ANP; FINEP; MCT; CAPES / Este trabalho objetiva otimizar a localização de um robô móvel para a inspeção dos tanques de armazenamento de GLP -- Gás Liquefeito de Petróleo -- e permitir a navegação autônoma deste robô por toda a superfície do tanque. A falta de marcos no interior dos tanques levou ao desenvolvimento de uma solução envolvendo detecção de pequenas estruturas com sensores LIDAR (Light Detection and Ranging - Detecção e Telemetria por Luz) aplicada aos cordões de solda, estes sim presentes no interior do tanque, que é apresentada na primeira parte do trabalho. Em seguida, aplicando uma técnica de fusão de dados, as diferentes fontes de odometria presentes no robô são combinadas, permitindo uma precisão mais elevada na inspeção de modo geral. Por fim, o mapeamento e navegação simultâneos do exterior da esfera é abordado, a fim de se adicionar uma camada suplementar ao mapa digital, indicando os locais onde existem falhas. Testes para validação de cada uma das técnicas foram efetuados e uma análise de desempenho é apresentada ao final do trabalho. / This work aims to optimize localization of a climbing inspection robot for spherical LPG -- Liquified Petroleum Gas -- tanks and allow autonomous navigation along the entire surface of these tanks. One solution envolving small structures detection using LIDAR sensors is applied to the weld beads that are present inside the tanks. This solution is developed on the first part of this work. Following, a data fusion technique is used to combine the diferent odometry sources on the robot, resulting in a better, higher precision on the inspection as a whole. Finnaly, simultaneous mapping and navigation on the exterior of the spherical tank was studied in order to add one extra layer to the digital map, pinpointing the places where failures and weld beams were found. Validation tests for each one of this techniques are carried out and a performance analysis is also documented herein.

ANP; FINEP; MCT; CAPES / Este trabalho objetiva otimizar a localização de um robô móvel para a inspeção dos tanques de armazenamento de GLP -- Gás Liquefeito de Petróleo -- e permitir a navegação autônoma deste robô por toda a superfície do tanque. A falta de marcos no interior dos tanques levou ao desenvolvimento de uma solução envolvendo detecção de pequenas estruturas com sensores LIDAR (Light Detection and Ranging - Detecção e Telemetria por Luz) aplicada aos cordões de solda, estes sim presentes no interior do tanque, que é apresentada na primeira parte do trabalho. Em seguida, aplicando uma técnica de fusão de dados, as diferentes fontes de odometria presentes no robô são combinadas, permitindo uma precisão mais elevada na inspeção de modo geral. Por fim, o mapeamento e navegação simultâneos do exterior da esfera é abordado, a fim de se adicionar uma camada suplementar ao mapa digital, indicando os locais onde existem falhas. Testes para validação de cada uma das técnicas foram efetuados e uma análise de desempenho é apresentada ao final do trabalho. / This work aims to optimize localization of a climbing inspection robot for spherical LPG -- Liquified Petroleum Gas -- tanks and allow autonomous navigation along the entire surface of these tanks. One solution envolving small structures detection using LIDAR sensors is applied to the weld beads that are present inside the tanks. This solution is developed on the first part of this work. Following, a data fusion technique is used to combine the diferent odometry sources on the robot, resulting in a better, higher precision on the inspection as a whole. Finnaly, simultaneous mapping and navigation on the exterior of the spherical tank was studied in order to add one extra layer to the digital map, pinpointing the places where failures and weld beams were found. Validation tests for each one of this techniques are carried out and a performance analysis is also documented herein.

