This work describes a framework for simultaneous estimation and modeling (SEAM) of dynamic systems using non-Gaussian belief fusion by first presenting the relevant fundamental formulations, then building upon these formulations incrementally towards a more general and ubiquitous framework. Multi-Gaussian belief fusion (MBF) is introduced as a natural and effective method of fusing non-Gaussian probability distribution functions (PDFs) in arbitrary dimensions efficiently and with no loss of accuracy. Construction of some multi-Gaussian structures for potential use in MBF is addressed. Furthermore, recursive Bayesian estimation (RBE) is developed for linearized systems with uncertainty in model parameters, and a rudimentary motion model correction stage is introduced. A subsequent improvement to motion model correction for arbitrarily non-Gaussian belief is developed, followed by application to observation models. Finally, SEAM is generalized to fully nonlinear and non-Gaussian systems. Several parametric studies were performed on simulated experiments in order to assess the various dependencies of the SEAM framework and validate its effectiveness in both estimation and modeling. The results of these studies show that SEAM is capable of improving estimation when uncertainty is present in motion and observation models as compared to existing methods. Furthermore, uncertainty in model parameters is consistently reduced as these parameters are updated throughout the estimation process. SEAM and its constituents have potential uses in robotics, target tracking and localization, state estimation, and more. / Doctor of Philosophy / The simultaneous estimation and modeling (SEAM) framework and its constituents described in this dissertation aim to improve estimation of signals where significant uncertainty would normally introduce error. Such signals could be electrical (e.g. voltages, currents, etc.), mechanical (e.g. accelerations, forces, etc.), or the like. Estimation is accomplished by addressing the problem probabilistically through information fusion. The proposed techniques not only improve state estimation, but also effectively "learn" about the system of interest in order to further refine estimation. Potential uses of such methods could be found in search-and-rescue robotics, robust control algorithms, and the like. The proposed framework is well-suited for any context where traditional estimation methods have difficulty handling heightened uncertainty.

Undoubtedly, state-of-the-art unit selection-based concatenative speech systems produce very high quality synthetic speech. this is due to a large speech database containing many instances of each speech unit, with a varied and natural distribution of prosodic and spectral characteristics. the join cost, which measures how well two units can be joined together is one of the main criteria for selecting appropriate units from this large speech database. The ideal join cost is one that measures percieved discontinuity based on easily measurable spectral properties of the units being joined, inorder to ensure smooth and natural sounding synthetic speech. During first part of my research, I have investigated various spectrally based distance measures for use in computation of the join cost by designing a perceptual listening experiment. A variation to the usual perceptual test paradigm is proposed in this thesis by deliberately including a wide range of qualities of join in polysyllabic words. The test stimuli are obtained using a state-of-the-art unit-selection text-to-speech system: rVoice from Rhetorical Systems Ltd. Three spectral features Mel-frequency cepstral coefficients (MFCC), line spectral frequencies (LSF) and multiple centroid analysis (MCA) parameters and various statistical distances - Euclidean, Kullback-Leibler, Mahalanobis - are used to obtain distance measures. Based on the correlations between perceptual scores and these spectral distances. I proposed new spectral distance measures, which have good correlation with human perception to concatenation discontinuities. The second part of my research concentrates on combining join cost computation and the smoothing operation, which is required to disguise joins, by learning an underlying representation from the acoustic signal. In order to accomplish this task, I have chosen linear dynamic models (LDM), sometimes known as Kalman filters. Three different initialisation schemes are used prior to Expectation-Maximisation (KM) in LDM training. Once the models are trained, the join cost is computed based on the error between model predictions and actual observations. Analytical measures are derived based on the shape of this error plot. These measures and initialisation schemes are compared by computing correlations using the perceptual data. The LDMs are also able to smooth the observations which are then used to synthesise speech. To evaluate the LDM smoothing operation, another listening test is performed where it is compared with the standard methods (simple linear interpolation). I have compared the best three join cost functions, chosen from the first and second parts of my research, subjectively using a listening test in the third part of my research. in this test, I also evaluated different smoothing methods: no smoothing, linear smoothing and smoothing achieved using LDMs.

