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

Receive updates for this collection

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.

Rapid Production of Nano-pixel-based Color Imageries

Author: 
Date created: 
2015-03-31
Abstract: 

Structural color pixels have attracted many interests as an alternate for colorant pigmentation. The structural coloration occurs as a result of light interaction with a physical structure containing nanometric features. In this study, novel techniques are presented to rapidly originate image master stamps for any given color image from a generic pixelated stamp named nanosubstrate. The nanosubstrate has gratings structure which gives colors due to light diffraction. Micro-patterning techniques were implemented to activate pixels of nanosubstrate and acquire image master stamps. The prepared master stamp can be replicated to provide several copies of image with the same quality. The given grating structure contains periodic hole arrays, which after replication convert to periodic pillars or vice versa. Both visible and invisible images can be easily stored and read using this method. This technique can be used for security and media applications, and can be complemented by combined information storage capabilities.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Bozena Kaminska
Department: 
Applied Sciences:
Thesis type: 
(Thesis) M.A.Sc.

Wavefront Sensorless Adaptive Optics Swept-Source Optical Coherence Tomography at 1060nm

Author: 
Date created: 
2015-12-23
Abstract: 

Optical Coherence Tomography (OCT) has revolutionized modern ophthalmology, providing depth resolved images of the retinal layers in a system that is suited to a clinical environment. A limitation on the performance and utilization of the OCT systems has been the lateral resolution; whereas the axial resolution is a function of the light source, the lateral resolution is dependent on the delivery optics. In this thesis, a compact lens based OCT system is presented that is capable of imaging the different retinal layers at a cellular lateral resolution with the combination of wavefront sensorless adaptive optics with dual variable optical elements. The central operating imaging wavelength of the wavelength swept OCT engine was 1060nm, close to the dispersion minimum of water (and the vitreous humor in the eye). A commercially available variable focal length lens is utilized to correct for a wide range of defocus commonly found in patient’s eyes, and a multi-actuator deformable lens for aberration correction to obtain near diffraction limited imaging at the retina. A parallel processing computational platform permitted real-time image acquisition and display. Cross-sectional images of the retinal layers and en face images of the cone photoreceptor mosaic acquired in vivo from research subjects are presented.

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

Applications of Reinforcement Learning to Routing and Virtualization in Computer Networks

Author: 
Date created: 
2016-03-17
Abstract: 

Computer networks and reinforcement learning algorithms have substantially advanced over the past decade. The Internet is a complex collection of inter-connected networks with a numerous of inter-operable technologies and protocols. Current trend to decouple the network intelligence from the network devices enabled by Software-Defined Networking (SDN) provides a centralized implementation of network intelligence. This offers great computational power and memory to network logic processing units where the network intelligence is implemented. Hence, reinforcement learning algorithms viable options for addressing a variety of computer networking challenges. In this dissertation, we propose two applications of reinforcement learning algorithms in computer networks.We first investigate the applications of reinforcement learning for deflection routing in buffer-less networks. Deflection routing is employed to ameliorate packet loss caused by contention in buffer-less architectures such as optical burst-switched (OBS) networks. We present a framework that introduces intelligence to deflection routing (iDef). The iDef framework decouples design of the signaling infrastructure from the underlying learning algorithm. It is implemented in the ns-3 network simulator and is made publicly available. We propose the predictive Q-learning deflection routing (PQDR) algorithm that enables path recovery and reselection, which improves the decision making ability of the node in high load conditions. We also introduce the Node Degree Dependent (NDD) signaling algorithm. The complexity of the algorithm only depends on the degree of the node that is NDD compliant while the complexity of the currently available reinforcement learning-based deflection routing algorithms depends on the size of the network. Therefore, NDD is better suited for larger networks. Simulation results show that NDD-based deflection routing algorithms scale well with the size of the network and outperform the existing algorithms. We also propose a feed-forward neural network (NN) and a feed-forward neural network with episodic updates (ENN). They employ a single hidden layer and update their weights using an associative learning algorithm. Current reinforcement learning-based deflection routing algorithms employ Q-learning, which does not efficiently utilize the received feedback signals. We introduce the NN and ENN decision-making algorithms to address the deficiency of Q-learning. The NN-based deflection routing algorithms achieve better results than Q-learning-based algorithms in networks with low to moderate loads.The second application of reinforcement learning that we consider in this dissertation is for modeling the Virtual Network Embedding (VNE) problem. We develop a VNE simulator (VNE-Sim) that is also made publicly available. We define a novel VNE objective function and prove its upper bound. We then formulate the VNE as a reinforcement learning problem using the Markov Decision Process (MDP) framework and then propose two algorithms (MaVEn-M and MaVEn-S) that employ Monte Carlo Tree Search (MCTS) for solving the VNE problem. In order to further improve the performance, we parallelize the algorithms by employing MCTS root parallelization. The advantage of the proposed algorithms is that, time permitting, they search for more profitable embeddings compared to the available algorithms that find only a single embedding solution. The simulation results show that proposed algorithms achieve superior performance.

