Engineering Science - Theses, Dissertations, and other Required Graduate Degree Essays

Receive updates for this collection

Visual-Inertial Sensor Fusion for Tracking in Ambulatory Environments

Date created: 
2016-08-22
Abstract: 

Tracking a high-velocity object through a cluttered environment is daunting for even the human-observer. Vision-based trackers will frequently lose their lock on the object as features on the object become distorted or faded as a result of motion-blurring imparted by the high-velocity of the object. Moreover, the frequent occlusions as the object passes through clutter only serves to compound the issue. To boot, the usual difficulties associated with most vision-based trackers still apply such as: nonuniform illumination, object rotation, object scale changes, etc... Inertial-based trackers provide useful complementary data to aid the vision-based systems. The higher sampling rates of the inertial measurements gives invaluable information to be able to track high-speed objects. With the IMU attached to the object, the inertial measurements are immune to occlusions unlike their visual counterparts. Efficient combination of visual as well as inertial sensors into a unified framework is coined visual-inertial sensor fusion. Visual-inertial sensor fusion is a powerful tool for many industries: it allows the medical practitioners to better understand and diagnose illnesses; it allows the engineer to design more flexible and immersive virtual reality environments; and it allows the film-director to fully capture motion in a scene. The complementary nature of visual and inertial sensors is well-toted throughout these industries, the faster sampling rate of the inertial sensors fits lock-and-key with the higher accuracy of the visual sensor to unlock the potential for algorithms capable of tracking high-velocity objects through cluttered environments. Inevitably, sensor fusion is accompanied by higher algorithmic complexity and requires careful understanding of the components involved. For this reason, the approach taken in this thesis is a ground-up approach towards a complete visual-inertial system: from camera calibration all the way to handling of asynchronous sensor measurements for sensor-fusion.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Ivan Bajic
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) M.A.Sc.

Transmission of JPEG2000 Images over Cognitive Radio Networks

Author: 
Date created: 
2016-08-12
Abstract: 

Cognitive Radio (CR) is an efficient way of spectrum utilization, because secondary users (SUs) with bandwidth-demanding applications such as multimedia can get access to licensed frequency resources opportunistically and resolve their bandwidth limitations. Among all multimedia formats, JPEG2000 is a suitable candidate for cognitive radio networks thanks to its unique features. In conventional resource allocations for CR systems, all data bits are assumed equally important. However, different parts of the JPEG2000 bit stream have different contributions to the quality of the received image. Therefore, in this thesis, an unequal power allocation method is used to allocate the available power to the coded bits based on their importance in the image quality. Furthermore, bits with higher significance are further protected by using sub-channels with better channel quality. Thus, the likelihood of significant bits being received correctly is increased. The optimal solution is obtained by minimizing the image distortion without violating the interference requirement of the primary users (PU). The performance of the proposed method is demonstrated by simulation results.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Jie Liang
Atousa Hajshirmohammadi
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) M.A.Sc.

A Context-Aware Model for Dynamic Adaptability of Software for Embedded Systems

Date created: 
2016-08-08
Abstract: 