O objetivo deste trabalho é desenvolver a arquitetura de um Sistema de Alarme Inteligente visando à aplicação deste em um Equipamento de Monitorização Multiparamétrica de Pacientes que deverá estar em conformidade com normas internacionais de segurança de equipamentos eletromédicos, tema este que surgiu durante o projeto de um Equipamento de Monitorização Multiparamétrica de Pacientes desenvolvido na Intermed Equipamento Médico Hospitalar Ltda. Esta arquitetura propõe o uso de um método robusto para fusão de sensores através de Filtro de Kalman, como apresentado por Ebrahim, Feldman e Bar-Kana (1997), onde a partir de diferentes sensores fornecendo valores de freqüência cardíaca obtemos um valor ótimo da freqüência cardíaca livre de artefatos e mais preciso do que o disponível por qualquer sensor individualmente. Os sinais fisiológicos do paciente, juntamente com o valor ótimo da freqüência cardíaca, são então transformados em variáveis semânticas e analisados através de lógica fuzzy a fim de se identificar condições de alarme presentes no paciente. O sistema desenvolvido é detalhado para um cenário clínico, correspondendo a uma Unidade de Tratamento Intensivo Cardíaca, onde os principais parâmetros de configuração do sistema foram obtidos através de entrevistas com profissionais da área de saúde. O desenvolvimento do sistema foi focado no atendimento às Normas Internacionais aplicáveis mais atuais, que passaram a permitir Sistemas de Alarme Inteligente em 2003. O sistema foi simulado para o cenário clínico detalhado, onde foram analisados três pacientes em diferentes casos. Durante a simulação foi mostrada a influência de um valor mínimo do grau de coincidência para validação das condições de alarme, onde obtivemos valores adequados utilizando para isso um valor de 50%. Ao final mostramos também a importância de se levar em consideração a prioridade das condições de alarme no sistema proposto, o que além de ser um requisito de norma, influencia diretamente no comportamento do sistema, e conseqüentemente na resposta esperada pelo operador em função da mensagem a ele apresentada. / The objective of this work is the development of an Intelligent Alarm System architecture aiming at its application on a Multiparameter Patient Monitoring Equipment, which must be in compliance to international electrical medical equipment safety standards. This theme was raised during the project of a Multiparameter Patient Monitoring Equipment developed at Intermed Equipamento Médico Hospitalar Ltda. This architecture proposes a robust method for sensor fusion using a Kalman Filter as presented by Ebrahim, Feldman and Bar-Kana (1997), in which sensor measurements of heart rate are used to derive a predicted value for this parameter free of artifacts and more precise than any of the individual measurements. The patient vital signs are then, together with the predicted value of the heart rate, transformed into semantic variables and analyzed through fuzzy logic in order to identify alarm conditions. The proposed system is detailed for a clinic scenario, corresponding to a Cardiac Intensive Care Unit, where the main configuration parameters were obtained through interviews with health care professionals. The system development was focused on the compliance to International Standards at present day, which allow the use of Intelligent Alarm Systems since 2003. The system was simulated for the detailed clinic scenario with three patients in different cases. The simulations show the influence of a minimal grade of coincidence value for the validation of the alarm conditions, where we obtained good results using a value of 50%. At the end of this work, it is shown the importance of considering the alarm priority, which not only is a standard requirement, but also interferes directly on the system behavior, and consequently, on the operator´s expected response to the message.

