Innovative Projects Realized

Understanding robot locomotion characteristics and planning in heterogeneous multi-robot coverage

To be safe and efficient, autonomous robots will need to consider and understand one another when performing their respective tasks. Communication and joint planning between robots might not be possible due to the number of robots, adversarial behaviors, or the lack of common protocol.
In this project, we start by tackling this large issue at a small scale by focusing on a specific situation: multi-robot coverage with heterogeneous robots. We consider a team of robots that have to be deployed to cover a large scale environment. Each robot has different locomotion characteristics but they all share a common objective: for the team to cover the whole area as efficiently as possible.
The idea of this work is for each robot to use visual information to gather a semantic understanding of other robots’ characteristics and of the environment. TO BE CONT’D

Combining cortical stimulation and ‘mirrored’ visual feedback to enhance recovery of upper-limb: a pilot study.

Stroke is one of the most prevalent neurological conditions worldwide, especially among the elderly.

Motor impairment is one of the main disabilities associated with stroke. The ability of training and

physical activity to restore motor function after neural injury has long been appreciated. Nevertheless,

even with intensive task-specific training and physical activity, 15–30% of patients with stroke are

permanently disabled. As a result,research aiming to develop novel therapies that enhance

neuroplasticity is currently being conducted to allow rehabilitation of these patients. A promising

approach is modulation of cortical excitability using non-invasive brain stimulation techniques. These

techniques may instantaneously induce plastic changes within the cortical network of sensorimotor

areas, thereby improving motor function of the affected hand after stroke. Our work attempts to

elucidate whether repetitive administration of inhibitory non-invasive brain stimulation of the

unaffected hemisphere in sub-acute stroke over several days combined with sensorimotor

behavioural training using optical left-right reversing prisms could result in an enhanced effectiveness.

The primary objective is………….TBC

The Development of Phonological Awareness in Vietnamese Children

Vietnamese is widely spoken by over 90 million people in Vietnam and around the world. The proposed project will be the first to study phonological awareness (PA) and reading outcomes in Vietnamese-speaking children. Phonological awareness is the ability to explicitly and/or implicitly manipulate small meaningful units of speech. Many studies with English-speaking children have found PA to be a strong predictor of reading ability. The proposed project aims to measure the growth of PA skills in Vietnamese-speaking children from kindergarten through second grade, and to identify relationships between PA skills and reading outcomes. Results from the analyses will contribute to the existing knowledge base about PA. In the future, the findings will help to identify Vietnamese-learning children who are at risk for reading difficulties and inform effective instructional strategies. Finally, the applicant will develop research skills relevant to clinical practice in her future career as a speech-language pathologist.

Extending artificial intelligence in the operating room

Assessment of surgical data from an operating room is a complex process that may require significant resources such as expert input and advanced technology. Automation brings a considerable opportunity to greatly reducing these significant resource requirements – e.g., using computer vision software to detect clinically relevant actions during surgery. However, those detections should be interpretable, or more actionable in order to be audited or reviewed. This project applies various ‘explainable AI’ (XAI) technologies, which will allow us to evaluate the impact of solutions, which generalizes across the healthcare syste

Enhancing steady state processing of plant based ingredients for the food industry

In partnership with KeyLeaf (KL), Saskatchewan Polytechnic (SP) is developing a system for the bioprocessing industry that will interface through a mobile app, collect data on a NoSQL database, and use analytics to mine it for new insights into all processing aspects. The Engage phase of the project will commence in tandem with a MITACS grant of the same scope, involving building the database (DB) and user interface, determining how to best structure the data for analysis, designing the analytic algorithms, then testing the system and prepping for the next phase. This will produce a faster bioprocessing system for KL capable of identifying areas in its own processes that can be improved, thus increasing product quality. SP will benefit from knowledge exchange with KL and plant-based ingredient processing techniques will be fine-tuned.

Pelletization of Forest Wildfire Fuels in Western Canada

Forests in British Columbia have significant areas that need to be managed to limit the danger of fire and disease. Under current management approaches, there are few cost-effective options available to areas that are not already regularly harvested. A holistic analysis of pelletization of these materials could show that the sale of wood pellets, as well as the reduction in fire suppression and forest management costs and the increase in tax base would generate a net profit to the province and the businesses associated with the management.

Development of the Parasocial Identifier for Social Media

Since the introduction of radio and television, fans have had a connection to those they know in the media. These connections are called “parasocial relationships” and they are one-sided relationships, usually between a fan and a celebrity, that is usually not face-to-face. With the advent of social media, any person, or company, can become a celebrity and gain fans. In this project, the intern will work with a marketing company to develop computer algorithms to automatically analyse comments on Instagram and determine how many account followers have parasocial relationships with the account owner. He will then develop a way to summarize these results in a parasocial identifier (PSI) score.

