Projets novateurs réalisés

Human Pose Estimation and Activity Monitoring in Hospital with Self-calibrating Cameras

The interns will work on improvements to algorithms using geometry and deep learning for estimating human pose of individuals and the distance between them. This is a difficult task to do from videos as it involves 1) the detection and 2) 3D metric reconstruction of persons in all kinds of poses and apparel. The interns will obtain hands-on experience in algorithmic development, programming, and running validation studies at UBC and HPC’s facilities.
The expected benefit to the partner organization are far-reaching. This work is laying the groundwork for Providence to establish a computer-vision based smart hospital where non-contact based detection models can benefit patient care in areas including but not limited to infection control and patient monitoring.

MATCHING VIDEO CONTENT TO THE DEVELOPMENTAL NEEDS OF PRESCHOOL CHILDREN

The proposed research will include an analysis of video content to determine its implicit learning content based on the social and emotional domains. Once the videos are analyzed, a parent profile will be used to determine what videos are most applicable to each individual child, based on their developmental profile. Further activities will be recommended to parents in addition to the video content. This project will aid the organization in becoming more individualized in terms of content and domain development, according to what the profile indicates for the child. This specificity will allow the organization to market their product as developmentally appropriate, and it will also promote the capabilities to foster particular developmental skills.

Monitoring, Analyzing, and Mitigating Electrical Vehicle (EV) Anomalies and Failures

This project is designed to build a system and software that will monitor and analyze the Electrical Vehicle (EV) bearing anomalies and failures. We will develop a framework that will address the EV bearing failure modes, its effect, and the key features of each failure mode. Later, we will collect the bearing data at “Solution Serafin”, and ingest it in an AI tool for the diagnosis and prognosis of the EV bearing failure. This tool will provide the EV driver and our partner the remaining useful bearing life, and consequently an enhanced maintenance planning strategy. Moreover, we will define the suitable actions to be taken to avoid the EV bearing breakdowns, and we will extend its functionality to reach the nearest suitable replacement time. The proposed system will help our partner to be in control of the EV performance and to find enhanced maintenance actions.

Real-Time Control of Integrated Stormwater Systems using a Model Predictive Control Approach

Burdened by aging infrastructures, urbanisation and climate change, municipalities are seeking innovative solutions to address urban water management. To mitigate flooding, riverbank erosion as well as stormwater-caused pollution, many authorities are now relying on green infrastructures and intelligent flow control instead of the traditional grey infrastructures (e.g., basins and pipes). The objective of this research is to develop an innovative approach to dynamically control emptying flows from stormwater systems to eliminate or reduce flooding and to limit the pollutant loads discharged in the receiving waters. For this, we propose to use a Model Predictive Control (MPC) approach. Using rainfall predictions, measurements and models, the MPC will control gate openings to optimize the use of the stormwater systems’ conveyance and storage capacities in view of minimizing flood risk, erosion and pollution.

Using Machine Learning to Predict 30-Day Risk of Hospitalization, Emergency Visit or Death Among Albertans Who Received Opioid Prescriptions

When utilizing and implementing ML for prediction using administrative health data, two key issues are ML algorithm evaluation and generalizability21. Current approaches evaluate model performance by quantifying how closely the prediction made by the model matches known health outcomes. Evaluation metrics include sensitivity, specificity, and positive predictive value, as well as measures such as the area under the receiver operating characteristic (ROC) curve, the area under the precision-recall curve, and calibration. Because no single measurement reflects all of the desirable properties of a model, several measurements typically are reported to summarize the performance of the model16. Furthermore, model performance ultimately comes down to discrimination and calibration22. Discrimination is usually quantified using a concordance statistic (area under ROC) while calibration is graphically represented as observed to expected ratios.
Generalizability is also an issue that must be acknowledged in ML prediction settings21. ML models trained in one setting may not be valid in another. The same is true for populations. Furthermore, even ML algorithms that are considered generalizable may quickly become outdated as treatment guidelines or the population changes thus requiring model updating and re-evaluation 21.

Integration of Simultaneous Localization And Mapping (SLAM) to improve workflow of reconstruction projects and space utilization

The focus of this project will be how modern technologies, specifically static and mobile laser scanners, drone photogrammetry, and Virtual Reality (VR) can be applied to solve issues related to renovating and utilizing (repurposing) old buildings. This is a multi-disciplinary approach with college interns from Geomatics Engineering and Architecture Engineering programs.