Emoções exercem influência direta sobre a vida humana, mediando a maneira como os indivíduos interagem e se relacionam, seja em âmbito pessoal ou social. Por essas razões, o desenvolvimento de interfaces homem-máquina capazes de manter interações mais naturais e amigáveis com os seres humanos se torna importante. No desenvolvimento de robôs sociais, assunto tratado neste trabalho, a adequada interpretação do estado emocional dos indivíduos que interagem com os robôs é indispensável. Assim, este trabalho trata do desenvolvimento de um modelo matemático para o reconhecimento do estado emocional humano por meio de expressões faciais. Primeiramente, a face humana é detectada e rastreada por meio de um algoritmo; então, características descritivas são extraídas da mesma e são alimentadas no modelo de reconhecimento de estados emocionais desenvolvidos, que consiste de um classificador de emoções instantâneas, um filtro de Kalman e um classificador dinâmico de emoções, responsável por fornecer a saída final do modelo. O modelo é otimizado através de um algoritmo de têmpera simulada e é testado sobre diferentes bancos de dados relevantes, tendo seu desempenho medido para cada estado emocional considerado. / Emotions have direct influence on the human life and are of great importance in relationships and in the way interactions between individuals develop. Because of this, they are also important for the development of human-machine interfaces that aim to maintain natural and friendly interactions with its users. In the development of social robots, which this work aims for, a suitable interpretation of the emotional state of the person interacting with the social robot is indispensable. The focus of this work is the development of a mathematical model for recognizing emotional facial expressions in a sequence of frames. Firstly, a face tracker algorithm is used to find and keep track of a human face in images; then relevant information is extracted from this face and fed into the emotional state recognition model developed in this work, which consists of an instantaneous emotional expression classifier, a Kalman filter and a dynamic classifier, which gives the final output of the model. The model is optimized via a simulated annealing algorithm and is experimented on relevant datasets, having its performance measured for each of the considered emotional states.

The world went through weeks of financial turbulence in stock markets and investors were overcome by fears fuelled by more bad news, while countries continued their attempts to calm the markets with more injection of funds. By these very disturbed times, even if traders hope extreme risk aversion has passed, an investor would like predict the future of the market in order to protect his portfolio and a speculator would like to optimize his tradings. This thesis describes the design of numerical models and algorithms for the forecasting of financial time series, for speculation on a short time interval. To this aim, we will use two models: - " Price Forecasting Model " forecasts the behavior of an asset for an interval of three hours. This model is based on Functional Clustering and smoothing by cubic-splines in the training phase to build local Neural models, and Functional Classification for generalization, - " Model of Trading " forecasts the First Stopping time, when an asset crosses for the first time a threshold defined by the trader. This model combines a Price Forecasting Model for the prediction of market trend, and a Trading Recommendation for prediction of the first stopping time. We use an auto-adaptive Dynamic State Space Model, with Particle Filters and Kalman-Bucy Filters for parameter estimation.

This work presents a new approach that combines the comprehensive capabilities of the ensemble Kalman filter (EnKF) and the flow path information from streamlines to eliminate and/or reduce some of the problems and limitations of the use of the EnKF for history matching reservoir models. The recent use of the EnKF for data assimilation and assessment of uncertainties in future forecasts in reservoir engineering seems to be promising. EnKF provides ways of incorporating any type of production data or time lapse seismic information in an efficient way. However, the use of the EnKF in history matching comes with its shares of challenges and concerns. The overshooting of parameters leading to loss of geologic realism, possible increase in the material balance errors of the updated phase(s), and limitations associated with non-Gaussian permeability distribution are some of the most critical problems of the EnKF. The use of larger ensemble size may mitigate some of these problems but are prohibitively expensive in practice. We present a streamline-based conditioning technique that can be implemented with the EnKF to eliminate or reduce the magnitude of these problems, allowing for the use of a reduced ensemble size, thereby leading to significant savings in time during field scale implementation. Our approach involves no extra computational cost and is easy to implement. Additionally, the final history matched model tends to preserve most of the geological features of the initial geologic model. A quick look at the procedure is provided that enables the implementation of this approach into the current EnKF implementations. Our procedure uses the streamline path information to condition the covariance matrix in the Kalman Update. We demonstrate the power and utility of our approach with synthetic examples and a field case. Our result shows that using the conditioned technique presented in this thesis, the overshooting/undershooting problems disappears and the limitation to work with non- Gaussian distribution is reduced. Finally, an analysis of the scalability in a parallel implementation of our computer code is given.

