Thesis (Ph.D.) -- Mississippi State University. Department of Computer Science and Engineering. / Title from title screen. Includes bibliographical references.

L'objectif de cette thèse est d'améliorer ou de préserver les performances d'un système décisionnel quand l’environnement peut impacter certains attributs de l'espace de représentation à un instant donné ou en fonction de la position géographique de l’observation. S'inspirant des méthodes d'ensemble, notre approche a consisté à prendre les décisions dans des sous-espaces de représentation résultant de projections de l'espace initial, espérant ainsi travailler dans des sous-espaces non impactés. La décision finale est alors prise par fusion des décisions individuelles. Dans ce contexte, trois méthodes de classification (one-class SVM, Kernel PCA et Kernel ECA) ont été testées en segmentation d'images texturées qui constitue un support applicatif parfaitement adéquat en raison des ruptures de modèle de texture aux frontières entre deux régions. Ensuite, nous avons proposé une nouvelle règle de fusion reposant sur un test du rapport de vraisemblance pour un ensemble de classifieurs indépendants. Par rapport au vote majoritaire, cette règle de fusion a montré de meilleures performances face à l'altération de l'espace de représentation. Enfin, nous avons établi un modèle conjoint pour l’ensemble des variables décisionnelles de Bernoulli corrélées associées aux décisions des classifieurs individuels. Cette modélisation doit permettre de lier les performances des classifieurs individuels à la performance de la règle de décision globale et d’étudier et de maîtriser l'impact des changements de l'espace initial sur la performance globale / The objective of this thesis is to improve or maintain the performance of a decision-making system when the environment can impact some attributes of the feature space at a given time or depending on the geographical location of the observation. Inspired by ensemble methods, our approach has been to make decisions in representation sub-spaces resulting of projections of the initial space, expecting that most of the subspaces are not impacted. The final decision is then made by fusing the individual decisions. In this context, three classification methods (one-class SVM, Kernel PCA and Kernel ECA) were tested on a textured images segmentation problem which is a perfectly adequate application support because of texture pattern changes at the border between two regions. Then, we proposed a new fusion rule based on a likelihood ratio test for a set of independent classifiers. Compared to the majority vote, this fusion rule showed better performance against the alteration of the performance space. Finally, we modeled the decision system using a joint model for all decisions based on the assumption that decisions of individual classifiers follow a correlated Bernoulli law. This model is intended to link the performance of individual classifiers to the performance of the overall decision rule and to investigate and control the impact of changes in the original space on the overall performance

This diploma thesis deals with the detection of the current floor during elevator ride. This functionality is necessary for robot to move in multi-floor building. For this task, a fusion of accelerometric data during the ride of the elevator and image data obtained from the information display inside the elevator cabin is used. The research describes the already implemented solutions, data fusion methods and image classification options. Based on this part, suitable approaches for solving the problem were proposed. First, datasets from different types of elevator cabins were obtained. An algorithm for working with data from the accelerometric sensor was developed. A convolutional neural network, which was used to classify image data from displays, was selected and trained. Subsequently, the data fusion method was implemented. The individual parts were tested and evaluated. Based on their evaluation, integration into one functional system was performed. System was successfully verified and tested. Result of detection during the ride in different elevators was 97%.