For the last twenty years, several assumptions have been expressed in the fields of information processing, neurophysiology and cognitive sciences. First, neural networks and their dynamical behaviors in terms of attractors is the natural way adopted by the brain to encode information. Any information item to be stored in the neural network should be coded in some way or another in one of the dynamical attractors of the brain, and retrieved by stimulating the network to trap its dynamics in the desired item’s basin of attraction. The second view shared by neural network researchers is to base the learning of the synaptic matrix on a local Hebbian mechanism. The third assumption is the presence of chaos and the benefit gained by its presence. Chaos, although very simply produced, inherently possesses an infinite amount of cyclic regimes that can be exploited for coding information. Moreover, the network randomly wanders around these unstable regimes in a spontaneous way, thus rapidly proposing alternative responses to external stimuli, and being easily able to switch from one of these potential attractors to another in response to any incoming stimulus. Finally, since their introduction sixty years ago, cell assemblies have proved to be a powerful paradigm for brain information processing. After their introduction in artificial intelligence, cell assemblies became commonly used in computational neuroscience as a neural substrate for content addressable memories. <p> <p>Based on these assumptions, this thesis provides a computer model of neural network simulation of a brain-like memory. It first shows experimentally that the more information is to be stored in robust cyclic attractors, the more chaos appears as a regime in the background, erratically itinerating among brief appearances of these attractors. Chaos does not appear to be the cause, but the consequence of the learning. However, it appears as an helpful consequence that widens the network’s encoding capacity. To learn the information to be stored, two supervised iterative Hebbian learning algorithm are proposed. One leaves the semantics of the attractors to be associated with the feeding data unprescribed, while the other defines it a priori. Both algorithms show good results, even though the first one is more robust and has a greater storing capacity. Using these promising results, a biologically plausible alternative to these algorithms is proposed using cell assemblies as substrate for information. Even though this is not new, the mechanisms underlying their formation are poorly understood and, so far, there are no biologically plausible algorithms that can explain how external stimuli can be online stored in cell assemblies. This thesis provide such a solution combining a fast Hebbian/anti-Hebbian learning of the network's recurrent connections for the creation of new cell assemblies, and a slower feedback signal which stabilizes the cell assemblies by learning the feed forward input connections. This last mechanism is inspired by the retroaxonal hypothesis. <p> / Doctorat en Sciences / info:eu-repo/semantics/nonPublished

Οι ηλεκτροµηχανικές συσκευές αποθήκευσης δεδοµένων πολύ µικρής κλίµακας που βασίζονται στη χρήση ανιχνευτών (probes) αποτελούν ανερχόµενες εναλλακτικές επιλογές για τη βελτίωση της πυκνότητας αποθήκευσης, του χρόνου πρόσβασης των δεδοµένων και της απαιτούµενης ισχύος σε σχέση µε τις συµβατικές αποθηκευτικές συσκευές. Μία υλοποίηση µιας τέτοιας συσκευής χρησιµοποιεί θερµοµηχανικές µεθόδους για την αποθήκευση πληροφορίας σε λεπτές µεµβράνες πολυµερών υλικών. Σε αυτή την περίπτωση, η ψηφιακή πληροφορία αποθηκεύεται µε τη µορφή κοιλωµάτων πάνω στο πολυµερές υλικό, οι οποίες δηµιουργούνται από τις άκρες των ανιχνευτών διαµέτρου µερικών nm. Με στόχο την αύξηση του ρυθµού εγγραφής και ανάγνωσης χρησιµοποιούνται διατάξεις από ανιχνευτές που λειτουργούν παράλληλα, µε κάθε ανιχνευτή να εκτελεί λειτουργίες εγγραφής/ανάγνωσης/διαγραφής σε ξεχωριστό αποθηκευτικό πεδίο. Βασικές απαιτήσεις κατά τη