O objetivo deste trabalho é desenvolver a arquitetura de um Sistema de Alarme Inteligente visando à aplicação deste em um Equipamento de Monitorização Multiparamétrica de Pacientes que deverá estar em conformidade com normas internacionais de segurança de equipamentos eletromédicos, tema este que surgiu durante o projeto de um Equipamento de Monitorização Multiparamétrica de Pacientes desenvolvido na Intermed Equipamento Médico Hospitalar Ltda. Esta arquitetura propõe o uso de um método robusto para fusão de sensores através de Filtro de Kalman, como apresentado por Ebrahim, Feldman e Bar-Kana (1997), onde a partir de diferentes sensores fornecendo valores de freqüência cardíaca obtemos um valor ótimo da freqüência cardíaca livre de artefatos e mais preciso do que o disponível por qualquer sensor individualmente. Os sinais fisiológicos do paciente, juntamente com o valor ótimo da freqüência cardíaca, são então transformados em variáveis semânticas e analisados através de lógica fuzzy a fim de se identificar condições de alarme presentes no paciente. O sistema desenvolvido é detalhado para um cenário clínico, correspondendo a uma Unidade de Tratamento Intensivo Cardíaca, onde os principais parâmetros de configuração do sistema foram obtidos através de entrevistas com profissionais da área de saúde. O desenvolvimento do sistema foi focado no atendimento às Normas Internacionais aplicáveis mais atuais, que passaram a permitir Sistemas de Alarme Inteligente em 2003. O sistema foi simulado para o cenário clínico detalhado, onde foram analisados três pacientes em diferentes casos. Durante a simulação foi mostrada a influência de um valor mínimo do grau de coincidência para validação das condições de alarme, onde obtivemos valores adequados utilizando para isso um valor de 50%. Ao final mostramos também a importância de se levar em consideração a prioridade das condições de alarme no sistema proposto, o que além de ser um requisito de norma, influencia diretamente no comportamento do sistema, e conseqüentemente na resposta esperada pelo operador em função da mensagem a ele apresentada. / The objective of this work is the development of an Intelligent Alarm System architecture aiming at its application on a Multiparameter Patient Monitoring Equipment, which must be in compliance to international electrical medical equipment safety standards. This theme was raised during the project of a Multiparameter Patient Monitoring Equipment developed at Intermed Equipamento Médico Hospitalar Ltda. This architecture proposes a robust method for sensor fusion using a Kalman Filter as presented by Ebrahim, Feldman and Bar-Kana (1997), in which sensor measurements of heart rate are used to derive a predicted value for this parameter free of artifacts and more precise than any of the individual measurements. The patient vital signs are then, together with the predicted value of the heart rate, transformed into semantic variables and analyzed through fuzzy logic in order to identify alarm conditions. The proposed system is detailed for a clinic scenario, corresponding to a Cardiac Intensive Care Unit, where the main configuration parameters were obtained through interviews with health care professionals. The system development was focused on the compliance to International Standards at present day, which allow the use of Intelligent Alarm Systems since 2003. The system was simulated for the detailed clinic scenario with three patients in different cases. The simulations show the influence of a minimal grade of coincidence value for the validation of the alarm conditions, where we obtained good results using a value of 50%. At the end of this work, it is shown the importance of considering the alarm priority, which not only is a standard requirement, but also interferes directly on the system behavior, and consequently, on the operator´s expected response to the message.