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

Photo-controlled bio-mimicking dry adhesive

Author: 
Date created: 
2016-01-29
Abstract: 

The goal of this work is to develop a photo-switchable dry adhesive. Spiropyran doped PDMS polymer was moulded into biomimetic mushroom-shaped fibrillar adhesive microstructures characterized using a variety of measurement techniques and compared with a flat control surface made of the same material. Using UV light to generate of charged merocyanine molecules within ‘mushroom’-shaped micro-structured PDMS films enhanced the adhesion of the film to glass surfaces. The strength of the dry adhesive property can be lowered back to the original state using visible light. Quick and efficient switching in the polymer was observed. Integrating this molecule increased normal adhesion of unstructured samples by a factor of ~4 when polymer was in the neutral spiropyran form and ~5 for the merocyanine zwitterionic isomer, which demonstrated that control over the adhesion strength was possible. Surface charge and contact angle measurements further confirmed the proper functionality of the switch inside the PDMS polymer.

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

Model based and intelligent monitoring and control of lithium-ion batteries

Date created: 
2016-02-19
Abstract: 

Increased concerns over the limited sources of energy and environmental impact of the petroleum-based transportation infrastructure have led to increasing interest in an electric transportation infrastructure. Thus, electrical vehicles (including electric vehicle (EV), hybrid electric vehicle (HEV), and plug-in hybrid electric vehicle (PHEV)) and related issues have gained a great deal of attention. Battery technology and battery management is a key component in this regard and has indeed remained as a central challenge in vehicle electrification. This thesis deals with monitoring and control of Lithium ion batteries. The objective is to provide novel solutions to some of the challenging issues from a control theoretic perspective. The research stream in this thesis is headed towards three general directions, i.e. monitoring, diagnostics, and control. The proposed monitoring approaches are introduced as model-based and data-based approaches. The main objective in model-based approaches is to employ the high-fidelity physics-based models of the battery for monitoring. In this thesis, two particle-filtering methods are proposed for state, and joint state and parameter estimation of such models. The data based approaches try to come up with new ideas to monitor the battery accurately but with minimum computational load. In this regard, two different approaches are considered. A Takagi-Sugeno fuzzy model is developed for Li-ion battery where by the virtue of multiple-model structure of T-S model, the non-linearities of battery dynamics and corresponding parameters can appropriately be accounted for, while keeping the local models linear and easy-to-implement control/estimation algorithms. As a completely different alternative, the "Dynamic Resistance" concept is introduced that is sensitive to the battery state of charge and aging. This parameter considers changes in states of active materials in the cell during charge and discharge as well as overall interface resistances that may develop during cell aging. It can bring a new dimension to battery monitoring by providing a new easy-to-monitor parameter where the aging of the battery is also taken into account. This parameter is modeled versus the state of charge and total power throughput of the battery using a Group Method of Data Handling (GMDH) neural network and the model is used to monitor the state of charge and state of health of the battery. A reliable fault diagnosis system for batteries can play an important role in enhanced performance and reliability of electric-based transportation. In this thesis, the physics of the problem is rather comprehensively reviewed, and some of the proposed models for failure mechanism are presented and some fault-detection algorithms for some common failure mechanism are developed. Finally, over-charge/discharge of the cells within a battery pack can affect the battery's health significantly, and would pose serious safety concerns as well. Thus, a cell balancing circuit is usually employed in battery packs in order to keep all the cells in balance. In this thesis, the control problem of a cell-balancing circuit, which is essentially a switched hybrid system, is addressed in a model-based framework by proposing a nonlinear model predictive control (NMPC) strategy.

