Projets novateurs réalisés

Problème de conception d’horaire en sécurité aéroportuaire

Le problème de conception d’horaire en contexte de sécurité aéroportuaire consiste à établir un horaire qui minimise le nombre d’employés qui sont nécessaires pour faire fonctionner un nombre de points de contrôle, dans le but de desservir le flot de passagers qui se manifeste en réponse aux multiples départs d’avion. Le problème s’apparente au Nurse Scheduling Problem, qui est connu comme étant NP-difficile. Nous cherchons donc à créer de nouveaux algorithmes spécialisés afin de résoudre le problème de manière plus efficace que l’approche courante basée sur la recherche locale.

Recognition of Vehicles in Satellite Views

The intern shall endeavour to develop a vehicle detection system for aerial satellite views with a reasonable accuracy and a reasonable execution time for a small urban area of 100 sq. km. The intern first will train, test, and evaluate a machine learning engine to automatically recognized the vehicles of interest in aerial satellite imagery. The sponsoring organization, LTAS Technologies will have an engine and a process that can be repeated for each state in the USA. It could then provide the service to its customers which can also benefit by collecting taxes due on those vehicles.

A Condition Based Monitoring Pilot for the Baggage Handling System at the Pearson International Airport

Maintenance of baggage handling equipment, passenger boarding bridges and aircraft support equipment currently costs the Pearson International Airport (PIA) $10M annually. Much of this cost is associated with routine inspections, replacing drives, lifts and similar industrial process equipment. Current maintenance undertaken by PIA is reactive; replacing or repairing parts on the equipment after break-down. This is an expensive process. In this proposal, the intern will implement a condition based maintenance pilot on the baggage handling system (BHS) at PIA. An online vibration monitoring system will be implemented to collect vibration data from the BHS and this data will be interrogated to provide valuable information regarding their state in addition to the event data. This condition information will be combined to calculate the reliability of the unit as a function of its current condition and age, through a proportional hazard model and will be used for inspection planning.

: Identification of organic disinfection byproduct precursors at the Waterhen Water Treatment Plant

There are about 180 public water treatment plants in Manitoba relying on surface water with high organic content as their source of supply. Organic carbon in the water reacts with chlorine during water disinfection. This reaction forms carcinogenic compounds which pose a health risk for people consuming the water.

The Waterhen water treatment plant is an example of typical Manitoba treatment plant that is suffering from high organic carbon resulting in high concentrations of these carcinogenic byproducts.

Experiments will be conducted to evaluate the types of organic compounds in the water and to evaluate their removal throughout the Waterhen plant. Improving the removal of these compounds will improve the overall quality of water and health in the region.

Centre for Operations Excellence Industry Projects 2015

The Centre for Operations Excellence Industry Projects consists of five sub-projects sponsored by four different industry partners. Each sub-project represents an important challenge for its sponsor. These sub-projects include using analytics to optimize sawmill production for Interfor; production planning for Tree Island; developing text-mining techniques to enable WorkSafeBC to predict and
prevent workplace accidents; using Twitter data to enrich Boeing Canada’s maintenance and operations planning tools; and performing human resources analytics to improve Boeing Canada’s workforce planning.

Spatial pattern of wildlife habitat across heterogeneous landscapes in Atlantic rainforest near São Paulo, Brazil

Transitions between adjacent plant communities are important features of landscapes that might harbour greater diversity. However, forest edges created by human activities can have negative consequences for wildlife habitat. Both types of transitions dominate fragmented agricultural landscapes in tropical forests. Understanding the vegetation structure in such heterogeneous landscape could help predict animal movements across natural and anthropogenic transitions. This project compares the relationship between structural diversity and bird movement across different types of landscape transitions as part of a larger joint study on animal movement in the Brazilian Atlantic Forest. The objectives are to determine the relationship between structural diversity and frugivorous bird assemblages and to compare seed distribution across forest edges and landscape transitions with different structure. The outcomes include an assessment of the usefulness of structural diversity as an indicator for wildlife habitat, which will be included
in the multidisciplinary research program on animal movement in Brazil’s forests.