The scientific contributions of this thesis are three-fold. Firstly, a novel specialized embedded systems software architecture for contextawareness is presented. This architecture is developed for use on a resource constrained hardware platform and is low latency. For firmware applications with many sources of context, a specialized architecture is important to achieve code readability, modularity, extensibility and maintainability. Context in embedded systems firmware development is defined as changeable and characterizing information such as sensor data (IR - infrared, GPS, accelerometer) or profile attributes (user, vehicle, device, etc.). A second focus was on dynamic architecture adaptability in the form of a cognitive engine which processes real-time updates to its user-configurable module. Dynamic adaptability improves the application software's flexibility and responsiveness according to different user requirements or varying operational conditions. Adaptability is defined as system changes according to changes in context and in terms of the four W's - why are there changes, what remains unchanged, when do the changes occur and who manages these changes. Thirdly, the concept of context-aware map logic (CAML) is introduced. Cognitive engine updates are performed using these logic maps which are derived from/inspired by fuzzy cognitive maps (FCM) and GPS (global positioning system) coverage maps. The logic maps feature phi, delta, timer, complement, latched and momentary operands. The logic maps were specifically designed for resource constrained hardware. No previous work has been done on the use of fuzzy cognitive maps specifically with linguistic weights for enabling dynamic, resource constrained firmware adaptability. Fuzzy cognitive maps are at the intersection of fuzzy logic and neural networks. A resource constrained hardware platform is defined as a single-processor microcontroller with low processing power and limited memory space as compared against large memory, multi-core, multi-media processors e.g. cell-phones. The targeted hardware platform could be a legacy processor or a low power processor typically found in wireless sensor networks or energy-aware or cost-aware solutions. Context-awareness is an important topic in the wireless sensor networks research field. Wireless sensor networks comprise wirelessly enabled embedded systems for data acquisition and control for a wide array of applications. In this thesis context is defined as changeable and characterizing information such as sensor data, profile attributes or explicitly provided user information. The embedded systems software architecture is a layered model with context and cognitive planes which focus on dynamic adaptability. The context plane features a microarchitecture, which includes context collectors, context controllers and a context task based coordinator. The cognitive plane is responsible for dynamic adaptable logic reconfiguration inspired by fuzzy cognitive maps. Proof-of-concept firmware was developed for a wireless physiological sensor showing context collector implementation. An ATE (automatic test equipment) test architecture was also developed for the sensor highlighting architecture development and providing the groundwork for the context controller development. The lead-up to the cognitive engine is explored in an introduction to fuzzy cognitive maps, its implementations and applications to current research. An industrial application, Novax's Accessible Pedestrian System (APS) and simulations using the Rapita suite of tools are presented.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Bozena Kaminska
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) Ph.D.

Enhanced Digital Imager Defect Analysis with Smaller Pixel Sizes

Date created: 
2016-08-03
Abstract: 

Digital imager pixels are shrinking resulting in an increased rate of pixel defects. These defects are known as “Hot Pixels” that are permanent in nature and develop in-field. The number of hot pixels in a given digital imager increases over time. This research experimentally measures defect rates for pixels from 7 μm to those in the cellphone camera range, as low as 1μm. New software algorithms and techniques have been developed to compensate for increasing noise levels in the 2 to 1 μm range. This has allowed the creation of an empirical model that provides accurate projections of defect rates as pixel size decreases and sensitivity increases. Results show that the hot pixel rate increases by 8.9 times as pixels shrink by a factor of 2. Additionally, digital imagers allow us to explore soft errors (known as single event upsets) in a way that can’t be done in traditional ICs.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Glenn Chapman
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) M.A.Sc.

Atomistic and Continuum Modelling of Strength and Adhesion of Graphene

Date created: 
2016-04-07
Abstract: 

Exceptional electromechanical properties of graphene (a single layer of graphite) have been used to develop the next generation of nanodevices. These state-of-the-art nanodevices, such as sensors and transistors, have a profound impact in numerous engineering disciplines, ranging from biomedicine to aerospace. Recent experiments show that graphene could also be used as an ultra-strong reinforcement for composite materials. In both nanodevices and composite materials, graphene is in contact with adjoining materials, creating mechanically weak interfaces between them. Therefore, understanding the mechanical behaviour of both graphene and graphene interfaces is critically important in designing reliable graphene-based systems. In this thesis, molecular dynamics simulation studies are conducted to gain a basic understanding of the mechanics of graphene-based systems. Then, based on this knowledge, computationally efficient continuum-based models are developed in order to further investigate the strength and adhesion of nanoscale systems. The continuum-based models are accurate and around one million times faster than the molecular dynamics simulations. In addition, using the concepts of kinetic analysis, an analytical model is developed to estimate the strength of defective graphene. Finally, a nonlinear spring model is developed to characterize the adhesion properties of defective graphene interfaces. Results show that defects and temperature significantly reduce the strength of graphene. Low concentrations of hydrogen adatoms degrade the interfacial adhesion of graphene interfaces, and highly hydrogen functionalized graphene completely loses its strength when subjected to higher temperatures. It is also found that molecular dynamics simulations, conducted at elevated temperatures and high strain rates, significantly over predict the strength. Furthermore, the study reveal that graphene with vacancy defects shows a singular stress field as in continuum fracture mechanics and moderate amount of lattice trapping prevails in graphene.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Nimal Rajapakse
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Dissertation) Ph.D.