Emoções exercem influência direta sobre a vida humana, mediando a maneira como os indivíduos interagem e se relacionam, seja em âmbito pessoal ou social. Por essas razões, o desenvolvimento de interfaces homem-máquina capazes de manter interações mais naturais e amigáveis com os seres humanos se torna importante. No desenvolvimento de robôs sociais, assunto tratado neste trabalho, a adequada interpretação do estado emocional dos indivíduos que interagem com os robôs é indispensável. Assim, este trabalho trata do desenvolvimento de um modelo matemático para o reconhecimento do estado emocional humano por meio de expressões faciais. Primeiramente, a face humana é detectada e rastreada por meio de um algoritmo; então, características descritivas são extraídas da mesma e são alimentadas no modelo de reconhecimento de estados emocionais desenvolvidos, que consiste de um classificador de emoções instantâneas, um filtro de Kalman e um classificador dinâmico de emoções, responsável por fornecer a saída final do modelo. O modelo é otimizado através de um algoritmo de têmpera simulada e é testado sobre diferentes bancos de dados relevantes, tendo seu desempenho medido para cada estado emocional considerado. / Emotions have direct influence on the human life and are of great importance in relationships and in the way interactions between individuals develop. Because of this, they are also important for the development of human-machine interfaces that aim to maintain natural and friendly interactions with its users. In the development of social robots, which this work aims for, a suitable interpretation of the emotional state of the person interacting with the social robot is indispensable. The focus of this work is the development of a mathematical model for recognizing emotional facial expressions in a sequence of frames. Firstly, a face tracker algorithm is used to find and keep track of a human face in images; then relevant information is extracted from this face and fed into the emotional state recognition model developed in this work, which consists of an instantaneous emotional expression classifier, a Kalman filter and a dynamic classifier, which gives the final output of the model. The model is optimized via a simulated annealing algorithm and is experimented on relevant datasets, having its performance measured for each of the considered emotional states.

La distraction du conducteur est un des facteurs importants à l’origine des accidents de la route. La détection de la distraction dans le contexte industriel et à faible coût conduit à privilégier des indicateurs reposant sur les capteurs déjà disponibles dans un véhicule série. Cependant, les systèmes actuels sont en général insuffisamment fiables, notamment parce que les grandeurs observées pour réaliser la détection sont assez éloignées du phénomène purement physiologique de distraction. L’approche étudiée ici a consisté à rajouter un modèle de comportement du conducteur (modèle cybernétique), rendant compte des fonctions perceptives et motrices support du contrôle latéral du véhicule. Les paramètres de ce modèle ont été estimés en procédant tour à tour à une identification par paquet de données d’entrée/sortie et à une identification récursive, cette dernière permettant de suivre continûment l'évolution paramétrique. Ensuite, trois approches ont été envisagées pour modéliser voire estimer l’état de distraction, considérant successivement la distraction comme une perturbation affectant les paramètres, la sortie ou l’entrée du modèle cybernétique du conducteur:Approche 1 - La distraction est modélisée comme une perturbation additive en sortie du modèle. Le couple produit par le conducteur est comparé au couple prédit par le modèle rendant compte de la conduite hors distraction. L’erreur de prédiction du couple constitue dans ce cadre le résidu dont la sensibilité à l’état de distraction du conducteur a été étudiée.Approche 2 - La distraction est modélisée par des perturbations multiplicatives, affectant certains paramètres du modèle. L’analyse des paramètres obtenus dans des phases de conduite avec et sans distraction a permis d’étudier leur capacité à rendre compte de la nature et de l’état de la distraction.Approche 3 - La distraction est modélisée comme une perturbation additive sur l’entrée du modèle. L’estimation de cette perturbation constitue un résidu également sensible à l’état de distraction. Les principes et algorithmes proposés pour estimer l’état de distraction ont été validés à partir de données expérimentales collectées pendant une campagne de tests effectuée sur un simulateur de conduite à base fixe, impliquant 35 conducteurs. Les conditions de test alternaient des phases de conduite normale et sujettes à des distractions de différentes natures : distractions cognitive, visuelle, visuomotrice et motrice. Les trois approches proposées donnent des résultats similaires et cohérents entre eux. / Distracted driving is one