Dissertations / Theses on the topic 'Artificial feedback'

Vision-based object stabilisation is an exciting and challenging area of research, and is one that promises great technical advancements in the field of computer vision. As humans, we are capable of a tremendous array of skilful interactions, particularly when balancing unstable objects that have complex, non-linear dynamics. These complex dynamics impose a difficult control problem, since the object must be stabilised through collaboration between applied forces and vision-based feedback. To coordinate our actions and facilitate delivery of precise amounts of muscle torque, we primarily use our eyes to provide feedback in a closed-loop control scheme. This ability to control an inherently unstable object by vision-only feedback demonstrates an exceptionally high degree of voluntary motor skill. Despite the pervasiveness of vision-based stabilisation in humans and animals, relatively little is known about the neural strategies used to achieve this task. In the last few decades, with advancements in technology, we have tried to impart the skill of vision-based object stabilisation to machines, with varying degrees of success. Within the context of this research, we continue this pursuit by employing the classic Cart Inverted Pendulum; an inherently unstable, non-linear system to investigate dynamic object balancing by vision-only feedback. The Inverted Pendulum is considered to be one of the most fundamental benchmark systems in control theory; as a platform, it provides us with a strong, well established test bed for this research. We seek to discover what strategies are used to stabilise the Cart Inverted Pendulum, and to determine if these strategies can be deployed in Real-Time, using cost-effective solutions. The thesis confronts, and overcomes the problems imposed by low-bandwidth USB cameras; such as poor colour-balance, image noise and low frame rates etc., to successfully achieve vision-based stabilisation. The thesis presents a comprehensive vision-based control system that is capable of balancing an inverted pendulum with a resting oscillation of approximately ±1º. We employ a novel, segment-based location and tracking algorithm, which was found to have excellent noise immunity and enhanced robustness. We successfully demonstrate the resilience of the tracking and pose estimation algorithm against visual disturbances in Real-Time, and with minimal recovery delay. The algorithm was evaluated against peer reviewed research; in terms of processing time, amplitude of oscillation, measurement accuracy and resting oscillation. For each key performance indicator, our system was found to be superior in many cases to that found in the literature. The thesis also delivers a complete test software environment, where vision-based algorithms can be evaluated. This environment includes a flexible tracking model generator to allow customisation of visual markers used by the system. We conclude by successfully performing off-line optimization of our method by means of Artificial Neural Networks, to achieve a significant improvement in angle measurement accuracy.<br>Goodrich Engine Control Systems and Balfour Beatty Rail Technologies

Simulations and Serious Games are powerful e-learning tools that can be designed to provide learning opportunities that stimulate their participants. To achieve this goal, the design of Simulations and Serious Games will often include some balance of three factors: motivation, engagement, and flow. Whilst many frameworks and approaches for Simulation and Serious Game design do provide the means for addressing a combination of these factors to some degree, few address how those factors might be affected by the presence of an out-of-game tutor. It is the position of some researchers that the presence of real-world tutors in a Simulation or Serious Game experience can be shown to have a detrimental effect on motivation, engagement, and flow as a continuously changing state for the participant from in-game to out-of-game breaks immersion. The focus of this study was to develop a framework for the design of Simulations and Serious Games that could provide the means to mitigate some of these identified negative effects of real world tutor. The framework itself, referred to as the Wrongness Framework, uses artificial intelligence techniques and practices to provide internal feedback and assessment to the participant as a foundation for the creation of a rudimentary in-game tutor. To achieve this goal it was necessary to develop the Wrongness Framework to include not only the findings of other scholars and researchers on the topic of feedback and assessment but also to introduce original refinements to existing artificial intelligence mechanisms. To test the abilities of the Wrongness Framework it was applied to two unique case studies each with a different purpose and scope. The first, the AdQuest case study, was a graphic design Serious Game scenario testing the ability of the Wrongness Framework's assessment mechanisms by having 102 postgraduate design students submit graphics for a luxury brand advertisement. These graphics were then assessed by the Wrongness Framework against expectations found in the Wrongness Framework's Intelligent System Knowledge Bank. The students were then surveyed for their responses to their assessments and individual rating scores for each design were taken. The second case study, Promasim, explored the possibilities of feedback tone and efficacy for non-player characters in a project management simulation. This was achieved with the use of expert interviews by both academics and working professionals to provide the information of experienced project managers to develop experiential interaction events for the Simulation. Despite the results of these case studies a full case for the success of the Wrongness Framework could not be made. However, many of the identified challenges for the Wrongness Framework were met and, as such, a case can be made that an adequate foundation for the framework has been successful and has provided the case for further refinement.

An ionic polymer actuator consists of a thin Nafion-117 sheet plated with gold or platinum on both sides. An ionic polymer actuator undergoes large deformation in the presence of low applied voltage across its thickness and exhibits low impedance. They can also be used as large displacement sensors by bending them to induce stresses and generate a voltage response. They operate best in a humid environment. Ionic polymer actuators have been used for various practical applications such as bio-mimetic robotic propulsion, flexible low mass robotic arms, propellors for swimming robotic structures, linear and platform type robotic actuators and active catheter systems. One of the disadvantages of ionic polymer actuators is that their settling time to a unit step voltage is on the order of 5-20 seconds in a cantilever configuration. The slow time constant of an ionic polymer limits the actuation bandwidth. The characteristics of ionic polymer actuators, low force and large displacement (as compared to other actuator technologies such as PZT or PVDF), cannot be used in applications requiring a faster response time for a given actuation signal. Due to this limitation, many applications will not be able to make use of the large displacement effectively because of the limited bandwidth of the actuator. Another disadvantage of using an ionic polymer actuator is that the stiffness of the actuator is a function of the hydration of the polymer. Difficulties in controlling the hydration, which changes with respect to time, results in inconsistencies in the mechanical response exhibited by the polymers during continual usage. Several physical models of ionic polymer actuators have been proposed. The physical phenomenon responsible for the bending is not completely understood and no clear set of principles have been able to explain the motion of the polymers completely. Physical phenomena like ionic motion, back diffusion of water and electrostatic force have been used to explain these models. This research demonstrates the use of feedback control to overcome the limitation of slow settling time. First, an empirical model of the ionic polymers developed by Kanno was modified by studying the step response of these actuators. The empirical model is used to design a feedback compensator by state space modeling techniques. Since the ionic polymer actuator has a slow settling time in the open-loop, the design objectives are to minimize the settling time and constrain the control voltage to be less than a prescribed value. The controller is designed using Linear Quadratic Regulator (LQR) techniques which reduced the number of design parameters to one variable. Simulations are performed which show settling times of 0.03 seconds for closed-loop feedback control are possible as compared to the open-loop settling time of 16-18 seconds. The maximum control voltage varied from 1.2 Volts to 3.5 Volts depending on the LQR design parameter. The controller is implemented and results obtained are consistent with the simulations. Closed-loop settling time is observed to be 4-8 seconds and the ratio of the peak response to the steady-state response is reduced by an order of magnitude. Discrepancies between the experiment and the simulations are attributed to the inconsistencies in the resonant frequency of the actuator. Experiments demonstrate that changes in the surface hydration of the polymer result in 20\% variations in the actuator resonance. Variations in the actuator resonance require a more conservative compensator design, thus limiting the performance of the feedback control system.<br>Master of Science