Exploration of Innovative Digital Consumer Research and Marketing Strategies

This project will analyze the current digital marketing strategy for an online jewelry seller. With this, a better understanding of the current competitive environment of online jewelry retailing will be developed, different customer profiles will be identified, and shopping behaviors will be identified. Finally, the research will examine the impact of online shopping experience on jewelry shopping
behavior. Based on the findings, new digital marketing and branding strategies will be recommended. This project will bring benefits to the organization by providing an external analysis and assessment of the current online marketing practices and improvement opportunities and new strategies to implement; the latter can bring potential improvements in sales and customer satisfaction.

Fiber Optic Temperature Sensors: Phosphor Deposition and Processing Improvements

Photon Control R&D LTD is a Burnaby, British Columbia company that designs and manufactures a wide range of optical sensors & instruments to measure temperature, pressure, position, and flow in the semiconductor, oil and gas, power, life science, and manufacturing industries. The operating principle of many of their products is based on the optical response of phosphor materials and is well suited to applications in harsh environments. Photon Control is currently seeking to improve the lifetime of their optical sensors by improving the quality and methods of deposition of their phosphor materials. Our research group has the infrastructure and expertise required to deposit and to protect these materials in robust thin film form on substrates of various size and shape, in order to address these material deposition challenges. This project will complement our current research efforts and allow Photon Control to achieve performance
improvements in their optical sensor systems, providing extended lifetime in their current applications, as well as extending their range of applications to include high temperature environments, where there is a demonstrated need.

Modeling of Current Collapse and Gate Leakage Phenomena in AlGaN/GaN HEMTs

AlGaN/GaN high electron mobility transistors (HEMTs) have emerged as promising candidates for high breakdown, high power output, and high operating temperature applications. However, there are several problems that hinder its practical use such as current collapse and high gate leakage current. Most of available studies use analytical formulas and only analyze a certain aspect. Physical models that can predict both phenomena consistently and describe a larger picture of device behavior are still lacking, which will be addressed in this work. This capability is important in the HEMT structure design, and will make APSYSTM a more competitive tool in the market.

Machine Vision Based System for the Recognition and Counting of Indian Coins

Counting coins, with speed and accuracy, has been a challenging issue for banks and stores. People used to count coins manually before the arrival of coin counting machines. The process of counting coins manually is a very time consuming and tedious job. Moreover, mistakes are likely to occur due to various reasons such as fatigue, eye tiredness and too many coins of nearly same shape and size cause confusion in sorting and counting. Coin sorters are common in North America and can be found in most commercial banks and even some grocery stores. By contrast, they are not available in India, where the number and similarity of the coins make for a very challenging problem. The objective of this project is to determine whether advanced machine vision techniques are able to sort coins from India with acceptable speed and accuracy. If the answer is yes, then the outcome will be used to develop a machine that can recognize and count Indian coins, with Indian banks as the initial market.

Competence Analytics in Gaming

Gaming is a multi-billion dollar industry. Canada has a solid footing in the entertainment angle of the industry. There is a growing trend to customize gaming towards education and training. Albatross Gaming Inc. is interested in such a customization that relies on techniques from competency-based learning analytics. Recognizing the skills and strategies of gamers, at real time, and offering technological guidance to help achieve their personal goals would not only open a new market for the company in the world of gaming but also reinforce Canada’s stature as a leader in advanced gaming technologies. Recent advances in data analytics and sensor technologies have started to influence gaming and game-based learning towards adaptive and personalised environments. This project aims to build industry partner solutions to offer personalised learning/training experiences to gamers by tracking gaming interactions at higher levels of granularity and translating them into opportunities for custom feedback, reflection, and regulation.

Recurrent Neural Networks for Video Similarity Metrics and Video Captioning.

The objective of this project is to develop a system that provides video recommendations to the user based on the content of the video they are currently watching. Existing methods generally create their recommendations based on the title or tags of the video or on the viewing patterns of other users who have watched the same video. However, text based methods are only accurate if the titles and text are correctly spelled, use similar colloquial language, and accurately reflect the important content of the video, which are factors that are not guaranteed. Furthermore, it is often difficult to infer meaningful viewing patterns and determine similar user groups when attempting to create video recommendations based on the viewing behavior patterns of similar users, particularly if the user base is small. Hence, a more accurate system for video recommendations could be created through an artificial intelligence method that is able to autonomously infer what the important content of a video is. The method would then use this “important content” to find other videos that contain similar “important content” in order to create video recommendations for the user as well as describing this “important content” in a natural language sentence.