λειτουργία τέτοιων συσκευών αποτελούν η εξαιρετικά µεγάλη ακρίβεια και η µικρή καθυστέρηση κατά τη µετακίνηση των ανιχνευτών πάνω από το πολυµερές υλικό. Η παρούσα διατριβή έχει ως αντικείµενο τη µελέτη των διατάξεων κίνησης και τη σχεδίαση πρωτότυπων αρχιτεκτονικών ελέγχου, που οδηγούν στη βελτίωση της απόδοσης των απαιτούµενων, σε συσκευές τέτοιου τύπου, λειτουργιών ελέγχου. Η µετατόπιση του αποθηκευτικού µέσου σε σχέση µε τη διάταξη των ανιχνευτών επιτυγχάνεται µε τη χρησιµοποίηση µικρής κλίµακας scanners, που έχουν δυνατότητες κίνησης σε δύο κατευθύνσεις (x/y). Πληροφορία για τη θέση του microscanner στις δύο κατευθύνσεις παρέχεται από θερµικούς αισθητήρες ανίχνευσης θέσης που κατασκευάζονται µαζί µε τη διάταξη µε τους ανιχνευτές και τοποθετούνται πάνω από το κινητό πλαίσιο. Η πλήρης κατανόηση της συµπεριφοράς των διατάξεων αυτών αποτελεί απαραίτητο στοιχείο για τον αποτελεσµατικό σχεδιασµό και την ανάλυση των συστηµάτων ελέγχου. Στα πλαίσια της διατριβής δηµιουργήθηκε ένα πλήρες µοντέλο της διάταξης του microscanner και των θερµικών αισθητήρων ανίχνευσης θέσης. Σύγκριση της απόκρισης του µοντέλου µε τις πειραµατικές µετρήσεις καταδεικνύει ότι το µοντέλο προσεγγίζει µε εξαιρετική ακρίβεια την απόκριση του συστήµατος. Το σύστηµα ελέγχου περιλαµβάνει, στην αρχή, τη λειτουργία αναζήτησης/ αποκατάστασης, κατά την οποία το σύστηµα εντοπίζει τη θέση όπου απαιτείται να πραγµατοποιηθεί εγγραφή ή ανάγνωση πληροφορίας µε εκκίνηση µία αυθαίρετη θέση του κινητού πλαισίου. Απαίτηση του συστήµατος κατά τη λειτουργία αυτή είναι η ελαχιστοποίηση του χρόνου πρόσβασης των δεδοµένων. Η γρήγορη πρόσβαση στα δεδοµένα αποτελεί µια σηµαντική πρόκληση στις συµβατικές αποθηκευτικές συσκευές. Με το πλεονέκτηµα των ελαφρύτερων µηχανικών µερών, οι υπό µελέτη συσκευές αποθήκευσης βασισµένες στην τεχνολογία MEMS θεωρούνται βασικές υποψήφιες για τη βελτίωση του χρόνου πρόσβασης των δεδοµένων. Οι σχετικές προοπτικές των συσκευών αυτών διερευνώνται αναλυτικά στα πλαίσια της διατριβής. Συγκεκριµένα, αρχικά µελετάται η απόδοση διαφόρων συστηµάτων µε βάση τη θεωρία ελέγχου βέλτιστου χρόνου. Τα αποτελέσµατα της µελέτης δίνουν το θεωρητικά βέλτιστο χρόνο πρόσβασης και την εξάρτησή του από τις παραµέτρους του κάθε συστήµατος. Στη συνέχεια, περιγράφεται η αρχιτεκτονική ελέγχου για τη λειτουργία αναζήτησης και παρουσιάζονται τα αποτελέσµατα που αντλήθηκαν από το περιβάλλον προσοµοίωσης και από την πειραµατική διάταξη. Τα αποτελέσµατα καταδεικνύουν ότι οι χρόνοι πρόσβασης των δεδοµένων που είναι δυνατό να επιτευχθούν µε τις συσκευές αυτές, είναι σηµαντικά µικρότεροι σε σχέση µε τις συµβατικές. Στη συνέχεια, ακολουθεί η λειτουργία παρακολούθησης, όπου η θέση των ανιχνευτών πρέπει να παραµένει στο κέντρο του επιθυµητού καναλιού, κατά τη διάρκεια εγγραφής/ανάγνωσης των δεδοµένων. Η απαίτηση για µεγάλη ακρίβεια στη µετακίνηση πάνω από τη νοητή γραµµή του κέντρου του καναλιού, της µίας ή περισσότερων κεφαλών που χρησιµοποιούνται κατά την εγγραφή/ανάγνωση, είναι σηµαντική για όλους τους τύπους αποθηκευτικών συσκευών. Οι απαιτήσεις για ακρίβεια γίνονται ακόµα πιο µεγάλες και κρίσιµες, στην περίπτωση των υπό µελέτη αποθηκευτικών συσκευών, όπου η ψηφιακή πληροφορία αποθηκεύεται σε µία περιοχή µε µέγεθος µερικών nm. Το σύστηµα ελέγχου, κατά τη λειτουργία αυτή, οφείλει να παρακολουθεί το επιθυµητό σήµα αναφοράς, και ταυτόχρονα να έχει ικανοποιητική απόρριψη των διαταραχών και να επιτυγχάνει την απαιτούµενη ακρίβεια ως προς τον προσδιορισµό της θέσης. Παράλληλα, σηµαντικό παράγοντα βελτιστοποίησης αυτής της λειτουργίας, αποτελεί ο ρυθµός εγγραφής/ανάγνωσης των δεδοµένων. Η πρώτη προσέγγιση για την αρχιτεκτονική ελέγχου, κατά τη λειτουργία αυτή, βασίζεται στην παρεχόµενη από τους θερµικούς αισθητήρες ανίχνευσης, πληροφορία της θέσης του microscanner. Η αρχιτεκτονική βασίζεται στον αλγόριθµο του γραµµικού τετραγωνικού ρυθµιστή (LQG) και η αξιολόγησή της γίνεται µε κριτήρια την ικανότητα παρακολούθησης της εισόδου, την απόρριψη των διαταραχών και την ακρίβεια ως προς τον προσδιορισµό της θέσης. Τα αποτελέσµατα που εξήχθησαν, κατά