies for Quality and Integrity Verification,” aims to tackle this important challenge and to measure and increase the reliability and trustworthiness of the content generated by artificial intelligence (AI).
We plan to develop new methods to check the quality and truthfulness of AI-generated text, making sure it meets high standards. For our partner organization, this means their AI platforms can produce better, more reliable content that users can trust. This will not only improve the organization’s reputation but also pave the way for safer, more effective use of AI in various sectors. Through this project, we’re trying to make AI a more dependable tool for generating high-quality content.

Sea Ice thickness distribution modeling

The aim of this internship is to carry out an analysis of the equations modelling the distribution function of sea ice thickness, particularly in the case of large deformations such as ridges. Among other things, we will be interested in the numerical solutions of these equations, their stability and their consistency.

Two-Stage BIM-Enabled Decision Support System for Selection of a Suitable Industrialized Building System

This study aims to promote and advance the application of Offsite Construction Manufacturing (OSCM) by
developing a BIM-enabled Decision Support System (BeDSS) to select and execute a suitable Industrial Building
System (IBS) for a given construction project.
The specific objectives of this research are as follows:
– To perform an OSC feasibility study of a building project by assessing the key decision-making factors
impacting the successful completion of the project.
– To assist decision-makers to select and execute a suitable IBS using the proposed Decision Support
System based on BIM data and decision-making factors which are associated with successful OSCM
project indicators.
Design Science Research is adopted as methodology for this project. Fuzzy AHP will be used to rank the different
types of Industrial Building System (IBS) alternatives.

Enhancing Autonomous Driving using Multiple Large Language Models (MLLMs)

Nowadays, leveraging advanced technologies like Generative Artificial Intelligence (AI), particularly Large Language Models (LLMs) such as GPT, holds promise in revolutionizing safety measures and resource optimization. However, while these general-purpose LLMs excel in various tasks, they may lack context specificity. Domain-specific LLMs, such as those tailored for biomedicine and transportation, are emerging to address this issue. For instance, in transportation, Multimodal Large Language Models (MLLMs) show the potential to enhance autonomous driving and traffic safety decision-making. Efforts also focus on using LLMs for accident prediction and prevention, with lightweight models proposed for real-time interventions. Despite advancements, there’s still a need for versatile AI agents capable of adapting to diverse scenarios. LLMs offer a foundation for such agents, with ongoing research exploring their potential. This project aims to integrate multiple LLM-based Intelligent agents into a framework for autonomous driving, enhancing decision-making and public safety by reducing accident risks. This initiative seeks to explore the use of MLLMs and AI agents in developing a novel autonomous driving framework.

Detecting atopic dermatitis with impedance measurements

This research improves atopic dermatitis (AD) detection, a common type of eczema, which affects about 20% of Canadians. AD causes dry and inflamed skin due to a lack of a key protein, making the skin barrier less effective against moisture loss and allergens. Current treatments like moisturizers and steroids can have side effects, and diagnosing AD accurately can be challenging, delaying treatment.

Traditional diagnosis involves a healthcare provider examining the skin and considering medical history, but this can be prone to errors. To improve this process, electrical impedance spectroscopy (EIS) can be used, which involves passing small electrical currents through the skin to detect abnormalities in cell structure. While this method has been used to detect skin cancer, its potential for AD diagnosis hasn’t been widely explored.

Our research will develop a mathematical model and simulate a device using computer programming. This device will be run through both healthy and affected skin areas, measuring changes in current strength and phase. The data will be used to create a graph, providing an overview of the skin’s condition.

The goal is to create a reliable and easy-to-use device that can accurately diagnose AD, leading to earlier treatment and improved outcomes for patients.

Quantum super-resolving lenses

Super-resolving lenses are able to beat the standard diffraction limit in optics and image objects smaller that the wavelength of light. Such devices are already important in areas such as the bio-medical sciences, and but could also find application in the emerging field of quantum technology by providing a way of coupling qubits with high fidelity. However, the operation and capabilities of some super-resolving lenses such as the Maxwell fisheye lens, which give perfect imaging in the limit of geometric optics, remains controversial in the more accurate wave theory of light. This theory project proposes to examine these issues by mapping to well-studied problems in quantum mechanics (such as the inverse square potential) that share similar properties, including non-hermitian features. This project would combine the expertise of two groups: one at the University of Birmingham which specializes in optics and topological techniques, and one at McMaster University that specializes in the quantum mechanics of singular and non-hermitian potentials. This project will benefit the larger optics and quantum information communities in both countries.

Laser Annealing is Full-Scale Electron Detectors

The primary objective of the Globalink research project is to develop and implement advanced techniques for the detection of energetic electrons in silicon detectors, with a focus on overcoming the limitations of traditional detection methods and mitigating radiation damage. We will try to investigate whether laser annealing and pixel migration techniques can effectively mitigate radiation damage in silicon detectors, restoring their functionality. The second question we are trying to answer is if optimization of detector design parameters will enhance the performance of silicon detectors in detecting energetic electrons and minimize radiation-induced damage. Max Planck Institute is at the forefront of physics and matter research and I would learn methods that would aid in Canada’s development in electronics and the technology sector.