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

Safe motion planning under uncertainty for mobile manipulators in unknown environments

Date created: 
2015-12-08
Abstract: 

For a mobile manipulator to operate and perform useful tasks in human-centered environments, it is important to work toward the realization of robust motion planners that incorporate uncertainty inherent in robot's control and sensing and provide safe motion plans for reliable robot operation. Designing such planners pose a significant challenge because of computational complexity associated with mobile manipulator planning and planning under uncertainty. Current planning approaches for mobile manipulation are often conservative in nature and the uncertainty is largely ignored. In this thesis, we propose sampling-based efficient and robust mobile manipulator planners that use smart strategies to deal with computational complexity and incorporate uncertainty to generate safer plans. The first part of the research addresses the design of an efficient planner for deterministic case, where robot state is fully known, and then subsequent extension to incorporate base pose uncertainty. In the first part, we propose a Hierarchical and Adaptive Mobile Manipulator Planner (HAMP) that plans both for the base and the arm in a judicious manner - allowing the manipulator to change its configuration autonomously when needed if the current arm configuration is in collision with the environment as the mobile manipulator moves along the planned path. We show that HAMP is probabilistically complete. We then propose an extension of HAMP (HAMP-U) to account for localization uncertainty associated with the mobile base position. The advantages of our planners are illustrated and discussed. The second part of the research deals with the computational complexity involved in planning under uncertainty. For that, we propose localization aware sampling and connection strategies that help to reduce the planning time significantly with little compromise on the quality of path. In the third part, we learnt from the shortcomings of HAMP-U and took advantage of our smart strategies developed to combat the computational complexity. We propose an efficient and robust mobile manipulator planner (HAMP-BAU) that plans judiciously and considers the base pose uncertainty and the effects of this uncertainty on manipulator motions. It uses our localization aware sampling and connection strategies to consider only those nodes and edges which contribute toward better localization. This helps to find the same quality of path in shorter time. We also extend HAMP-BAU to incorporate task space constraints (HAMP-BAU-TC). Finally, in the last part of the work, we incorporate our planners (HAMP-BAU and HAMP-BAU-TC) within an integrated and fully autonomous system for mobile pick-and-place tasks in unknown static environments. We demonstrate our system both in simulation and real experiments on SFU mobile manipulator.

Document type: 
Thesis
File(s): 
HAMP demonstration corresponding to scenario B (simulation)
HAMP demonstration corresponding to scenario C (simulation)
HAMP demonstration corresponding to scenario D (simulation)
HAMP demonstration corresponding to scenario E (simulation)
HAMP-U demonstration on SFU mobile manipulator
HAMP-BAU demonstration using autonomous system for mobile pick-and-place task in unknown environment (simulation)
HAMP-BAU-TC demonstration using autonomous system for mobile pick-and-place task in known environment (simulation)
HAMP-BAU demonstration using autonomous system for mobile pick-and-place task in unknown environment (real experiment on SFU Mob
HAMP-BAU-TC demonstration using autonomous system for mobile pick-and-place task in known environment (real experiment on SFU Mo
Senior supervisor: 
Kamal Gupta
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) Ph.D.

Development of printed-circuit-board based industry-compatible point-of-care biosensing and bioprocessing technology with applications

Author: 
Date created: 
2015-11-03
Abstract: 