την υλοποίηση της αρχιτεκτονικής ελέγχου στην πειραµατική διάταξη, αναδεικνύουν ότι η αρχιτεκτονική πληρεί τις απαιτήσεις και η ακρίβεια µερικών nm που επιτυγχάνεται στον προσδιορισµό της θέσης επιτρέπει την αξιόπιστη εγγραφή και κατόπιν ανάγνωση δεδοµένων από την αποθηκευτική συσκευή. Μειονέκτηµα της παραπάνω προσέγγισης αποτελεί ο χαµηλής συχνότητας θόρυβος των θερµικών αισθητήρων, που επηρεάζει τη σωστή λειτουργία του κλειστού συστήµατος σε µεγάλες περιόδους λειτουργίας της συσκευής. Το πρόβληµα αυτό επιλύεται µε µία πρωτότυπη προσέγγιση που αναπτύχθηκε στα πλαίσια της διατριβής και βασίζεται στην πληροφορία, την προερχόµενη από τους θερµικούς αισθητήρες ανίχνευσης θέσης, σε συνδυασµό µε το προερχόµενο από το αποθηκευτικό µέσο σήµα σφάλµατος θέσης. Ο σχεδιασµός του συστήµατος ελέγχου, στην περίπτωση αυτή, εκµεταλλεύεται την εκ των προτέρων γνώση των χαρακτηριστικών θορύβου ως προς τη συχνότητα των δύο αισθητήρων ανίχνευσης θέσης, έτσι ώστε το σύστηµα ελέγχου που προκύπτει να χρησιµοποιεί την πιο αξιόπιστη µέτρηση σε κάθε περιοχή συχνοτήτων. Το πλαίσιο του σθεναρού ελέγχου, H∞, χρησιµοποιείται κατά το σχεδιασµό αυτής της αρχιτεκτονικής ελέγχου, µε διαχωρισµό ως προς τη συχνότητα. Με χρήση αυτής της µεθόδου, το σύστηµα ελέγχου δεν επηρεάζεται από τον χαµηλής συχνότητας θόρυβο των θερµικών αισθητήρων. Τα αποτελέσµατα που εξήχθησαν κατά την υλοποίηση της αρχιτεκτονικής ελέγχου στην πειραµατική διάταξη επιβεβαιώνουν τα παραπάνω. Η µέθοδος αυτή είναι πιο γενική και µπορεί να εφαρµοστεί σε κάθε πρόβληµα ελέγχου, που έχει δύο ή και περισσότερους αισθητήρες µε διαφορετικά χαρακτηριστικά απόδοσης σε διαφορετικές περιοχές συχνοτήτων. / Micro-electro-mechanical-system (MEMS)-based scanning-probe storage devices are emerging as potential ultra-high-density, low-access-time, and low-power alternatives to conventional data storage. One implementation of probe-based storage uses thermomechanical means to store and retrieve information in thin polymer films. Digital information is stored by making indentations on the thin polymer film with the tips of atomic force microscope (AFM) cantilevers, which are a few nanometers in diameter. To increase the data rate, an array of probes is used, in which each probe performs read/write/erase operations over an individual storage field. One of the primary challenges in building such devices is the extreme accuracy and the short latency required in the navigation of the probes over the polymer medium. This dissertation describes the design of novel control architectures and the characterization of their performance. The associated modelling effort, theoretical analysis, simulation work and experimental results are presented. Displacement of the storage medium relative to the array of cantilevers is achieved by using silicon-based micro-scanners with x/y-displacement capabilities. The x/y positional information can be provided by thermal position sensors that are fabricated on the cantilever-array chip and positioned directly above the scan table. A thorough understanding of the dynamics of these parts of the device is essential for effective design and analysis of the control architectures. In this dissertation a complete model of the micro-scanner and the thermal position sensors was developed. Comparison of the model response with the experimental data have shown that the model approximates the system response with an excellent accuracy. In general, the servo system in such a storage device has two functions. First, it locates the target track to which information is to be written or read back from, starting from an arbitrary initial position of the scan table carrying the storage medium. This is achieved by the so-called seek-and-settle procedure. The data access time depends on the duration of this operation, and therefore the minimization of its duration constitutes an important optimization factor. The speed of data access is a significant bottleneck in today’s computing systems. With the advantage of the lighter moving stage