The mammalian cochlea has been the inspiration to develope contemporary cochlear implants and active dynamic sensors that operate in the sensor's resonance region and possess favorable nonlinear characteristics. In the present work, multi-channel and self-sensing active artificial hair cells (AHCs) made of piezoelectric cantilevers and controlled by a cubic damping feedback controller are developed numerically and experimentally. These novel AHCs function near a Hopf bifurcation and amplify or compress the output by a one-third power-law relationship with the input, analogous to the mammalian cochlear amplifier. The multi-channel AHCs have extended frequency bandwidth to sense over multiple resonant frequencies, unlike conventional single-channel AHCs. Therefore, the adoption of these AHCs reduces the number of required sensors to cover the desired bandwidth of interest in an array format. Furthermore, a novel self-sensing active AHC is created in this study using quadmorph beams for future cochlear implants or sensor design applications. The self-sensing scheme allows miniaturization of the system, embedding AHCs in a limited space, and fabrication of AHC arrays by omitting external sensors from the system for practical implementation. Preliminary research on the extension of this research to MEMS AHCs and arrays of AHCs is also presented. The active AHCs can lead to transformative improvements in the dynamic range, sharpness of the response, and threshold of sound detection in cochlear implants to aid individuals with sensorineural hearing loss. Additionally, they can enhance the dynamic properties of sensors such as fluid flow sensors, microphones, and vibration sensors for various applications.<br>Doctor of Philosophy<br>In the mammalian auditory system, the acoustic wave that enters the ear canal is transmitted to the cochlea of the inner ear where it is decomposed into its frequency components. The cochlea then amplifies faint sounds and compresses high-level signals and as these processes stop due to damage, severe hearing loss occurs. Therefore, the present work is focused on developing artificial hair cells (AHCs) that can accurately replicate cochlea's behavior and aid the creation of prostheses for hearing restoration. In this work, the AHC is a beam with piezoelectric layers that is integrated with a control system designed to apply the cochlea-like amplification/compression on the beam. Experimental and simulation results show that the AHC is able to amplify or compress the output based on its input level similar to the mammalian cochlea. In contrast to previous designs of AHCs where each AHC could sense a single frequency, the system developed in this work possesses multiple sensing channels to increase the frequency range of the AHC. Furthermore, the development of a novel self-sensing scheme allows the omission of the external sensor that was required for the AHC operation in previous devices. This advancement in the self-sensing AHC design paves the way for creating fully implantable AHCs to replace the damaged parts of the cochlea. These multi-channel self-sensing AHCs have the potential to be used in the creation of cochlear implants, or sensors such as accelerometers, microphones, and hydrophones with improved dynamic properties. AHCs with different lengths, i.e. different sensing frequencies, can be mounted in an array format to cover the speech frequency range for speech recognition in individuals with hearing loss.

Artificial Intelligence Lab, Department of MIS, University of Arizona<br>Information retrieval using probabilistic techniques has attracted significant attention on the part of researchers in information and computer science over the past few decades. In the 1980s, knowledge-based techniques also made an impressive contribution to â â intelligentâ â information retrieval and indexing. More recently, information science researchers have turned to other newer inductive learning techniques including symbolic learning, genetic algorithms, and simulated annealing. These newer techniques, which are grounded in diverse paradigms, have provided great opportunities for researchers to enhance the information processing and retrieval capabilities of current information systems. In this article, we first provide an overview of these newer techniques and their use in information retrieval research. In order to familiarize readers with the techniques, we present three promising methods: The symbolic ID3 algorithm, evolution-based genetic algorithms, and simulated annealing. We discuss their knowledge representations and algorithms in the unique context of information retrieval. An experiment using a 8000-record COMPEN database was performed to examine the performances of these inductive query-by-example techniques in comparison with the performance of the conventional relevance feedback method. The machine learning techniques were shown to be able to help identify new documents which are similar to documents initially suggested by users, and documents which contain similar concepts to each other. Genetic algorithms, in particular, were found to out-perform relevance feedback in both document recall and precision. We believe these inductive machine learning techniques hold promise for the ability to analyze usersâ preferred documents (or records), identify usersâ underlying information needs, and also suggest alternatives for search for database management systems and Internet applications.

Thesis (Ph.D)--Electrical and Computer Engineering, Georgia Institute of Technology, 2010.<br>Committee Chair: Dr. Ayanna M. Howard; Committee Member: Dr. Charles Kemp; Committee Member: Dr. Magnus Egerstedt; Committee Member: Dr. Patricio Vela. Part of the SMARTech Electronic Thesis and Dissertation Collection.

The fitness device market is young and rapidly growing. More people than ever before take count of how many steps they walk, how many calories they burn, their heart rate over time, and even their quality of sleep. New, and as of yet, unreleased fitness devices have promised the next evolution of functionality with exercise technique analysis. These next generation of fitness devices have wrist and armband style form factors, which may not be optimal for barbell exercises such as back squat, bench press, and overhead press where a sensor on one arm may not provide the most relevant data about a lift. Barbell path analysis is a well-known visual tool to help diagnose weightlifting technique deficiencies, but requires a camera pointed at the athlete that is integrated with motion-tracking software. This camera set up is not available at most gyms, so this motivates the use of a small, unobtrusive sensor to obtain data about an athlete's weightlifting technique. Researchers have shown that an accelerometer attached to a barbell while the athlete is lifting yields just as accurate acceleration information as a camera. The LIFT (Leveraging Information For Training) automated weight lifting coach attempts to implement a simple, unobtrusive system for analyzing and providing feedback on barbell weight lifting technique.

Orientadores Vera Lúcia da Silveira Nantes Button, Percy Nohama<br>Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação<br>Made available in DSpace on 2018-08-22T09:56:14Z (GMT). No. of bitstreams: 1 NogueiraNeto_GuilhermeNunes_D.pdf: 3511169 bytes, checksum: eab59dd67c7bd6370efd2d475664860e (MD5) Previous issue date: 2013<br>Resumo: A lesão medular afeta a condução dos sinais motores e sensoriais na medula. Com isso, a realização de movimentos voluntários por pessoas com lesão medular é prejudicada quando não impossível. A estimulação elétrica neuromuscular (EENM) é uma técnica que pode auxiliá-los a contornar essas dificuldades, pois possibilita ativar a musculatura de um membro paralisado. Embora a EENM em malha aberta seja capaz de produzir movimentos em um membro paralisado, ele não é capaz de compensar automaticamente desvios angulares devidos a distúrbios. Fenômenos supressores de desempenho muscular como a fadiga muscular e a adaptação neuronal podem comprometer a qualidade de um movimento funcional evocado artificialmente. Por isso, sistemas de controle são incorporados a estimuladores elétricos. A mecanomiografia (MMG) pode fornecer informações sobre a condição de um músculo em contração. Este trabalho investiga a viabilidade do emprego de sinais de MMG no controle da aplicação de EENM visando a sua incorporação em estratégias de controle de membros inferiores (MMII) em lesados medulares sem treinamento. Para isso, foi desenvolvido um sistema integrado de monitoração multiparamétrica e EENM com capacidade de incorporação do sinal de MMG na estratégia de controle. Um ensaio com voluntários hígidos e contrações isométricas de membros superiores (MMSS) apontou variações em parâmetros de MMG no início e final da contração. Outro ensaio analisou o sinal de MMG e de ângulo com controle de EENM em malha e cadeia abertas para indivíduos hígidos e lesados medulares. Sinais de MMG e EENM em contrações dinâmicas de MMII foram registrados. Com o sistema integrado, aplicou-se EENM com controle de ângulo de joelho em malha fechada obtendo- se estabilidade angular do joelho de um lesado medular. Finalmente, o sistema integrado aplicou controle para estabilizar o ângulo de joelho e monitorar sinais de MMG. Os resultados de ensaios experimentais mostraram que parâmetros espectrais, como a frequência média de potência, apresentaram variações entre o início e o fim da contração quando a amplitude de EENM apresentou taxas de incremento mais elevadas. Conclui-se que existe viabilidade no emprego de MMG em estratégias de controle de EENM para MMII de lesados medulares em estabilidade angular, observando-se a taxa de incremento/decremento da saída do estimulador<br>Abstract: Spinal cord injury affects conduction of motor and sensory signals that flow through the spinal cord. Therefore, the performance of voluntary movements by spinal cord injured (SCI) persons is impaired, if not impossible. Neuromuscular electrical stimulation (NMES) is a technique that aids SCI individuals to overcome these difficulties, because it allows activation of paralyzed limb muscles. Though NMES in open loop and in open chain tasks is able to generate movements in paralyzed limbs, it is not capable to automatically compensate for angle deviations due to disturbances. Suppressive phenomena of muscle performance, such as muscle fatigue and motor neuron adaptation, can compromise the quality of an artificially elicited functional moment. Thus, closed-loop control systems are merged with electrical stimulators. Mechanomyography (MMG) can provide information about a contracting muscle condition. This work proposes to investigate the feasibility of using MMG signals in the control of NMES application aiming its incorporation in lower limb control strategies of untrained spinal cord injured persons. An integrated system was built, comprising a multiparametric acquisition system and a closed-loop controlled neuromuscular electrical stimulator. The integrated system is able to incorporate MMG signals in the control strategy. A test with healthy volunteers performing upper limb isometric contractions indicated variations in MMG parameters between the beginning and end of contraction. Another test investigated MMG signals and joint angles with NMES in open loop and in open chain to healthy and SCI individuals. NMES and MMG signals during lower limb dynamic contractions were registered. Using the integrated system, NMES was applied to the control of knee angle in closed loop configuration ensuring angular stability of both knees on a single SCI volunteer. Finally, the integrated system applied closed loop control to stabilize the knee angle, and monitored MMG signals. Experimental results revealed that spectral parameters such as mean power frequency presented variations between the beginning and end of contraction when NMES amplitude showed higher amplitude increase rates. In conclusion, there is feasibility of using MMG signals in control strategies of NMES for lower limbs of SCI persons in reasonable angle stability. However, one must observe the stimulator output increase/decrease rate<br>Doutorado<br>Engenharia Biomedica<br>Doutor em Engenharia Elétrica