This dissertation explores data fusion methodology to deduce an overall inference from the data gathered from multiple heterogeneous sources. Typically, if there existed a data source in which the data were reliable and unbiased, then data fusion would not be necessary. Data fusion methodology combines data form multiple diverse sources so that the desired information - such as the population mean - is improved despite redundancies, inaccuracies, biases, and inflated variability in the data. Examples of data fusion include estimating average demand from similar sources, and integrating fatality counts from different media sources after a catastrophe. The approach in this study combines "inputs" from distinct sources so that the information is "fused." Another way of describing this process is "data integration." Important assumptions are 1. Several sources provide "inputs" for information used to estimate parameters of a probability distribution. 2. Since distributions for the data from the sources are heterogeneous, some sources are less reliable. 3. Distortions, bias, censorship, and systematic errors may be more prominent in data from certain sources. 4. The sample size of sources data, number of "inputs," may be very small. Examples of information from multiple sources are abundant: traffic information from sensors at intersections, multiple economic indicators from various sources, demand data for product using similar retail stores as sources, polling data from various sources, and disaster count of fatalities from different media sources after a catastrophic event. This dissertation seeks to address a gap in the operations literature by addressing three research questions regarding entropy base data fusion (EBDF) approaches to estimation. Three separate, but unifying, essays address the research questions for this dissertation. Essay 1 provides an overview of supporting literature for the research questions. A numerical analysis of airline maximum wait time data illustrates the underlying issues involved in EBDF methods. This essay addresses the research question: Why consider alternative entropy-based weighting methods? Essay 2 introduces 13 data fusion methods. A Monte Carlo simulation study examines the performance of these methods in estimating the mean parameter of a population with either a normal or lognormal distribution. This essay addresses the following research questions: 1. Can an alternative formulation for Shannon's entropy enhance the performance of Sheu (2010)'s data fusion approach? 2. Do symmetric and skewed distributions affect the 13 data fusion methods differently? 3. Do negative and positive biases affect the performance of the 13 methods differently? 4. Do entropy based data fusion methods outperform non-entropy based data fusion methods? 5. Which data fusion methods are recommended for symmetric and skewed data sets when no bias is present? What is the recommendation under conditions of few data sources? Essay 3 explores the use of the data fusion method estimates of the population mean in a newsvendor problem. A Monte Carlo simulation study investigates the accuracy of the using the estimates provided in Essay 2 as the parameter estimate for the distribution of demand that follows an exponential distribution. This essay addresses the following research questions: 1. Do data fusion methods with relatively strong performance in estimating the parameter mean estimate also provide relatively strong performance in estimating the optimal demand under a given ratio of overage and underage costs? 2. Do any of the data fusion methods deteriorate or improve with the introduction of positive and negative bias? 3. Do the alternative entropy formulations to Shannon's entropy enhance the performance of the methods on a relative basis? 4. Is the relative rank ordering performance of the data fusion methods different in Essay 2 and Essay 3 in the resulting performances of the methods? The contribution of this research is to introduce alternative EBDF methods, and to establish a framework for using EBDF methods in supply chain decision making. A comparative Monte Carlo simulation analysis study will provide a basis to investigate the robustness of the proposed data fusion methods for estimation of population parameters in a newsvendor problem with known distribution, but unknown parameter. A sensitivity analysis is conducted to determine the effect of multiple sources, sample size, and distributions.

