Spatially-aware lighting estimation

We present a method for automatically estimating the lighting conditions from a single image. As opposed to most previous works which proposed methods that estimate only global lighting or use a limited illumination representation (low frequency SSH, parametric model), the proposed method attempt to use a new spatially-varying light representation with realist texture to render shiny objects. Our method will estimate a coarse (cuboid) geometry of an indoor scene from a single image and use this geometric information to hallucinate a realistic room texture used for illumination.

Microelectromechanical Low-power Strain Sensor for structural health monitoring applications – Phase 2

Structural health monitoring (SHM) of airplanes requires very compact and low-power stain sensors. Therefore, IPR wants to investigate how a commercial micro-fabrication process can be used to implement its MEMS sensor design, particularly using the electro-conductive properties of doped silicon vs. metal-coated crystalline silicon or polysilicon. The project will consist of a conceptual study of the current design provided by IPR and the design and evaluation through simulations of that design implemented in different technology.

Water Monitoring using Flocks of Flying Robots

This project aims at developing an automated system for water sampling along the coast of Nova Scotia using a swarm of flying robots. Currently, water is sampled by hand by teams of 2-5 people each on a number of boats, that must regularly travel to different sites, collect samples, and bring them to land for analysis. This task is slow and costly and exposes people to the inevitable dangers of the ocean.

Real-time Bus Routing and Traffic Prediction Via Machine Learning-based Methods

With the development of advance telecommunication systems, new opportunities for real-time public transport monitoring has been created. Traffic congestion in the vehicular ad-hoc network can be typically caused by an accident, construction zones, special events, and adverse weather. This research presents a cognitive framework to address real-time routing problem and and arrival time prediction for bus system using a machine learning method.

Millimeter-Wave Photonic Component Packaging and Interconnect

With the increasing demand for data rates in modern high-speed links come new requirements for the simulation environments that are used for their design. With optical modulator now achieving beyond-100-GHz large-signal modulation bandwidth in hybrid silicon photonics, the main challenges that such systems are currently facing is the lack of efficient interconnects to interface with the outside world. These interconnects are designed and optimized using full-wave simulations.

Weakly Supervised Behavioral Modeling for Controllable AI Agents in Video Games

The project aims at developing a new type of Reinforcement Learning algorithm that would allow to retain more control over the artificial agent once its training is completed. This framework would combine modern unsupervised modeling techniques to capture the variability of a set of demonstrations and user-defined programmatic functions that can characterise particularly important factors of variations. Ultimately, the user would be able to specify to the learned agent what type of behavior should be executed at test-time.

Fly-By-Wire INDI-Based Generic Control Laws for Flexible Civil Transport Aircraft: Enhanced Verification

New generation of civil transport aircraft presents interaction between flight mechanics and structural dynamics. Innovative CLaws have been developed to address this issue. They need to be verified thoroughly by high-fidelity simulations. For a research and development project, the traditional industrial verification process is too demanding and would be too time consuming. Indeed, each high-fidelity simulation is very slow.

RISC-V Vector Processor for High-throughput Multidimensional Sensor Data Processing & Machine Learning Acceleration at the Edge

The computing solutions of tomorrow must be more energy efficient than those of today, which requires combined efforts to be conducted on multiple research areas: from new transistor technologies to innovative software algorithms by way of original processor architectures. This project enters into this last research area by revisiting the vector processing model, which provides a highly efficient way of exploiting data parallelism in scientific computations, sensor processing, and machine learning algorithms.

Reconnaissance automatique des charges énergétiques et des événements pour une solution de domotique résidentielle.

L’intérêt pour la gestion intelligente de la consommation d’énergie résidentielle prend une force importante grâce à l’avancement de la domotique et la dissémination des technologies connectées facilitant la configuration et la gestion à distance des appareils. Ces progrès sont aussi supportés par le développement des nouvelles techniques de l’intelligence artificielle, la miniaturisation des systèmes embarqués et la réduction des coûts d’acquisition et d’installation des capteurs intelligents.

Anomaly detection to identify insider threats in the banks and companies

The project will identify and test combinations of AI (machine learning, neural network) algorithms for detection of suspicious activity (anomaly detection) on the IT network of a large bank or company. The project will benchmark which algorithms work best to identify instances of insider fraud based on high-volume continuous streaming event and content data from all endpoints (computers, laptops, tablets, phones) on a corporate IT network.