El problema d'investigació que s'ha desenvolupat en aquesta tesi pretén donar resposta a algunes de les preguntes obertes actualment al voltant de dos grans temes, "Emotion Awareness in elearning" i "Emotional feedback in e-learning", proposant un model conceptual i computacional de anàlisi de l'emoció en un entorn CSCL i, en particular, en el discurs educacional que té lloc en aquests entorns. En aquesta línia, els objectius d'aquesta tesi han estat mostrar les emocions que els alumnes senten durant el seu procés d'aprenentatge col·laboratiu en entorns virtuals i com aquestes emocions influeixen en el seu aprenentatge. Avaluant les emocions obtingudes, classificant-les i observant la seva evolució dins el període de temps que duri l'acció formativa. Estudiant, a més, per quines situacions formatives, els tutors virtuals afectius poden resultar efectius i apropiats i determinar quines capacitats o habilitats han de tenir per optimitzar els processos d'aprenentatge dels alumnes. Els resultats mostren que un enfocament integrat que inclou solucions i estratègies metodològiques constructivistes per a la detecció i interpretació de l'emoció pot ser desenvolupat i aplicat. Entre els resultats obtinguts es destaca la capacitat del sistema de proporcionar consciència emocional i retroalimentació emocional, analitzar els efectes d'aquest tipus de consciència i la influència d'aquest tipus de retroalimentació en aspectes relacionats amb l'aprenentatge, la motivació, el compromís, l'autoregulació i els resultats de l'aprenentatge dels estudiants així com la gestió del temps (time management) i la gestió d'un mateix (self-management) en termes d'implicació conductual i cognitiva, autoregulació i rendiment dels estudiants.<br>El problema de investigación desarrollado en esta tesis pretende dar respuesta a algunas de las preguntas abiertas actualmente en torno a dos grandes temas, <<Emotion Awareness in elearning>> y <<Emotional feedback in elearning>>, proponiendo un modelo conceptual y computacional de análisis de la emoción en un entorno CSCL y, en particular, en el discurso educacional que tiene lugar en este contexto. En esta línea, los objetivos de esta tesis han sido mostrar las emociones que los alumnos sienten durante su proceso de aprendizaje colaborativo en entornos virtuales y cómo estas emociones influyen en su aprendizaje. Evaluando las emociones obtenidas, clasificándolas y observando su evolución dentro del periodo de tiempo que dure la acción formativa. Estudiando, además, para qué situaciones formativas, los tutores virtuales afectivos pueden resultar efectivos y apropiados y determinar qué capacidades o habilidades deben tener para optimizar los procesos de aprendizaje de los alumnos. Los resultados muestran que un enfoque integrado que incluye soluciones y estrategias metodológicas constructivistas para la detección e interpretación de la emoción se puede desarrollar y aplicar en este contexto. Entre los resultados obtenidos se destaca la capacidad del sistema de proporcionar conciencia y retroalimentación emocional, y de analizar los efectos de este tipo de conciencia y la influencia de este tipo de retroalimentación en aspectos relacionados con el aprendizaje, la motivación, el compromiso, la autorregulación y los resultados del aprendizaje de los estudiantes así como la gestión del tiempo (time management) y la gestión de uno mismo (self-management) en términos de implicación conductual y cognitiva, autorregulación y rendimiento de los estudiantes.<br>The research problem that has been developed in this thesis aims to answer some of the currently open questions on two big issues, "emotion awareness in e-learning" and "emotional feedback in e-learning". It also aims to propose a conceptual and computational model of emotion analysis in a CSCL environment and, in particular, in the educational discourse that takes place in this context. Therefore, the goals of this thesis have been to reveal and represent the emotions that students feel during their process of collaborative learning in virtual environments, show how these emotions influence their learning, analyse and evaluate the obtained emotions, classify them, and observe their evolution within the period of time that the learning activity lasts. Moreover, studying particular learning situations, an online affective pedagogical tutor (APT) has been developed that provides efficient and appropriate affective feedback. In addition, our study determined which skills or abilities the APT should have in order to optimize the student’s learning processes. The results show that an integrated approach including solutions and constructivist methodological strategies for the detection and interpretation of emotion can be developed and applied. Among the results obtained, we highlight the system's ability to provide emotional awareness and emotional feedback and the effects of this kind of awareness and the influence of this type of feedback on issues related to learning, motivation, commitment, self-regulation and student learning outcomes as well as to time management and self-management in terms of behavioural and cognitive involvement, self-regulation and student performance.

Le développement des interfaces cerveau-machine (ICM) apporte une nouvelle perspective aux patients en perte d'autonomie motrice. En combinant des enregistrements en ligne de l'activité cérébrale avec un algorithme de décodage, les patients peuvent apprendre à contrôler un bras robotique afin d'effectuer des actions simples. Cependant, contrairement à la grande quantité d'informations somatosensorielles fournie par les membres physiologiques vers le cerveau, les ICM actuelles sont dépourvues de capteurs de toucher et de force. Les patients doivent donc se fier uniquement à la vision et à l'audition, qui sont inadaptées au contrôle d'une prothèse. Cela contraste avec le fait que dans le cas d’un membre sain, les entrées somatosensorielles seules peuvent guider efficacement la manipulation d'un objet fragile, ou assurer une trajectoire précise. Une caractéristique intéressante des entrées somatosensorielles est leur organisation topologique à la surface corticale. Cette carte corticale semble jouer un rôle déterminant dans la perception sensorielle. Par conséquent, l'intégration d'une rétroaction somatosensorielle artificielle alignée sur cette carte corticale pourrait aider grandement l’intégration avoir un impact déterminant dans le transporter un grand nombre d'informations. Pour vérifier cette hypothèse, nous avons développé un ICM chez la souris qui inclut une riche rétroaction corticale artificielle de type somatosensoriel. Notre installation comprend des enregistrements en ligne de l'activité de plusieurs neurones dans le cortex moteur primaire du whisker (wM1), et fournit une rétroaction simultanée via une photostimulation du cortex somatosensoriel primaire du whisker (wS1), à faible latence, haute fréquence et structure spatiale, basée sur une cartographie obtenue par imagerie intrinsèque. Nous démontrons le fonctionnement de la boucle et montrons que les souris peuvent détecter l’activité des neurons dans wS1 déclenchée par les photostimulations. Surtout, nous montrons qu'en utilisant l'ICM en boucle fermée, les souris peuvent avoir une meilleure performance dans une tâche comportementale lorsque la structure du feedback artificiel est respecté somatotopie connue de wS1<br>The development of brain-machine interfaces (BMIs) brings a new perspective to patients with a loss of motor autonomy. By combining online recordings of brain activity with a decoding algorithm, patients can learn to control a robotic arm in order to perform simple actions. However, in contrast to the vast amounts of somatosensory information channeled by limbs to the brain, current BMIs are devoid of touch and force sensors. Patients must therefore rely solely on vision and audition, which are maladapted to the control of a prosthesis. In contrast, in a healthy limb, somatosensory inputs alone can efficiently guide the handling of a brittle object, or ensure a smooth trajectory. One interesting feature of somatosensory inputs is its topological organization at the cortical surface. This cortical map plays a role in sensory perception. Therefore, integrating artificial somatosensory feedback aligned and consistent with this cortical map could potentially help the subject to decode the information conveyed by the feedback. To test this hypothesis, we have developed a BMI in the mouse model that includes a rich artificial somatosensory-like cortical feedback. Our setup includes online recordings of the activity of multiple neurons in the whisker primary motor cortex (wM1), and delivers feedback simultaneously via a low-latency, high-refresh rate photo-stimulation of the whisker primary somatosensory cortex (wS1) that is spatially structured at the mesoscopic scale, based on a mapping obtained by intrinsic imaging. We demonstrate the operation of the loop and show that mice can detect the wS1 neuronal spiking triggered by the photostimulations. Remarkably, we show that in the closed loop BMI, mice can have a significantly better performance in a behavioral task when the structure of the artificial feedback abides to the known wS1 somatotopy