Développement d’un outil d’analyse de texte assistée par ordinateur permettant de classifier les entreprises dans l’écosystème de la 5G au Canada

Prompt doit identifier les PME qui ont un fort potentiel de croissance au sein de l’écosystème canadien de la 5G. Dans ce contexte, la recherche propose d’exploiter une base de données textuelles. Elle est constituée par les sites internet de 7 000 entreprises. Cette base de données a déjà été construite par Prompt. Le premier but de l’analyse est l’exploration des spécificités thématiques des sites internet et leur classification par groupe d’entreprise. Plusieurs algorithmes et méthode statistiques venant du domaine de la fouille de texte (ou text mining) et du traitement automatique du langage naturel (ou Natural Language Processing – NLP) seront utilisés à cet effet. En particulier, les méthodes de fouille de thème (ou topic modeling) seront utilisées pour explorer les données textuelles.

Analysis Tools for Automated Rehabilitation Assessment and Progress Tracking

In this project, Cardon Rehabilitation and Medical Equipment (CRME) and the University of Waterloo (UW) will test the Automated Rehabilitation System (ARS) for lower body physiotherapy in a clinical environment. ARS provides the ability to measure the human pose during the performance of rehabilitation exercises and provide real time feedback. CRME and UW are interested in developing algorithms to automatically assess the changes in patient recovery rate. The intern will develop numerical metrics of exercise performance which can be used by clinicians to better understand patient performance and progression as a supplement to current subjective scales that are common in clinics and reduce inter-rater variability.

Characterizing intracellular interactions of carcinoembryonic-antigen-related cell adhesion molecule 1-4L using high resolution microscopy

Preliminary work has shown that adding excess CEACAM1-specific antibody, 26H7, to hela cells expressing carcinoembryonic-antigen-related cell-adhesion moleule (CEACAM) 1-4L decreases CEACAM1-4L in the plasma membrane and shifts their surface expression from a monomer-oligomer mix to mainly monomers. A super resolution microscope will be used to co-localize CEACAM1-4L with 26H7 in order to determine the optimal antibody dose, and how different doses affect CEACAM1-4L spatial distribution and surface expression. A microscope equipped for 3D imaging of live thick samples will be modified to be able to determine the the monomer-oligomer mix of CEACAM1-4L upon addition of 26H7, in 3D cell structures. Intracellular vesicles leaving the plasma membrane will be tracked and quantified. It is predicted that CEACAM1-4L dimers are being internalized, so I expect to see mainly monomers at the membrane, and an increase in CEACAM1-4L dimer containing vesicles leaving the membrane.

Enhancements to Smart Disease and Pest Prediction System Through the Use of Machine Learning Techniques

Given the current global environmental crisis, developing sustainable solutions to enhance or replace our current agricultural practices is critical: the agricultural sector exerts important environmental pressure through its aggressive land, water and pesticide usage combined with the ever increasing demand on food supply. Mitigating this problem requires developing more sustainable and efficient agricultural techniques. Precisely, advances in networking and sensing technologies allows us to gather vast amounts of data in the fields that can be fed to complex machine learning and data mining algorithms to enhance current pest and disease prediction models. Such models feature unprecedented accuracy and flexibility, ultimately allowing farmers to significantly reduce their water and pesticide usage. This is exactly the kind of solution the current project will focus on by helping Ukko Agro, a Canadian company, extend the capabilities of their current precision agriculture solution.

Cognitive performance in patients with coronary artery disease undertaking exercise: imaging the role of brain-derived neurotrophic factor

Patients with coronary artery disease often present clinically with a cluster of vascular risk factors, which predispose them to stroke and can cause subtle deficits in cognitive performance. These deficits are believed to pertain, in part, to disease of the small blood vessels providing blood to the brain. Fitness is associated with better cognitive performance, but it is unclear if fitness interacts with small vessel disease and what the biological mechanisms might be. This project will test cognitive performance in patients with coronary artery disease, utilizing magnetic resonance imaging to quantify small vessel disease. In addition, we will investigate possible effects of a variant in the brain-derived neurotrophic factor gene to understand how this gene fits into the mechanism of cognitive resilience to small vessel disease. This project will advance two key foci of the Centre for Stroke Recovery; exercise and small vessel disease in the patients of the Toronto……………….