Thesis (MScEng)--Stellenbosch University, 2013. / ENGLISH ABSTRACT: This thesis describes the development of a distributed control system for SSL soccer robots. The project continues on work done to develop a robotics research platform at Stellenbosch University. The wireless communication system is implemented using Player middleware. This enables high level programming of the robot drivers and communication clients, resulting in an easily modifiable system. The system is developed to be used as either a centralised or decentralised control system. The software of the robot’s motor controller unit is updated to ensure optimal movement. Slippage of the robot’s wheels restricts the robot’s movement capabilities. Trajectory tracking software is developed to ensure that the robot follows the desired trajectory while operating within its physical limits. The distributed control architecture reduces the robots dependency on the wireless network and the off-field computer. The robots are given some autonomy by integrating the navigation and control on the robot self. Kalman filters are designed to estimate the robots translational and rotational velocities. The Kalman filters fuse vision data from an overhead vision system with inertial measurements of an on-board IMU. This ensures reliable and accurate position, orientation and velocity information on the robot. Test results show an improvement in the controller performance as a result of the proposed system. / AFRIKAANSE OPSOMMING: Hierdie tesis beskryf die ontwikkeling van ’n verspreidebeheerstelsel vir SSL sokker robotte. Die projek gaan voort op vorige werk wat gedoen is om ’n robotika navorsingsplatform aan die Universiteit van Stellenbosch te ontwikkel. Die kommunikasiestelsel is geïmplementeer met behulp van Player middelware. Dit stel die robotbeheerders en kommunikasiekliënte in staat om in hoë vlak tale geprogrameer te word. Dit lei tot ’n maklik veranderbare stelsel. Die stelsel is so ontwikkel dat dit gebruik kan word as óf ’n gesentraliseerde of verspreidebeheerstelsel. Die sagteware van die motorbeheer eenheid is opgedateer om optimale robot beweging te verseker. As die robot se wiele gly beperk dit die robot se bewegingsvermoëns. Trajekvolgings sagteware is ontwikkel om te verseker dat die robot die gewenste pad volg, terwyl dit binne sy fisiese operasionele grense bly. Die verspreibeheerargitektuur verminder die robot se afhanklikheid op die kommunikasienetwerk en die sentrale rekenaar. Die robot is ’n mate van outonomie gegee deur die integrasie van die navigasie en beheer op die robot self te doen. Kalman filters is ontwerp om die robot se translasie en rotasie snelhede te beraam. Die Kalman filters kombineer visuele data van ’n oorhoofse visiestelsel met inertia metings van ’n IMU op die robot. Dit verseker betroubare en akkurate posisie, oriëntasie en snelheids inligting. Toetsresultate toon ’n verbetering in die beheervermoë as ’n gevolg van die voorgestelde stelsel.

Cette thèse porte sur l'étude des techniques de fusion de données réparties et incertaines au sein d’un réseau de véhicules pour gérer la confiance dans les autres véhicules ou dans les données reçues. L'algorithme de fusion distribuée proposé est basé sur les fonctions de croyance et est appliqué par chaque nœud à la réception des messages. In se base sur la gestion d'une connaissance directe, locale à chaque nœud et d'une connaissance distribuée diffusée dans le réseau. Cette dernière résulte de la fusion des messages par un opérateur adapté prenant en compte les cycles éventuels et limitant l'effet de "data incest". Chaque nœud peut être autonome pour estimer la confiance mais la coopération entre les véhicules permet d'améliorer et de rendre plus robuste cette estimation. L'algorithme peut être adapté au cas d'étude en considérant un ou plusieurs éléments d'observation et en prenant en compte l'obsolescence des données. Lorsqu'il y a plusieurs éléments d'observation, se pose le problème de l'association de données nécessaire avant l'étape de combinaison. Un nouvel algorithme d'association a été formalisé dans le cadre des fonctions de croyance. Il a été démontré que ce problème est équivalent à un problème d'affectation linéaire, qui peut être résolu en temps polynomial. Cette solution est à la fois optimale et beaucoup plus efficace que d'autres approches développées dans ce formalisme. La gestion de la confiance dans les nœuds et dans les données échangées ont été illustrées par la mise en œuvre de deux applications : la détection de faux nœuds dans une attaque Sybil et la gestion de la confiance dans les cartes dynamiques pour la perception augmentée. / This thesis focuses on the study of fusion techniques for distributed and uncertain data in a vehicle network in order to manage the confidence in other vehicles or in received data. The proposed distributed fusion algorithm is based on belief functions and is applied by each node when it receives messages. It is based on the management of direct knowledge, local for each node, and the management of a distributed knowledge broadcasted over the network. The distributed knowledge is the result of the fusion of messages by a suitable operator taking into account the possible cycles and limiting the effect of "data incest". Each node can be autonomous to estimate confidence but cooperation between vehicles can improve and make more robust this estimation. The algorithm can be adapted to the case of study by considering one or more elements of observation and taking into account the data obsolescence. When there are multiple elements of observation, the data association is necessary before the combination step. A new association algorithm was formalized in the framework of belief functions.It has been shown that this problem is equivalent to a linear assignment problem which can be solved in polynomial time. This solution is both optimal and more effective than other approaches developed in this formalism. The confidence management in the nodes and in the received data were illustrated by the implementation of two applications : the detection of false nodes in a Sybil attack and the distributed dynamic maps for enhanced perception