Doctor of Philosophy<br>Department of Civil Engineering<br>Yacoub M. Najjar<br>Access to databases after the digital revolutions has become easier because large databases are progressively available. Knowledge discovery in these databases via intelligent data analysis technology is a relatively young and interdisciplinary field. In engineering applications, there is a demand for turning low-level data-based knowledge into a high-level type knowledge via the use of various data analysis methods. The main reason for this demand is that collecting and analyzing databases can be expensive and time consuming. In cases where experimental or empirical data are already available, prediction models can be used to characterize the desired engineering phenomena and/or eliminate unnecessary future experiments and their associated costs. Phenomena characterization, based on available databases, has been utilized via Artificial Neural Networks (ANNs) for more than two decades. However, there is a need to introduce new paradigms to improve the reliability of the available ANN models and optimize their predictions through a hybrid decision system. In this study, a new set of ANN modeling approaches/paradigms along with a new method to tackle partially missing data (Query method) are introduced for this purpose. The potential use of these methods via a hybrid decision making system is examined by utilizing seven available databases which are obtained from civil engineering applications. Overall, the new proposed approaches have shown notable prediction accuracy improvements on the seven databases in terms of quantified statistical accuracy measures. The proposed new methods are capable in effectively characterizing the general behavior of a specific engineering/scientific phenomenon and can be collectively used to optimize predictions with a reasonable degree of accuracy. The utilization of the proposed hybrid decision making system (HDMS) via an Excel-based environment can easily be utilized by the end user, to any available data-rich database, without the need for any excessive type of training.

Orientador: Romis Ribeiro de Faissol Attux<br>Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação<br>Made available in DSpace on 2018-08-15T11:06:22Z (GMT). No. of bitstreams: 1 Wada_Cristina_M.pdf: 4155297 bytes, checksum: 8683ce097ae6eab2c4bc3f3cae097225 (MD5) Previous issue date: 2010<br>Resumo: Na transmissão de dados através de um canal ocorrem distorções que podem eventualmente levar a níveis inaceitáveis de degradação.Uma distorção bastante comum nesse cenário é a interferência intersimbólica ,que é a conseqüência do espalhamento temporal do sinal de informação.Para mitigar essa Interferência ,é usual empregar um equalizador ,que pode ser adaptado de modo supervisionado ou não upervisionado. Uma solução clássica no caso não supervisionado é fazer uso do critério de mínimo erro quadrático médio de predição.Sabe-se que tal abordagem ,no contexto linear,é eficiente apenas para canais de fase mínima ou máxima.Para lidar com canais de fase mista ,é preciso recorrer a estruturas não lineares. Neste trabalho ,investigaremos a relevância ,nesse contexto,do uso de preditores não lineares contendo laços de realimentação .Analizar-se-á o desempenho de estruturas neurais recorrentes sob um conjunto representativo de canais ,de modo a permitir a investigação dos efeitos da memória sobre o processo de desconvolução.O processo adaptativo será conduzido por um sistema imunológico artificial,dotado de significativo potencial de busca global e robustez a soluções instáveis<br>Abstract: When data is transmitted trough a channel, it may be subject to several sorts of distortion that might cause unacceptable level of degradation. A very usual type of distortion is the intersymbol interference ,which is a consequence of the temporal spread of the information-bearing signal .To mitigate this interference ,it is usual to employ an equalizer ,which can be adapted either in a supervised or an unsupervised manner. For the latter case, a predictive structure, optimized according to the mean squared error criterion, is a classical solution. In the linear context, it is known that this approach is efficient only for minimum- or maximum-phase channels: to deal with mixed-phase channels, it is necessary to resort to nonlinear structures. In this work, we investigate the relevance, in this context, of the use of nonlinear predictors with feedback loops. The performance of nonlinear neural structures is analyzed in asset of representative channels, in order to form a better understanding of the effect of the channel memory on the signal and to make use of it in the deconvolution process. An optimization algorithm based on the concept of artificial immune systems is applied in the adaptation of predictors, due to its powerful global search capabilities and robustness to unstable solutions<br>Mestrado<br>Engenharia de Computação<br>Mestre em Engenharia Elétrica

Self-adaptive systems have limited time to adjust their configurations whenever their adaptation goals, i.e., quality requirements, are violated due to some runtime uncertainties. Within the available time, they need to analyze their adaptation space, i.e., a set of configurations, to find the best adaptation option, i.e., configuration, that can achieve their adaptation goals. Existing formal analysis approaches find the best adaptation option by analyzing the entire adaptation space. However, exhaustive analysis requires time and resources and is therefore only efficient when the adaptation space is small. The size of the adaptation space is often in hundreds or thousands, which makes formal analysis approaches inefficient in large-scale self-adaptive systems. In this thesis, we tackle this problem by presenting an online learning approach that enables formal analysis approaches to analyze large adaptation spaces efficiently. The approach integrates with the standard feedback loop and reduces the adaptation space to a subset of adaptation options that are relevant to the current runtime uncertainties. The subset is then analyzed by the formal analysis approaches, which allows them to complete the analysis faster and efficiently within the available time. We evaluate our approach on two different instances of an Internet of Things application. The evaluation shows that our approach dramatically reduces the adaptation space and analysis time without compromising the adaptation goals.

A finalidade desse trabalho é desenvolver um protótipo de uma prótese antropomórfica multifuncional para membros superiores para pacientes amputados. Seu objetivo é substituir a mão natural perdida, de forma a auxiliar a realização de algumas tarefas diárias do usuário. A prótese possuirá características antropomórficas, tais como aparência e movimentação semelhantes às da mão humana, e características naturais inerentes à mesma, por exemplo, o arco reflexo. Além disso, contará também com meios de realimentação táteis das informações de forças aplicadas pela prótese em objetos, bem como sua temperatura para o paciente, suprindo assim uma das carências das próteses convencionais. Esse dispositivo terá incorporado na sua construção sensores diversos para realizar as funções propostas e contará com um algoritmo baseado em redes neurais artificiais, capaz de identificar padrões dos sinais mioelétricos do paciente, que serão utilizados como sinais de controle, possibilitando ao paciente um comando natural. Todas essas implementações visam contribuir para a redução da taxa de rejeição de próteses para membros superiores e possibilitar uma maior reabilitação e reintegração do paciente à sociedade.<br>The purpose of this assignment is to develop a multifunctional and anthropomorphic upper limb prosthesis prototype for amputated patients. Its objective is to substitute the natural lost hand, in a way to improve the performance of regular activities. This prosthesis will have anthropomorphic characteristics, like appearance and movement, similar to the ones of the human hand, and natural characteristics inherent to it, for example the reflected arc. Another characteristic will be the tactile feedback ways of obtaining the information of the forces applied by the prosthesis in objects, as well as their temperature for the patient, overcoming therefore one of the traditional prosthesis\' deficiency. This device will have incorporated in its construction many sensors in order to do the proposed functions and it will use an algorithm based on the artificial neural network that is able to recognize patterns of myoelectric signals of the patient, which will be used as control signals, making possible to the patient a natural command. All of these implementations objective to contribute for the reduction of the rejection rate of prostheses for upper limb members and make possible a better rehabilitation and reintegration of the patient in the society.

Animals exploit soft structures to move smoothly and effectively in complex natural environments. These capabilities have inspired robotic engineers to incorporate soft actuating technologies into their designs. Developing soft muscle-like actuation technology is one of the grand challenges in the creation of soft-body robots that can move, deform their body, and modulate body stiffness. This paper presents the development of smart pneumatic McKibben muscles woven and reinforced by using conductive insulated wires to equip the muscles with an inherent sensing capability, in which the deformation of the muscles can be effectively measured by calculating the change of wire inductance. Sensing performance of a variety of weaving angles is investigated. The ideal McKibben muscle models are used for analysing muscle performance and sensing accuracy. The experimental results show that the contraction of the muscles is proportional to the measured change of inductance. This relationship is applied to a PID control system to control the contraction of smart muscles in simulation, and good control performance is achieved. The creation of smart muscles with an inherent sensing capability and a good controllability is promising for operation of future soft robots.

