Projets novateurs réalisés

Control of multi-agent robots: an opinion dynamics approach

The objective of this project is the development of a novel control strategy for multi-agent robots. The objectives are energy saving and efficient robots movements and manoeuvres within clustered environments. To this end, the research intern (with the help of the host and home supervisor) will be in charge of developing a novel distributed control strategy based on the exploitation of the opinion dynamics among the involved groups of vehicles. To this end, the intern will to start from recently published contributions of the host and home supervisors in order to derive a new approach that leverages the tools developed by both research groups. The resulting solution will be tested via laboratory experiments using a team of differential-drive robots available in the host supervisors research lab at Concordia University.

A Machine Learning-Guided Platform to Engineer Functional Analogues of Therapeutic Peptides

Comprised of a small chain of amino acids, therapeutic peptides are a unique class of drugs that have played a significant role in medicine since the discovery of insulin. Despite their importance in Pharma, the discovery of new peptides and the improvement of existing therapeutics remains a highly specialized domain and has not seen the cost barrier fall yet. Additionally, the pharmaceutical industry is heavily patent-protected allowing only a few major corporations to monopolize the market.
Novel platforms for designing patent-free mimetics of high-value peptide-based therapeutics can potentially mitigate this problem. Combining the advantages of synthetic biology-enabled cell-free systems with microfluidics, I propose to create a machine learning-guided platform for designing functional analogs of high-value peptide therapeutics that share less than 50% sequence identity with the original therapeutic.

On-the-fly World Models for Quadrupedal Robots

Deep reinforcement learning (DRL) for quadrupedal robot control has recently become tractable. In under an hour, serviceable control policies can be obtained in simulation. However, transferring reinforcement learning from simulation to the real world is still an arduous task. We propose a novel two-agent setup that leverages the good sides of both model-free and model-based DRL. We propose introducing a new agent is to the model-free control policy. By tasking the new agent with regularizing differences in operating environments by generating world models on-the-fly that will act as regularizers over the control policy’s inputs, we will obtain a reliable sim2real strategy to help take quadrupedal control polices out of simulation and into the real world.

Ethical Awareness of Japanese Computer Science and Software Engineering Graduate Students on Text-to-Image AI-Generated Art

Text-to-image AI models for AI-generated art has recently gained extensive media coverage and has had a significant impact on Japanese art communities and Japanese digital content sites. I am particularly interested in investigating Japanese computer science and software engineering graduate students’ awareness of ethical issues embedded in the technology necessary to create AI-generated art. This is because these students are the primary demographic hired to work at research and development (R&D) labs at the forefront of such innovations. By understanding the amount of ethical values held by these students or the lack thereof, there will be a clear understanding of what gaps in digital literacy and ethical awareness are held by prospective researchers in the field. This will better inform what areas in computer science or software engineering higher education can be improved to increase ethical and critical thinking towards the technological development, creation, public evaluation, and moderation of AI-generated content and related technologies. These gaps will be addressed through the creation of an international workshop for computer science and software engineering graduate students to assess and critique text-to-image AI-generated art.

Heart-lung synchronizations during exercise

Patients diagnosed with heart failure often require the implantation of a pacemaker to improve cardiac function, which in turn, improves exercise tolerance and quality of life. Current pacemaker models work at a fixed time interval for a given heart rate (i.e. 1 second between heart beats at 60 beats per minute). However, the body’s natural homeostatic system creates variations in the time interval in line with one’s breathing cycle (termed respiratory sinus arrhythmia; RSA). A novel pacemaker model has restored RSA when implanted, showing large improvements in cardiac function in animal models of heart failure. While restoration of RSA at rest may improve cardiac function, this synchronization between heart rate and breathing disappears during exercise. Thus, Dr. Stembridge has set out to explore alternative heart-lung synchronization mechanisms in healthy populations during exercise, to guide future pacing principles for novel pacemakers. We will evaluate these mechanisms and their effect on cardiovascular function during exercise in healthy individuals. We hypothesize that cardiorespiratory phase synchronization will remain different across exercise domains in young healthy individuals, and hypo- and hyperventilation will lead to a decrease in left heart output during exercise and exaggerated heart rate response relative to exercise stage.

Santana Airfield: linking place, identity, and memories through performance into civic participation

This project explores how an artistic oral history-based audio walk can be used as a tool for connecting locals, visitors and immigrants with a place and engage them as participants in public debates concerning local development projects in the Azores. Santana Airfield, Azores, is an abandoned complex risking demolition. Zoning laws are changing the landscape. The personal narratives of elders linked to it are also in danger of being lost. The intent of this project is to have participants explore the land and buildings of Santana Airfield in a walking performance before and while listening to the sound art followed by a post-walk discussion on how this artistic practice can be used to foster community engagement in transformative activities as compared to other forms of calls to action. The research ends with the documenting of the participants’ initiatives such as media output, petitions, and town hall meetings to turn Santana Airfield into a heritage site.