This dissertation investigates novel middleware mechanisms for building streaming applications. Developing streaming applications is a challenging task because (i) they are continuous in nature; (ii) they require fusion of data coming from multiple sources to derive higher level information; (iii) they require efficient transport of data from/to distributed sources and sinks; (iv) they need access to heterogeneous resources spanning sensor networks and high performance computing; and (v) they are time critical in nature. My thesis is that an intuitive programming abstraction will make it easier to build dynamic, distributed, and ubiquitous data streaming applications. Moreover, such an abstraction will enable an efficient allocation of shared and heterogeneous computational resources thereby making it easier for domain experts to build these applications. In support of the thesis, I present a novel programming abstraction, called DFuse, that makes it easier to develop these applications. A domain expert only needs to specify the input and output connections to fusion channels, and the fusion functions. The subsystems developed in this dissertation take care of instantiating the application, allocating resources for the application (via the scheduling heuristic developed in this dissertation) and dynamically managing the resources (via the dynamic scheduling algorithm presented in this dissertation). Through extensive performance evaluation, I demonstrate that the resources are allocated efficiently to optimize the throughput and latency constraints of an application.

Thesis (Ph.D)--Electrical and Computer Engineering, Georgia Institute of Technology, 2009. / Committee Chair: Rusell M. Mersereau; Committee Member: Biing Hwang (Fred) Juang; Committee Member: Christopher E. Heil; Committee Member: Georgia Vachtsevanos; Committee Member: James H. McClellan. Part of the SMARTech Electronic Thesis and Dissertation Collection.

The objective of this research is to demonstrate how an underlying system's state vector distribution can be determined in a distributed heterogeneous sensor network with reduced subspace observability at the individual nodes. We show how the network, as a whole, is capable of observing the target state vector even if the individual nodes are not capable of observing it locally. The initialization algorithm presented in this work can generate the initial state vector distribution for networks with a variety of sensor types as long as the measurements at the individual nodes are known functions of the target state vector. Initialization is accomplished through a novel distributed implementation of the particle filter that involves serial particle proposal and weighting strategies, which can be accomplished without sharing raw data between individual nodes in the network. The algorithm is capable of handling missed detections and clutter as well as compensating for delays introduced by processing, communication and finite signal propagation velocities. If multiple events of interest occur, their individual states can be initialized simultaneously without requiring explicit data association across nodes. The resulting distributions can be used to initialize a variety of distributed joint tracking algorithms. In such applications, the initialization algorithm can initialize additional target tracks as targets come and go during the operation of the system with multiple targets under track.

Ubiquitous computing application developers have limited options for a practical activity and location sensing technology that is easy-to-deploy and cost-effective. In this dissertation, I have developed a class of activity monitoring systems called infrastructure mediated sensing (IMS), which provides a whole-house solution for sensing activity and the location of people and objects. Infrastructure mediated sensing leverages existing home infrastructure (e.g, electrical systems, air conditioning systems, etc.) to mediate the transduction of events. In these systems, infrastructure activity is used as a proxy for a human activity involving the infrastructure. A primary goal of this type of system is to reduce economic, aesthetic, installation, and maintenance barriers to adoption by reducing the cost and complexity of deploying and maintaining the activity sensing hardware. I discuss the design, development, and applications of various IMS-based activity and location sensing technologies that leverage the following existing infrastructures: wireless Bluetooth signals, power lines, and central heating, ventilation, and air conditioning (HVAC) systems. In addition, I show how these technologies facilitate automatic and unobtrusive sensing and data collection for researchers or application developers interested in conducting large-scale in-situ location-based studies in the home.