en català:<br/><br/>Un camp de recerca important dins del paradigma del Computer-Supported Collaborative Learning (CSCL) és la importància en la gestió eficaç de la informació d'esdeveniments generada durant l'activitat de l'aprenentatge col·laboratiu virtual, per a proporcionar coneixement sobre el comportament dels membres del grup. Aquesta visió és especialment pertinent en l'escenari educatiu actual que passa d'un paradigma tradicional - centrat en la figura d'un instructor magistral - a un paradigma emergent que considera els estudiants com actors centrals en el seu procés d'aprenentatge. En aquest nou escenari, els estudiants aprenen, amb l'ajuda de professors, la tecnologia i els altres estudiants, el que potencialment necessitaran per a desenvolupar les seves activitats acadèmiques o professionals futures.<br/>Els principals aspectes a tenir en compte en aquest context són, primer de tot, com dissenyar una plataforma sota el paradigma del CSCL, que es pugui utilitzar en situacions reals d'aprenentatge col·laboratiu complexe i a llarg termini, basades en el model d'aprenentatge de resolució de problemes. I que permet al professor una anàlisi del grup més eficaç així com donar el suport adequat als estudiants quan sigui necessari. <br/>En segon lloc, com extreure coneixement pertinent de la col·laboració per donar consciència i retorn als estudiants a nivell individual i de rendiment del grup, així com per a propòsits d'avaluació. <br/>L'assoliment d'aquests objectius impliquen el disseny d'un model conceptual d'interacció durant l'aprenentatge col·laboratiu que estructuri i classifiqui la informació generada en una aplicació col·laborativa en diferents nivells de descripció. A partir d'aquesta aproximació conceptual, els models computacionals hi donen resposta per a proporcionar una extracció eficaç del coneixement produït per l'individu i per l'activitat del grup, així com la possibilitat d'explotar aquest coneixement com una eina metacognitiva pel suport en temps real i regulat del procés d'aprenentatge col·laboratiu.<br/>A més a més, les necessitats dels entorns CSCL han evolucionat en gran mesura durant els darrers anys d'acord amb uns requisits pedagògics i tecnològics cada cop més exigents. Els entorns d'aprenentatge col·laboratius virtuals ara ja no depenen de grups d'estudiants homogenis, continguts i recursos d'aprenentatge estàtics, ni pedagogies úniques, sinó que exigeixen una forta personalització i un alt grau de flexibilitat. En aquest nou escenari, les organitzacions educatives actuals necessiten estendre's i moure's cap a paradigmes d'ensenyament altament personalitzats, amb immediatesa i constantment, on cada paradigma incorpora el seu propi model pedagògic, el seu propi objectiu d'aprenentatge i incorpora els seus propis recursos educatius específics. <br/>Les demandes de les organitzacions actuals també inclouen la integració efectiva, en termes de cost i temps, de sistemes d'aprenentatge llegats i externs, que pertanyen a altres institucions, departaments i cursos. Aquests sistemes llegats es troben implementats en llenguatges diferents, suportats per plataformes heterogènies i distribuïdes arreu, per anomenar alguns dels problemes més habituals. Tots aquests problemes representen certament un gran repte per la comunitat de recerca actual i futura. Per tant, els propers esforços han d'anar encarats a ajudar a desenvolupadors, recercaires, tecnòlegs i pedagogs a superar aquests exigents requeriments que es troben actualment en el domini del CSCL, així com proporcionar a les organitzacions educatives solucions ràpides i flexibles per a potenciar i millorar el rendiment i resultats de l'aprenentatge col·laboratiu. Aquesta tesi proposa un primer pas per aconseguir aquests objectius.<br>An important research topic in Computer Supported Collaborative Learning (CSCL) is to explore the importance of efficient management of event information generated from group activity in collaborative learning practices for its further use in extracting and providing knowledge on interaction behavior. <br/>The essential issue here is first how to design a CSCL platform that can be used for real, long-term, complex collaborative problem solving situations and which enables the instructor to both analyze group interaction effectively and provide an adequate support when needed. Secondly, how to extract relevant knowledge from collaboration in order to provide learners with efficient awareness and feedback as regards individual and group performance and assessment. The achievement of these tasks involve the design of a conceptual framework of collaborative learning interaction that structures and classifies the information generated in a collaborative application at several levels of description. Computational models are then to realize this conceptual approach for an efficient management of the knowledge produced by the individual and group activity as well as the possibility of exploiting this knowledge further as a metacognitive tool for real-time coaching and regulating the collaborative learning process.<br/>In addition, CSCL needs have been evolving over the last years accordingly with more and more demanding pedagogical and technological requirements. On-line collaborative learning environments no longer depend on homogeneous groups, static content and resources, and single pedagogies, but high customization and flexibility are a must in this context. As a result, current educational organizations' needs involve extending and moving to highly customized learning and teaching forms in timely fashion, each incorporating its own pedagogical approach, each targeting a specific learning goal, and each incorporating its specific resources. <br/>These entire issues certainly represent a great challenge for current and future research in this field. Therefore, further efforts need to be made that help developers, technologists and pedagogists overcome the demanding requirements currently found in the CSCL domain as well as provide modern educational organizations with fast, flexible and effective solutions for the enhancement and improvement of the collaborative learning performance and outcomes. This thesis proposes a first step toward these goals.<br/><br/>Índex foliat:<br/>The main contribution in this thesis is the exploration of the importance of an efficient management of information generated from group activity in Computer-Supported Collaborative Learning (CSCL) practices for its further use in extracting and providing knowledge on interaction behavior. To this end, the first step is to investigate a conceptual model for data analysis and management so as to identify the many kinds of indicators that describe collaboration and learning and classify them into high-level potential categories of effective collaboration. Indeed, there are more evident key discourse elements and aspects than those shown by the literature, which play an important role both for promoting student participation and enhancing group and individual performance, such as, the impact and effectiveness of students' contributions, among others, that are explored in this work. By making these elements explicit, the discussion model proposed accomplishes high students' participation rates and contribution quality in a more natural and effective way. This approach goes beyond a mere interaction analysis of asynchronous discussion in the sense that it builds a multi-functional model that fosters knowledge sharing and construction, develops a strong sense of community among students, provides tutors with a powerful tool for students' monitoring, discussion regulation, while it allows for peer facilitation through self, peer and group awareness and assessment.<br/>The results of the research described so far motivates the development of a computational system as the translation from the conceptual model into a computer system that implements the management of the information and knowledge acquired from the group activity, so as to be efficiently fed back to the collaboration. The achievement of a generic, robust, flexible, interoperable, reusable computational model that meets the fundamental functional needs shared by any collaborative learning experience is largely investigated in this thesis. The systematic reuse of this computational model permits a fast adaptation to new learning and teaching requirements, such as learning by discussion, by relying on the most advanced software engineering processes and methodologies from the field of software reuse, and thus important benefits are expected in terms of productivity, quality, and cost.<br/>Therefore, another important contribution is to explore and extend suitable software reuse techniques, such as Generic Programming, so as to allow the computational model to be successfully particularized in as many as situations as possible without losing efficiency in the process. In particular, based on domain analysis techniques, a high-level computational description and formalization of the CSCL domain are identified and modeled. Then, different specific-platform developments that realize the conceptual description are provided. It is also explored a certain level of automation by means of advanced techniques based on Service-Oriented Architectures and Web-services while passing from the conceptual specification to the desired realization, which greatly facilitates the development of CSCL applications using this computational model.<br/>Based on the outcomes of these investigations, this thesis contributes with computational collaborative learning systems, which are capable of managing both qualitative and quantitative information and transforming it into useful knowledge for all the implicated parties in an efficient and clear way. This is achieved by both the specific assessment of each contribution by the tutor who supervises the discussion and by rich statistical information about student's participation. This statistical data is automatically provided by the system; for instance, statistical data sheds light on the students' engagement in the discussion forum or how much interest drew the student's intervention in the form of participation impact, level of passivity, proactivity, reactivity, and so on. The aim is to provide both a deeper understanding of the actual discussion process and a more objective assessment of individual and group activity.<br/>This information is then processed and analyzed by means of a multivariate statistical model in order to extract useful knowledge about the collaboration. The knowledge acquired is communicated back to the members of the learning group and their tutor in appropriate formats, thus providing valuable awareness and feedback of group interaction and performance as well as may help identify and assess the real skills and intentions of participants. The most important benefit expected from the conceptual model for interaction data analysis and management is a great improvement and enhancement of the learning and teaching collaborative experiences.<br/>Finally, the possibilities of using distributed and Grid technology to support real CSCL environments are also extensively explored in this thesis. The results of this investigation lead to conclude that the features provided by these technologies form an ideal context for supporting and meeting demanding requirements of collaborative learning applications. This approach is taken one step further for enhancing the possibilities of the computational model in the CSCL domain and it is successfully adopted on an empirical and application basis. From the results achieved, it is proved the feasibility of distributed technologies to considerably enhance and improve the collaborative learning experience. In particular, the use of Grid computing is successfully applied for the specific purpose of increasing the efficiency of processing a large amount of information from group activity log files.