of the important factors that cause road accidents. The detection of the driver’s state of distraction in the industrial context and at low-cost leads to privilege the indicators based on sensors that are already available on the vehicle. However,current systems are generally not reliable enough, especially because the observed magnitudes to achieve detection are quite far from a purely physiological phenomenon distraction. This led us to propose solutions based on a cybernetic driver model that represent the visual and motor process involved in the lateral control of the vehicle. The parameters of this model have been estimated by conducting successively identification exploiting data packets and recursive identification, the latter allowing to track continuously the parametric evolution over time. Then, three approaches were considered to model or estimate the state of distraction, by modeling alternately thedistraction as a disturbance affecting parameters, the output or the input of the cybernetic model of the driver:Approach 1 - The distraction is modeled as an additive disturbance on the model output. The experimental output, the driver steering wheel torque, is then compared with the predicted steering wheel torque to generate the torque prediction error that is sensitive to the state of distraction.Approach 2 - The distraction is modeled as disturbances that affect the model parameters. The analysis of these parameters identified during normal and distracted driving periods showed that the parameters’ variation depends effectively on the driver’s state of distraction.Approach 3 - Distraction is modeled as an additive disturbance on the input of the model. The estimate of this disturbance is also a significant residue, sensitive to the state of distraction. The principles and algorithms proposed for estimating the state of distraction were validated using experimental data collected during a test campaign conducted on a fixed-base driving simulator, involving 35 drivers. The test conditions alternated normal driving phases and prone to distractions of various kinds: cognitive distractions, visual, visual-motor and motor. The three proposed approaches give similar and consistent results between them.

Unmanned Underwater Vehicles (UUVs) are widely used in commercial, scientific, and military missions for various purposes. What makes this technology challenging is the increasing mission duration and unknown environment. It is necessary to embed fault-tolerant control paradigms into UUVs to increase the reliability of the vehicles and enable them to execute and finalize complex missions. Specifically, fault-tolerant control (FTC) comprises fault detection, identification, and control reconfiguration for fault compensation. Literature review shows that there have been no systematic methods for fault-tolerant control of UUVs in earlier investigations. This study presents a hierarchical methodology of fault detection, identification and compensation (HFDIC) that integrates these functions systematically in different levels. The method uses adaptive finite-impulse-response (FIR) modeling and analysis in its first level to detect failure occurrences. Specifically, it incorporates a FIR filter for on-line adaptive modeling, and a least-mean-squares (LMS) algorithm to minimize the output error between the monitored system and the filter in the modeling process. By analyzing the resulting adaptive filter coefficients, we extract the information on the fault occurrence. The HFDIC also includes a two-stage design of parallel Kalman filters in levels two and three for fault identification using the multiple-model adaptive estimation (MMAE). The algorithm activates latter levels only when the failure is detected, and can return back to the monitoring loop in case of false failures. On the basis of MMAE, we use multiple sliding-mode controllers and reconfigure the control law with a probability-weighted average of all the elemental control signals, in order to compensate for the fault. We validate the HFDIC on the steering and diving subsystems of Naval Postgraduate School (NPS) UUVs for various simulated actuator and/or sensor failures, and test the hierarchical fault detection and identification (HFDI) with realistic data from at-sea experiment of the Florida Atlantic University (FAU) Autonomous Underwater Vehicles (AUVs). For both occasions, we model actuator and sensor failures as additive parameter changes in the observation matrix and the output equation, respectively. Simulation results demonstrate the ability of the HFDIC to detect failures in real time, identify failures accurately with a low computational overhead, and compensate actuator and sensor failures with control reconfiguration. In particular, verification of HFDI with FAU data confirms the performance of the fault detection and identification methodology, and provides important information on the vehicle performance. / Ph. D.