Distortion Estimation and Graph-based Transform for Visual Communications

Author: 
Date created: 
2016-04-29
Abstract: 

In this thesis, we study several visual communications problems, including joint source-channel coding for single view video transmission, transmission distortion estimation for multiview video coding, and depth video coding for multiview video applications. The first contribution in this thesis is the design and implementation of an error-resilient video conferencing system. We first develop an algorithm to estimate the decoder-side distortion in the presence of packet loss. We then design a family of very short systematic forward error correction (FEC) codes to recover lost packets. Finally, FEC codes are dynamically optimized to minimize the distortion from packet loss. The proposed scheme is demonstrated on a real-time embedded video conferencing system. A similar joint source channel coding framework can also be applied to multiview video coding applications such as free-viewpoint TV. Therefore an algorithm is needed for the encoder to estimate the distortion of the synthesized virtual view. We first derive a graphical model to analyze how random errors in the reference depth image affect the synthesized virtual view. We then consider the case where packet loss occurs in both the encoded texture and depth images during transmission, and develop a recursive algorithm to calculate the pixel level texture and depth probability distributions in the reference views. The recursive algorithm is then integrated with the graphical model method to estimate the distortion in the synthesized view. The graph-based transform has been extensively used for depth image coding in multiview video applications. In this thesis, we aim to develop a single graph-based transform for a class of depth signals. We first propose a 2-D first-order autoregression (2-D AR1) model and a 2-D graph to analyze depth signals with deterministic discontinuities. We show that the inverse of the biased Laplacian matrix of the proposed 2-D graph is exactly the covariance matrix of the proposed 2-D AR1 model. Therefore the optimal transform are the eigenvectors of the proposed graph Laplacian. Next, we show that similar results hold when the locations of the discontinuities are randomly distributed within a confined region. The theory in this thesis can be used to design both pre-computed and signal-dependent transforms.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Jie Liang
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) Ph.D.

A customizable, scalable control solution for digitally-based reconfigurable magnetic microfluidic systems

Author: 
Date created: 
2015-10-06
Abstract: 

Recent trends in microfluidic technologies are leaning to more streamlined and integratedplatforms that can perform a variety of tasks. In order to achieve this, a continuous-flowintegrated microfluidics system needs to be made portable through the use of componentsthat are digitally controllable. The proposed device will use magnetic-based microfluidicscomponents, such as valves and mixers, which will require an electromagnetic based modelof actuation.The scope of this thesis is to design and optimize an FPGA-based control system comprisedof a user interface, device libraries and circuitry to connect to the physical components.Particular focus is given to optimizing the actuation system for magnetic microvalves toensure power efficiency, a trait that is paramount for a portable device such as the proposedmicrofluidics platform. Theoretical models and simulations are evaluated throughexperimentation to determine which best correlate with the physical system. This enablesthe selection of a set of parameters that result in a power-efficient actuation system. Thesimulations and evaluations are used to define a procedure for parameter selection.The selection criteria for these parameters are evaluated for an example system and theresulting actuation system behaves as predicted in a physical demonstration. The actuationsystem is integrated with the user interface through a software framework designed to bemodular, scalable and easy to upgrade.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Lesley Shannon
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) M.A.Sc.