A prevalent morphology in the microscopic world of artificial microswimmers, bacteria and viruses is that of a helix. The intriguingly different physics at play at the small scale level make it necessary for bacteria to employ swimming strategies different from our everyday experience, such as the rotation of a helical filament. Bio-inspired microswimmers that mimic bacterial locomotion achieve propulsion at the microscale level using magnetically actuated, rotating helical filaments. A promising application of these artificial microswimmers is in non-invasive medicine, for drug delivery to tumours or microsurgery. Two crucial features need to be addressed in the design of microswimmers. First, the ability to selectively control large ensembles and second, the adaptivity to move through complex conduit geometries, such as the constrictions and curves of the tortuous tumour microvasculature. In this dissertation, a mechanics-based selective control mechanism for magnetic microswimmers is proposed, and a model and simulation of an elastic helix passing through a constricted microchannel are developed. Thereafter, a theoretical framework is developed for the propulsion by stiff elastic filaments in viscous fluids. In order to address this fluid-structure problem, a pertubative, asymptotic, elastohydrodynamic approach is used to characterise the deformation that arises from and in turn affects the motion. This framework is applied to the helical filaments of bacteria and magnetically actuated microswimmers. The dissertation then turns to the sub-bacterial scale of bacteriophage viruses, 'phages' for short, that infect bacteria by ejecting their genetic material and replicating inside their host. The valuable insight that phages can offer in our fight against pathogenic bacteria and the possibility of phage therapy as an alternative to antibiotics, are of paramount importance to tackle antibiotics resistance. In contrast to typical phages, flagellotropic phages first attach to bacterial flagella, and have the striking ability to reach the cell body for infection, despite their lack of independent motion. The last part of the dissertation develops the first theoretical model for the nut-and-bolt mechanism (proposed by Berg and Anderson in 1973). A nut being rotated will move along a bolt. Similarly, a phage wraps itself around a flagellum possessing helical grooves, and exploits the rotation of the flagellum in order to passively travel along and towards the cell body, according to this mechanism. The predictions from the model agree with experimental observations with respect to directionality, speed and the requirements for succesful translocation.

Cette thèse traite du problème de la stabilisation d'un système de lancement aéroporté (éventuellement non habité) pour satellites. Le lancement aéroporté consiste à ramener, à l'aide d'un avion, un satellite et son lanceur (fusée) à une certaine hauteur, et d'exécuter son lancement dans les airs (souvent en larguant la fusée). Ceci est similaire au lancement d'un missile par un avion chasseur. La plus grande différence réside dans le rapport de masse entre l'avion et le lanceur qui est beaucoup plus proche de l'unité (fusée lourde comparée à la masse de l'avion). Le système est composé de deux étages: le premier étage est dit avion porteur qui est un véhicule aérien automatisé. Il porte le lanceur qui constitue le deuxième étage (la fusée). Dans la première partie, sont proposées des approches de modélisation pour le système de largage pendant et après le largage. La première approche considère que la phase de séparation est instantanée, mais imparfaite. Par conséquent le système est vu comme un modèle d'aéroplane dont les variables d'état sont avec des larges conditions initiales dues à la séparation imparfaite. Une deuxième approche considère la séparation elle-même, représentée par une forte perturbation (un extrême cas) sur les forces et couples aérodynamiques du modèle au cours d'un intervalle de temps. Dans la deuxième partie, afin de stabiliser le système de largage après la séparation, la commande à intégrateur conditionnel modifié est développée dans un premier temps pour une classe des systèmes non-linéaires multi-entrées multi-sorties, avec comme point de départ la théorie introduite par Khalil et co-auteurs pour des systèmes mono entrée mono sortie. Cette commande a été ensuite étendue pour la commande à servo-compensateur conditionnel modifié pour une classe de systèmes non-linéaires multi-entrées multi-sorties. Les deux stratégies ont été appliquées pour stabiliser le système de largage pendant et après la phase de séparation. Ces techniques ont l'avantage d'être robustes et de pouvoir utiliser des modèles approximatifs. D'un autre côté, il était important d'examiner la possibilité d'obtenir de meilleures performances en utilisant de meilleurs modèles. Pour cette raison, la commande de linéarisation par bouclage dynamique a été étudiée. Finalement, les performances de toutes ces méthodes de commande (ainsi que certaines commandes de base additionnelles) ont été illustrées par des simulations sous Matlab/Simulink sur un modèle non-linéaire de F-16<br>This thesis addresses the problem of the stabilization of an (unmanned) airlaunch system. Air launching consists in bringing a satellite and its launcher (rocket) to a certain height using an aircraft, and then launching it from the air (often by dropping the rocket), in a similar way of launching a missile from a fighter. The main difference is that the envisaged mass ratio is much closer to one (heavy rocket compared to aircraft mass). It is then composed of two stages: the first stage called carrier aircraft consists of an aerial vehicle that carries the launcher which constitutes the second stage (rocket). This thesis starts by introducing the problem and objectives, continues by presenting several approaches to model the airlaunch system, and ends by developing different advanced control methods to stabilize it after the launching phase. In the modeling part we propose a firstly approach called the initial condition model which assumes that the separation phase is instantaneous, and then the airlaunch system is composed of an aircraft model after the launching phase but with large initial conditions on its state variables, caused by a non-perfect split phase. A second approach assumes that the separation phase itself is modeled by a disturbance on aerodynamic forces and moments (from a worst case) during a time interval. In the control part a modified conditional integrator controller for a class of nonlinear multi-input multi-output systems is first developed starting from the conditional integrator theory developed by Khalil and co-workers. It is then extended to a modified conditional servocompensator control for a class of nonlinear multi-input multi-output systems. Both control strategies were then applied to stabilize the airlaunch system after the separation phase. They have the advantage of being very robust, and they don't depend so much on reliable models. Even if these control strategies gave good results, it was investigated in this thesis another control approach much more dependent on detailed and reliable models. This approach was based on dynamic feedback linearization theory, and the main idea is to obtain better performance in trade off better models. Finally, all proposed control methods (plus some standard ones) were compared and illustrated by simulations under Matlab/Simulink on a nonlinear F-16 model. These simulations have shown that the results were as expected, and that each control strategy was well fit for a particular situation

The use of autonomous robots in our society is increasing every day and a robot is no longer seen as a tool but as a team member. The robots are now working side by side with us and provide assistance during dangerous operations where humans otherwise are at risk. This development has in turn increased the need of robots with more human-awareness. Therefore, this master thesis aims at contributing to the enhancement of human-aware robotics. Specifically, we are investigating the possibilities of equipping autonomous robots with the capability of assessing and detecting activities in human teams. This capability could, for instance, be used in the robot's reasoning and planning components to create better plans that ultimately would result in improved human-robot teamwork performance. we propose to improve existing teamwork activity recognizers by adding intangible features, such as stress, motivation and focus, originating from human behavior models. Hidden markov models have earlier been proven very efficient for activity recognition and have therefore been utilized in this work as a method for classification of behaviors. In order for a robot to provide effective assistance to a human team it must not only consider spatio-temporal parameters for team members but also the psychological.To assess psychological parameters this master thesis suggests to use the body signals of team members. Body signals such as heart rate and skin conductance. Combined with the body signals we investigate the possibility of using System Dynamics models to interpret the current psychological states of the human team members, thus enhancing the human-awareness of a robot.<br>Användningen av autonoma robotar i vårt samhälle ökar varje dag och en robot ses inte längre som ett verktyg utan som en gruppmedlem. Robotarna arbetar nu sida vid sida med oss och ger oss stöd under farliga arbeten där människor annars är utsatta för risker. Denna utveckling har i sin tur ökat behovet av robotar med mer människo-medvetenhet. Därför är målet med detta examensarbete att bidra till en stärkt människo-medvetenhet hos robotar. Specifikt undersöker vi möjligheterna att utrusta autonoma robotar med förmågan att bedöma och upptäcka olika beteenden hos mänskliga lag. Denna förmåga skulle till exempel kunna användas i robotens resonemang och planering för att ta beslut och i sin tur förbättra samarbetet mellan människa och robot. Vi föreslår att förbättra befintliga aktivitetsidentifierare genom att tillföra förmågan att tolka immateriella beteenden hos människan, såsom stress, motivation och fokus. Att kunna urskilja lagaktiviteter inom ett mänskligt lag är grundläggande för en robot som ska vara till stöd för laget. Dolda markovmodeller har tidigare visat sig vara mycket effektiva för just aktivitetsidentifiering och har därför använts i detta arbete. För att en robot ska kunna ha möjlighet att ge ett effektivt stöd till ett mänskligtlag måste den inte bara ta hänsyn till rumsliga parametrar hos lagmedlemmarna utan även de psykologiska. För att tyda psykologiska parametrar hos människor förespråkar denna masteravhandling utnyttjandet av mänskliga kroppssignaler. Signaler så som hjärtfrekvens och hudkonduktans. Kombinerat med kroppenssignalerar påvisar vi möjligheten att använda systemdynamiksmodeller för att tolka immateriella beteenden, vilket i sin tur kan stärka människo-medvetenheten hos en robot.<br><p>The thesis work was conducted in Stockholm, Kista at the department of Informatics and Aero System at Swedish Defence Research Agency.</p>