This thesis presents the development of a technology employing printed circuit board (PCB) technology to facilitate the performance and translation of point-of-care (POC) biosensing and bioprocessing devices toward practical products. Key features of the proposed technology are a universal, standardized platform and a set of techniques, featuring integrated functional units, three-dimensional (3D) configurations, convenient device-instrumentation interconnections, and industry-compatible precision manufacturing. The developed technology aims to incorporate and fabricate multiple functional units into a POC device with a compact configuration to perform bio/chemical sensing or processing that requires complex experimental conditions. In this thesis, PCB technology is employed to facilitate the development of three example biosensing and bioprocessing applications for proof of concept. First, the capability of using PCB substrates for complex assembly of functional components is demonstrated to facilitate the development of a chemistry-based enzyme assay. Proof-of-concept glucose-6-phosphate dehydrogenase (G6PD) deficiency assays are developed with integrated pH sensing units and temperature control units on boards. The assay is found to determine the G6PD level of a sample within 2 minutes. PCB technology is demonstrated to not only form an integrated platform but is also utilized in the fabrication of functional elements for biosensing and bioprocessing devices and systems. The second demonstrator is a molecule-based quantitative polymerase chain reaction (qPCR) device. A method is employed in this work to produce arrays of electrochemical biosensors and thermal cyclers using a three-metal PCB technology. The electrochemical performance and surface morphology of the biosensor microelectrodes are characterized and evaluated. The qPCR experiments are performed with 95% PCR efficiency and the detection limit of 59 deoxyribonucleic acid (DNA) copies. The third demonstrator is a cell-based on-board cooling rate controlled cryopreservation device. The possibility of meso-scale integration between the platform, sample storage and instrumentation is demonstrated in this work to facilitate the development of bioprocessing applications. On-board cooling-rate-controlled cryopreservation devices for use in low-temperature (-80°C) environments are developed with disposable, biocompatible polydimethylsiloxane (PDMS) storage chambers on top of localized feedback-controlled heaters on boards. These devices were able to maintain a stable cooling rate as low as 1°C per minute. Based on the work presented in this thesis, the future development plan and possible business models for the proposed technology are envisioned from academic and industrial perspectives to realize POC biosensing and bioprocessing applications toward commercialization.

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

Design and development of functional dry adhesives and their applications

Date created: 
2014-10-17
Abstract: 

Functional dry adhesives are dry adhesives that rely on dry adhesive structures for adhesion but also include additional functionality that enables adhesion switching or sensing capabilities. This thesis describes the design and testing of functional dry adhesives. Electro-dry-adhesives with flexible electrodes were fabricated. When a high voltage was applied to the flexible electrodes, fabricated from mixing and curing Carbon Black (CB) and polydimethylsiloxane (PDMS), an electrostatic field was generated between opposing electrodes and between the Electro-dry-adhesive and the surface it was attached to. The generated electrostatic field resulted in an increased shear adhesion force over the shear adhesion measured without the applied electrostatic field applied as well as the ability to self-preload. Magnetic field switchable dry adhesives were designed with a backing layer composed of iron oxide particles embedded within PDMS. The design of the dry adhesive backing layer allowed increased or decreased measured adhesion forces when the magnetic field was present during only the pull-off portion of the normal dry adhesion test cycle depending on the orientation of the magnetic field. Decreased adhesion was observed when the magnetic field was present during either the entire adhesion test cycle or when the magnetic field was present during only the preload portion of the dry adhesion test cycle regardless of the orientation of the magnetic field. Force and torque sensing dry adhesives were designed and fabricated by molding CB-PDMS. Force sensing was observed when the device was both compressed and extended by measuring a change in the resistance across the device terminals. Torque sensing was observed when the dry adhesive backing layer was twisted again by comparing resistance changes across the device terminals. The design of the force and torque sensing dry adhesives allowed the user to differentiate between forces in compression and extension as well as torques. Finally, a low cost method of fabricating dry adhesives was developed that utilizes commercially available meshes as a mold. The ability to utilize commercially available meshes instead of cleanroom fabrication techniques may save on overall fabrication costs and allow dry adhesives to be fabricated in large sheets.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Carlo Menon
Department: 
Applied Sciences:
Thesis type: 
(Thesis) Ph.D.

Head pose estimation and its application in TV viewers' behavior analysis

Author: 
Peer reviewed: 
No, item is not peer reviewed.
Date created: 
2015-12-04
Abstract: 

As a reliable indicator of visual gaze direction, head pose implies a person's visual attention and interest. Therefore, head pose information extracted from face images serves as important input in many applications. In this thesis, a coarse-to-fine head pose estimation method is proposed, by decomposing the original pose space in a hierarchical structure. The estimation begins with a coarse step to identify a subspace that encompasses a set of head pose candidates. Then a subsequent fine estimation is conducted within the subspace, generating a refined result. Besides, to eliminate irrelevant information within a face image, we propose to detect Region of Interest (ROI) by exploring importance degree of image points. Furthermore, we build an application of analyzing TV viewers' behaviors from video recordings, by integrating face detection, face tracking and head pose estimation. Based on head pose and face motion, a viewer's behavior is identified to be focused or unfocused.

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