Implementing lessons learned from the pandemic to support our vulnerable populations

The purpose of this project is to improve the lives of citizens in The Region of Durham through a partnership between the Regional government and Ontario Tech University. This project will tackle issues related to aiding vulnerable populations in our communities such as members of our homeless population and seniors living in long-term care homes. It will accomplish these goals by:
• Examine how feasible and effective it is to set up community hubs for the homeless and other vulnerable populations, where they can meet all their health, mental health, and other needs in one place.
• Determine what factors/attributes make nurses and personal support workers decide whether or not they want to be employed in long-term care workforce, with the goal of attracting more prospective nurses and PSW to work in this sector.

Plant growth response to growth promoting rhizobacteria

Numerous species of soil bacteria flourish in the rhizosphere of plants, which may grow in, on, or around plant tissues and stimulate plant growth by a plethora of mechanisms. These bacteria are collectively known as plant growth promoting rhizobacteria (PGPR). Bacillus velezensis is a PGPR that promotes plant growth, enhances drought stress tolerance, and suppresses plant pathogens. However, little is known about the interactive effects of exogenous orange peel amendments and B. velezensis PGPR strains on plants growth and productivity. The project aims to (i) identify elite strains of B. velezensis for plant growth promotion, (ii) evaluate the best mode of application of B. velezensis PGPR strains amended with orange peel powder to enhance plant growth and physiological parameters. We will also test the effect of the growth promoting rhizobacteria on seed viability, germinability and seedling vigor and establishment. The test plant will be kale.

Acylketenes in Catalytic Cycloaddition Reactions

The reactivity of acylketenes in carbon-carbon bond forming reaction has been scarcely explored, despite their ease of access and stability. The student funded through this program will investigate the reactivity of these unexplored species in intramolecular transition metal-catalyzed cycloaddition reactions.

These studies will establish the reactivity of acylketenes in cycloadditions, and reveal novel reactivity pathways, leading to tunable catalytic carbon-carbon bond forming reactions, and allowing to access a diversity of structures by simply modifying the catalysts/ketenophiles. The resulting synthetic strategies will provide expedient entries into an array of complex carbocyclic frameworks found in biologically active natural products and pharmaceuticals. During the course of this project on the exploration of the synthetic potential of acylketenes, the student will receive an excellent theoretical and practical training in synthesis and catalysis.

Feasibility Study of Wave Energy Conversion with Grid Connectivity

A renewable energy source that has not received much attention is tidal and wave energy, although numerous studies have concluded that wave power, and to a lesser extent tidal power, could contribute massive amounts to the overall energy picture. This project aims to explore the feasibility of tidal/wave energy absorption and its storage in the form of on-shore energy bank and using that to power on shore devices/generators. The primary activities in the proposed R&D activity include developing a proof-of-concept system in the lab as a pilot system for performing certain experiments using a hardware-in-the-loop platform. The lab-scale system can help to identify feasibility and performance of a scaled-up system under more practical conditions and the possibility of commercialization. The concept to be explored is an offshore surface floating foil that acts as a buoy, which acts to raise and lower a mechanical arm located at a nearby onshore pivot point.
This study will produce proof-of-concept results and help the team and the partner organization, Greenergy, in making commercialization decisions on the developed technology that has desirable features such as high-efficiency, economic viability, and environmental friendliness.

Market making for digital assets

Market makers facilitate trading in electronic financial markets by simultaneously offering to buy and sell the same asset at any given time. Their role is to provide price stability and increase market liquidity to improve its overall efficiency. Digital assets markets are extremely fragmented and present both challenges and opportunities for market makers. These latter must offer participants accurate prices while balancing their asset inventory on many venues at the same time, what represents a difficult synchronization task. At the same time, this complex environment leaves room to arbitrage opportunities, which market makers can use to offset inventory risk. The project consists of developing dynamic programming and reinforcement algorithms that take advantage of this environment to improve the profitability of market makers while reducing their risk. To achieve this goal, a large volume of historical data is used in conjunction with a machine learning-based market simulator.

Leveraging Deep Learning in Asset Pricing in a Multi-Factor Modelling Framework

The purpose of this project is leverage Machine Learning technology to develop and test effective trading strategies in order to properly hedge an investment strategy. Hedging is an integral part of the investment process and allows portfolio managers to protect their positions against any adverse change in asset prices. As it is a task that requires solving a range of highly complex problems, using Artificial intelligence is proving to outperform traditional techniques currently used, especially when it comes to decreasing costs and improving returns.
By optimizing the effectiveness of the trading models, this project will allow Quantolio to integrate a superior hedging technique in its final product (platform). As a result, the organization will be able to offer solutions that help portfolio managers enhance their investment strategies.