The principal problem is the control of a nonlinear system with uncertainty. We will consider a robot manipulator system, which is nonlinear, and uncertain (unknown parameters and modeling errors). Our goals are to come up with a design of controllers that insure the stability of the system and provide robustness to parameters changes and modeling errors. We will use the theory developed for uncertain linear systems after carrying out an exact linearization of the original system. This linearization which is not an approximation, has been recently developed. The linear part of the controller has been designed so as to guarantee tracking and disturbance rejection. However, additional constraints resulting from the original nonlinear system have to be taken care of. Our design is tested by simulation on a two degree of freedom robot manipulator, which is simple enough to simulate but has all the properties of more general manipulators.

This thesis covers two topics of robotic grasping: stable grasping and semantic grasping. The first part of the thesis is dedicated to the stable grasping problem, where we focus on a grasping pipeline that robustly executes a planned-to-be stable grasp under uncertainty. To this end, we first present a learning method which estimates the stability of a grasp based on tactile feedback and hand kinematic data. We then show our hand adjustment algorithm which works with the grasp stability estimator and synthesizes hand adjustments to optimize a grasp towards a stable one. With these two methods, we obtain a grasping pipeline with a closed-loop grasp adjustment process which increases the grasping performance under uncertainty. The second part of the thesis considers how robotic grasping should be accomplished to facilitate a manipulation task that follows the grasp. Certain task-related constraints should be satisfied by the grasp in use, which we refer to as semantic constraints. We first develop an example-based method to encode semantic constraints and to plan stable grasps according to the encoded semantic constraints. We then design a task description framework to abstract an object manipulation task. Within this framework, we also present a method which could automatically construct this manipulation task abstraction from a human demonstration.

This paper provides an analysis ofhow technostress influences the technological acceptanceof machine feedback systems and tools.To testthe relationship, data was collected witha survey, in which the participants were introducedto a workplace scenario which utilizes an Artificial Intelligence-enabled machine feedback system. The results of the executed analysis (N = 286) suggestthat technostress, especially techno-complexity,has asignificantinverse relationship with technology acceptance’core construct perceived ease of use. For thesecond core construct, perceived usefulness, no relationships with technostress have been identified. From these results, this paper derives managerial implications.

碩士<br>國立雲林科技大學<br>電機工程系<br>103<br>In this thesis, we mainly probe into the design of an active noise controller (ANC).The ANC system is equipped with a functional link artificial neural network (FLANN) structure in which filtered-s least mean square (FSLMS) algorithm is used. To improve system performance, we propose a rear adaptive decision feedback functional link artificial neural network (RADF-FLANN) for more efficiently updating the convergence factor of the neural network. An evolutionary mechanism for the learning convergence factor has been added to the RADF-FLANN structure. In this mechanism, learning factor will be self-adjusted. It not only makes the weight updating process smoother, but also helps the system achieve desired status of convergence. Moreover, a return mechanism for convergence factor has been included in the algorithm. Basically, the return mechanism keeps analyzing the residual noise and sends its statistics back for further processing. If the statistical properties of residual noise change dramatically, system will force the convergence factor back to its original value and start the re-learning process. Experiment results show that, when the noise signal changes suddenly, it takes only about 5000 iterations for RADF-FLANN to return stability. Conversely, DF-FLANN architecture may become divergent under the same situation. In addition, we have noticed that the anti-noise signal may sometimes be unpredictably instable when the learning algorithm is not yet stabilized. This phenomenon generally produces makes unnecessary peaks in residual. To suppress the instable signal, we have added an output limiter in our proposed architecture. For example, when the peak values of initial residual noise in a system without limiter is in range of ±20, we can suppress residual noise wave within to be ±10. Experiment results confirmed that an output limiter can suppress the initial instability signal to half amount of the original. In addition, the output limiter will destroy integrity of residual noise waveform. So we propose the concept of double cancellation points. It makes the system able to use the output limiter without destroying the waveform of residual noise.

碩士<br>國立雲林科技大學<br>電機工程系<br>102<br>In this thesis, we mainly probe into the design of an active noise controller (ANC). The principle of an ANC is based on the superposing of two destructively interfered waves to subdue the noise. That is, a sound wave of equal energy and 180° phase difference to the noise source is produced to cancel the original noise wave. The ANC system is equipped with a functional link artificial neural network (FLANN) structure in which filtered-s least mean square (FSLMS) algorithm is used. To improve the performance of the system, decision feedback (DF) mechanisms are commonly used in the FLANN structure. Although the DF-FLANN structure has better performance than FLANN, its complexity is also much higher. So we propose a simple structure which is named as RDF-FLANN, but the RDF-FLANN structure has worse performance than DF-FLANN structure almost. In this thesis, we use DF-FLANN structure foundation for improvement, decision a fast convergence, high performance architecture, known ODF-FLANN structure, And DF-FLANN structure differences, one amplified using trigonometric functions in the function expansion, and the second adding adaptive learning rate. From the simulation results of the MAE (mean absolute error) value of the performance checks ODF-FLANN structure, we put the amount of noise suppression is less than 1 as performance benchmark for the convergence, using trigonometric expand from the 22 iterations remaining after the noise is less than 1, and began to stabilize, compared to 374 iterations after FLANN structure, DF-FLANN structure after 94 iterations and RDF-FLANN structure 125 remaining after the noise is less than 1, and its to improve the original algorithm significantly more stable and faster convergence. In addition, we also consider a sudden change in the size of the noise sources, a sudden change in the frequency of the noise sources,and noise source is the complex frequency noise. Other issues on the ODF-FLANN structure effects and in different sizes to verify the effectiveness of the noise source and improving results.

碩士<br>國立雲林科技大學<br>電機工程系<br>103<br>In computer science, front-end and rear-end respectively stand for the beginning and the end of processes, that is, the direct communications with users and the outputs of results. In this thesis, the adaptive mechanism is titled “front-end” because it basically makes judgments and corrections on input ends, and the results are then used for further noise control. We have proposed a modified system based on the adaptive concepts from DF-FLANN (decision-feedback functional link artificial neural network) and ODF-FLANN (optimized DF-FLANN) structures. The system not only deals with input signals on both linear and nonlinear considerations, but also includes a noise reduction mechanism for suppressing noise peaks at the start. For the above reasons, we call this proposed system a front-end advanced adaptive DF-FLANN (FAADF-FLANN). The purpose of this thesis is to improve the data types of DF-FLANN architecture. Specifically, we have shortened its computation time, and added the noise reduction mechanism of the second channel so that the noise peaks can be lessened. The final goal is then to make adaptive learning more accurate. From experiment results, we observe that the processing time of FAADF-FLANN on 1000 input noise samples is about 40 seconds faster than DF-FLANN. As the mounted of data increases, the time difference also increases substantially. For example, the time difference can be up to 600 seconds when 20000 data samples are processed. Another advantage is that the secondary channel output values shrinks to within 2 to 4 times of the original noise amplitude, while for DF-FLANN it is about 6 times. Most important of all, adaptive learning associated with front-end judgment mechanism makes FAADF-FLANN not only work well at different settings of μ values, but also achieve better learning and convergence conditions. From these results, we conclude that FAADF have obvious advantages over DF-FLANN.