A continued optical function analysis for thin-film device structures

Informed thin-film device design, wherein layers of thin-films are deposited onto an underlying substrate, requires an understanding of the spectral dependence of the optical functions associated with the thin-film layers which constitute such a device. In this project, we aim to continue to develop our understanding of the form of the optical functions for the various types of thin-film materials and then use this knowledge in order to aid in the interpretation of experimentally acquired transmittance and reflectance spectra. We will start by continuing to expand, curate, and complete a library of optical functions corresponding to a plethora of thin-film materials. A series of models that aims to capture the underlying physicality captured in this library will be devised. We will then use these models in order to narrow the parameter space that must be probed in determining the spectral dependence of the optical functions associated with the thin-film layers from measurements of the transmittance spectrum at normal incidence and the reflectance spectrum at near-normal incidence. Finally, we aim to develop a protype software package that aims to capture the results of our investigations.

Encrypted control for Cyber-Physical Systems

The project is devoted to make encrypted control systems more practical and valuable to the industry and ensure the security of safety-critical industrial control systems.

The research outcome is the development of a novel computationally efficient and secure MPC-based encrypted control system that is appealing to a wide range of safety-critical industrial sectors, including but not limited to Energy, Smart Manufacturing, and Smart Transportation. To this end, an non-encrypted computationally low-demanding MPC strategy developed by the home supervisor and the encrypted control solutions developed by the host supervisor will be investigated, and a new encrypted design will be developed.

Personalization with Integration of Sensor Signals in Cloud Architecture

$50B of unrealized GDP will occur by 2030 in Canada if the skill gap challenge is not addressed. With the acceleration and uncertainty of a borderless economy, Employers, Educators and Jobseekers will continue to be in constant flux. The greatest cost to Small & Medium Sized Businesses (SMBs) is the hiring, skilling and training of new hires, and the cost when they leave after a day or after 30 days.
Recent seismic talent layoffs at the conglomerates of Amazon, Twitter, Facebook have now started this ripple effect and won’t take long to impact SMBs, Jobseeker and Educators.

ELSTec : Efficient lightweight structures made by automated fiber placement (AFP) of thermoplastic composites

This project deals with the continuation of a project at Concordia University in 2020 via Mitacs Globalink research
award. The project investigated different manufacturing scenarios of a thermoplastic skin/grind structure using automated fiber placement (AFP) in the Concordia Center for Composites labs. The thermoplastic skin/grin structure was successfully fabricated via AFP in-situ consolidation.
Now, this project will evaluate the plate with the help of the existing experimental facilities at Concordia University. The goal is to evaluate the performance of the in-situ consolidated skin/grid structure through mechanical testing and microscopy analysis. Mechanical test will be performed using universal testing machine and stiffness and strength properties will be extracted. Void
content and delamination of the specimen will be investigated with microscopic analysis. Finally, the results are compared with a Finite Element (FE) analysis to get further insight of the failure behaviour. With the results it is possible to benchmark the developed skin/grid structures with already existing’s ones and it should be possible to draw conclusions to improve the structure itself and the manufacturing process with AFP.

Swine feed efficiency improvement and meat quality co-selection using genome wide technologies Year Two

Improving feed efficiency (FE) and meat quality (MQ) are Canadian swine industry priorities that will increase sustainability and competitiveness. It is difficult to improve these traits simultaneously by traditional breeding methods due to limited knowledge of genetic interactions, limitations on animals that can be measured and the high cost of measurement. High throughput genomic technology has the potential to positively impact these issues. Our overall purpose is to develop a genomic assisted breeding program to simultaneously improve FE and MQ. This collaboration will benefit the partner’s long term competitiveness through implementation of genomic technology into existing breeding programs and training of high qualified personnel. An example of the potential economic benefits to the industry by improving FE: herd feed conversion (kg feed/kg pork) is now ~3:1, each 0.01 improvement in FE (e.g. 3.00 to 2.99) represents ~200,000 tonnes of feed/yr saved, worth approximately $55 million to the Canadian pork industry.

Development of Copper Based Additive Manufacturing Techniques

The goal of the project is the development of copper for Rapidia’s innovative paste extrusion process. Copper is a highly sought-after material with unique properties that has a wide range of applications in various industries including biomedical, electrical, manufacturing, and energy storage. The ability to 3D print copper parts on our system will thus have a large impact on several manufacturing industries.