Robo-Wrist Controller

Date created: 
2015-04-23
Abstract: 

Robo-wrist controller is a Graphical User Interface (GUI) based application designed and developed to control, and monitor a wrist exoskeleton driven by micro-controllers. The software application is developed to better assist a physiotherapist in administering physical therapy to stroke patients with the help of the exoskeleton. Since the micro-controller device that directly controls the exoskeleton is not very intuitive to use and lacks visual feedback, a software application fills these gaps and makes the device more user-friendly. It also provides the benefits of a wireless controller. The software application comprises two major aspects: an Android application for use in smartphones and a Windows PC version. The major differences between these two are the number of devices that can be controlled simultaneously and the communication media. The android application can control at most one device at any given time via Bluetooth, while the PC version is capable of controlling multiple devices simultaneously because it uses both Bluetooth and Wi-Fi communication media. Both types of software applications have features such as: ability to add new devices, view the added devices, create an exercise protocol and store it, view stored protocols, a data visualizer to plot incoming feedback data from the exoskeleton, and multiple user profiles with different levels of administrative privileges. Each module in the applications is independent of each other as well allowing for easy modifications and additions to the application in the future. The GUI of the software makes it very easy for the therapist to use the exoskeleton using visual controls and feedback allowing for easy administering of physical therapy using the exoskeleton to his/ her patients.

Document type: 
Graduating extended essay / Research project
File(s): 
Senior supervisor: 
Carlo Menon
Fabio Campi
Department: 
Applied Sciences:
Thesis type: 
(Project) M.Eng.

Implementation of a Wearable Feedback System Monitoring the Activities of Upper-extremities

Author: 
Date created: 
2015-04-21
Abstract: 

In the previous work done by ZG.Xiao and C.Menon, the novel idea of using a strap with 8 Force Sensing Resistor (FSR) sensors for monitoring activities of upper-extremities was proposed. The goal of my research is to implement such a system for a low-cost, low power embedded system, a band module, together with a hand-held user interface for rehabilitation related application. For hand gesture classification, the Linear Discriminate Analysis (LDA) method is used. The training and predicting can be done on either a band module or a hand-held user interface. Two system configurations are proposed: real-time band module data sampling and hand-held user interface data analysis, or real-time band module data sampling and analysis. On top of this, the LDA algorithm in C language used in our system has been profiled on an Intel Galileo Gen2 board in order to evaluate its performance on 32-bit embedded platforms for the next generation of our system.

Document type: 
Graduating extended essay / Research project
File(s): 
Senior supervisor: 
Carlo Menon
Department: 
Applied Sciences:
Thesis type: 
(Project) M.Eng.

Faster Tuberculosis Diagnosis with Low Cost Public Health Devices

Date created: 
2015-04-07
Abstract: 

Tuberculosis (TB) is a deadly infectious disease that usually affects the lungs, but may affect almost any tissue of the body. The primary symptoms of TB include cough of more than three weeks duration, weight loss, fever, blood-stained sputum, chest pain, and if left untreated, can result in death. In 2013 alone, nine million people fell ill with TB, and one point five million people died from the disease, making TB the second greatest killer worldwide as a single infectious agent. TB is both preventable by implementing appropriate infection control measures, and is treatable through the use of specific antibacterial drugs. A common method for detecting active tuberculosis and measuring its bacillary load is through the use of acid fast staining, such as Ziehl Neelsen (ZN) staining, which involves applying specific chemicals to a sputum sample or other smear from a patient, and analyzing the smear under a microscope. In the case of ZN staining, TB bacteria appear as pink rod-shaped objects against a bright blue background. By counting the number of bacteria visible in the stained smear, the examiner can deduce approximately how advanced the TB infection is, and act upon it appropriately. At this time, it is the norm for the bacteria to be counted manually by lab technicians working in TB clinics. The goal of this thesis is to identify the problem that is TB, and explain a set of methods and devices which can be used to further reduce the global impact of TB. Using a combination of image processing methods and image acquisition devices combined with microscopy, we are able to rapidly count the number of bacteria present in a ZN stained slide. The result is a portable, low cost, hand-held device that can perform hours of manual analysis in under a minute, while providing ivgreater consistency across individual tests. This has the potential to decrease the lead time on diagnosis and treatment of TB in the field from weeks to minutes, helping to reduce the impact of this deadly disease.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Dr. Mirza Faisal Beg
Dr. Pankaj Sadaphal
Department: 
Applied Sciences:
Thesis type: 
(Thesis) M.A.Sc.