碩士<br>立德管理學院<br>資源環境研究所<br>94<br>Recently, the rapid development of Taiwan’s industry has not only brought the growth of economy and the prosperity of our society, but also exposed the potential environmental crisis gradually. Treatment with trash incinerator has become the trend of dealing with trash in Taiwan in recent years. By the end of 2005, there have been twenty incinerators in the country. Due to the difficulties in sorting trash thoroughly, it is relatively hard to control the incinerators stably. This study uses the back-propagation Neural Network to predict the NOx and CO in the flue gas emitting from the incinerators and uses the sensitivity analysis as the feedback of control system, with the expectation to get the most efficient control on the minimum release of the waste air. The sensitivity analysis was performed to find the correlationship among the controllable variables of incineration system and to get optimal operating values that were helpful to control the incinerator. The biggest factor to influence NOx is the feeding rate of ammonia solution . The second one is the water-supply rate of boiler, i.e., the loading rate of the boiler. The third one is the oxygen concentration in the flue gas, which is also related to the flow rate of primary air and the pH value in the pool of neutralization. .The biggest factor to influence CO is the oxygen in the flue gas. The second one is the pressure of the primary air. The third one is the water content in the flue gas, which is also related to the flow rate of primary air and the loading rate of the boiler. The incinerator system has played an important role on the problems of the air environment, such as acid precipitation, photochemical smog, the decrease of visibility, greenhouse effect…etc. In order to reduce the treatment, operation and maintenance cost of the flue-gas pollutant, this study aims to find out the relationship between different operational conditions and the NOx and CO concentrations in the flue gas. According to the effluent concentrations of NOx and CO, we can change the operational conditions to reach the effluent standard and make sure that the efficiency of the incinerator system is in the best conditions.

abstract: For this master's thesis, an open learner model is integrated with Quinn, a teachable robotic agent developed at Arizona State University. This system is represented as a feedback system, which aims to improve a student’s understanding of a subject. It also helps to understand the effect of the learner model when it is represented by performance of the teachable agent. The feedback system represents performance of the teachable agent, and not of a student. Data in the feedback system is thus updated according to a student's understanding of the subject. This provides students an opportunity to enhance their understanding of a subject by analyzing their performance. To test the effectiveness of the feedback system, student understanding in two different conditions is analyzed. In the first condition a feedback report is not provided to the students, while in the second condition the feedback report is provided in the form of the agent’s performance.<br>Dissertation/Thesis<br>Masters Thesis Computer Science 2016

La biologie moléculaire et, plus spécifiquement, la régulation de l’expression génique ont été révolutionnées par la découverte des microARN (miARN). Ces petits ARN d’une vingtaine de nucléotides sont impliqués dans la majorité des processus cellulaires et leur expression est dérégulée dans plusieurs maladies, comme le cancer. Un miARN reconnaît ses cibles principalement par son noyau, ce qui lui permet de réguler simultanément la traduction de centaines d’ARN messagers. Nos travaux ont montré l’existence d’une boucle de rétro-activation négative, entre deux miARN du polycistron miR-17-92 et trois facteurs de transcription de la famille E2F. E2F1, 2 et 3 induisent la transcription de miR-20 et miR-17 qui par la suite inhibent leur traduction. Nos résultats suggèrent l’implication de cette boucle dans la résistance à l’apoptose induite par E2F1 dans les cellules du cancer de la prostate, ce qui expliquerait en partie le potentiel oncogénique du polycistron miR-17-92. L’étude de ce motif de régulation nous a donc permis de réaliser le potentiel incroyable qu’ont les miARN à inhiber la traduction de plusieurs gènes. Basé sur les règles de reconnaissance des miARN, nous avons développé et validé MultiTar. Cet outil bioinformatique permet de trouver la séquence d’un miARN artificiel ayant le potentiel d’inhiber la traduction de gènes d’intérêts choisis par l’utilisateur. Afin de valider MultiTar, nous avons généré des multitargets pouvant inhiber l’expression des trois E2F, ce qui nous a permis de comparer leur efficacité à celle de miR-20. Nos miARN artificiels ont la capacité d’inhiber la traduction des E2F et de neutraliser leur fonction redondante de la progression du cycle cellulaire de façon similaire ou supérieur à miR-20. La fonctionnalité de notre programme, ouvre la voie à une stratégie flexible pouvant cibler le caractère multigénique de différents processus cellulaires ou maladies complexes, tel que le cancer. L’utilisation de miARN artificiels pourrait donc représenter une alternative intéressante aux stratégies déjà existantes, qui sont limitées à inhiber des cibles uniques. En plus d’élucider un réseau de régulation complexe impliquant les miARN, nous avons pu tirer profit de leur potentiel d’inhibition par la conception de miARN artificiels.<br>miRNAs are powerful regulators of gene expression in mammals. These small RNAs of around 20 nucleotides are involved in several cellular processes and diseases. MiRNAs recognize their targets mainly by a region comprising nucleotides 2-8, known as the seed. This characteristic gives them the potential to inhibit hundreds of messenger RNAs. Our first goal was to better characterize the complex network involving miRNAs in the regulation of gene expression. To achieve this, we studied the relation between a family of transcription factors, the E2Fs, and a family of miRNAs, the miR-17-92 cluster. Our results suggest a negative feedback loop involving miR-17, miR-20a, E2F1, E2F2 and E2F3. In this loop E2F1, 2 and 3 activate the transcription of the two miRNAs that inhibit their translation in return. The inhibition of the antiapoptotic function of E2F1 by miR-17 and miR-20 in a prostate cancer context, could explain the oncogenic potential of the miR-17-92 cluster that was previously reported. Studying the miR-20/E2F feedback loop made us realize how powerful was the ability of miRNAs to inhibit several targets. To overcome the lack of efficient tools able to inhibit simultaneously the expression of multiple genes, our second goal was to develop MultiTar, an algorithm able to design artificial miRNAs that target a set of predetermined genes. MultiTar was validated in silico, using known targets of endogenous miRNAs and in vivo, taking advantage of our experience with the E2F context. We designed artificial miRNAs against E2F1-3 and expressed them both in normal human fibroblasts and prostate cancer cells where they inhibited cell proliferation and induced cellular senescence. The observed phenotypes were precisely those known for inhibiting E2F activities. Hence, MultiTar can efficiently design artificial micro RNAs able to target multiple genes and is thus a flexible tool that can address the issue of multigenic diseases and complex cellular processes. The use of multitargets could be an alternative to overcome the limits of drugs or siRNAs that are designed generally to regulate only one target.

Indiana University-Purdue University Indianapolis (IUPUI)<br>This thesis proposes an automated grid-based alignment system utilizing lasers and an array of light-detecting photodiodes. The intent is to create an inexpensive and scalable alignment system for pick-and-place robotic systems. The system utilizes the transformation matrix, geometry, and trigonometry to determine the movements to align the robot with a grid-based array of photodiodes. The alignment system consists of a sending unit utilizing lasers, a receiving module consisting of photodiodes, a data acquisition unit, a computer-based control system, and the robot being aligned. The control system computes the robot movements needed to position the lasers based on the laser positions detected by the photodiodes. A transformation matrix converts movements from the coordinate system of the grid formed by the photodiodes to the coordinate system of the robot. The photodiode grid can detect a single laser spot and move it to any part of the grid, or it can detect up to four laser spots and use their relative positions to determine rotational misalignment of the robot. Testing the alignment consists of detecting the position of a single laser at individual points in a distinct pattern on the grid array of photodiodes, and running the entire alignment process multiple times starting with different misalignment cases. The first test provides a measure of the position detection accuracy of the system, while the second test demonstrates the alignment accuracy and repeatability of the system. The system detects the position of a single laser or multiple lasers by using a method similar to a center-of-gravity calculation. The intensity of each photodiode is multiplied by the X-position of that photodiode. The summed result from each photodiode intensity and position product is divided by the summed value of all of the photodiode intensities to get the X-position of the laser. The same thing is done with the Y-values to get the Y-position of the laser. Results show that with this method the system can read a single laser position value with a resolution of 0.1mm, and with a maximum X-error of 2.9mm and Y-error of 2.0mm. It takes approximately 1.5 seconds to process the reading. The alignment procedure calculates the initial misalignment between the robot and the grid of photodiodes by moving the robot to two distinct points along the robot’s X-axis so that only one laser is over the grid. Using these two detected points, a movement trajectory is generated to move that laser to the X = 0, Y = 0 position on the grid. In the process, this moves the other three lasers over the grid, allowing the system to detect the positions of four lasers and uses the positions to determine the rotational and translational offset needed to align the lasers to the grid of photodiodes. This step is run in a feedback loop to update the adjustment until it is within a permissible error value. The desired result for the complete alignment is a robot manipulator positioning within ±0.5mm along the X and Y-axes. The system shows a maximum error of 0.2mm in the X-direction and 0.5mm in the Y-direction with a run-time of approximately 4 to 5 minutes per alignment. If the permissible error value of the final alignment is tripled the alignment time goes down to 1 to 1.5 minutes and the maximum error goes up to 1.4mm in both the X and Y-directions. The run time of the alignment decreases because the system runs fewer alignment iterations.