MEMS-based storage devices are widely touted to improve access times. In this dissertation these perspectives are examined in detail. Initially the time-optimal control theory has been studied for different system models and their performance has been examined regarding the optimal access time. The results of this study have provided the theoretically optimal access time for each model and its dependence on the system parameters. The control architecture for the seek operation has been designed. The simulation and experimental results show that the possible access times that can be achieved are significantly smaller than the conventional storage devices. The second function of the control system is to maintain the position of the read/write probes on the centre of the target track as they are being scanned along the length of this track during the normal read/write operation. This is achieved by the so-called track-follow procedure. Precise positioning and navigation of the read/write head(s) on the track centerlines is of paramount importance in all types of storage devices. The requirements become more crucial in the devices under study, where in order to achieve reliable storage and retrieval of data, accuracy in the order of a few nanometers in the scanner motion is needed. Therefore, the tracking of the reference signal, the disturbance rejection capabilities and the positioning resolution are considered as performance measures for the control system in this operation. Similarly, the read/write data rate constitutes an important optimization factor for this operation. The first approach of the control architecture for the track-follow procedure uses the position information from the thermal sensors. The control of the position in the x/y directions is realized using two independent feedback loops and each controller is based on the linear quadratic Gaussian regulator (LQG). For the evaluation of the proposed control architecture a detailed analysis has been performed in terms of the tracking performance, the disturbance rejection and the positioning resolution. The proposed architecture has been implemented in the experimental set-up and the analytical results are in agreement with those obtained experimentally. The experimental results show that the accuracy in the motion of the micro-scanner obtained with the proposed control architecture allows reliable storage and retrieval of data in the storage device. The disadvantage of the above control scheme originates from the low frequency noise of the thermal sensors that affects the closed loop performance for long term operation of the device. A novel control architecture was developed that addresses this problem by using medium-derived position information along with the thermal positioning sensor. The objective of this method is, using the a priori knowledge of the noise characteristics of the two sensors, to create a control structure that utilizes the best measurement in different frequency regions. The framework of the H∞ robust control was used for the design of this new frequency separated control architecture. Using this method the control system is not affected from the low frequency noise of the thermal sensors. The experimental results validate the performance of the proposed method. The developed methodology is more general and can be applied to any control problem that has two or more sensors with different performance characteristics in different frequency regions.

A crescente demanda por ambientes de virtualização de larga escala, como os usados em datacenters e nuvens computacionais, faz com que seja necessário um gerenciamento eficiente dos recursos computacionais utilizados. Um dos recursos mais exigidos nesses ambientes é a memória RAM, que costuma ser o principal fator limitante em relação ao número de máquinas virtuais que podem executar sobre o mesmo host físico. Recentemente, hipervisores trouxeram mecanismos de compartilhamento transparente de memória RAM entre máquinas virtuais, visando diminuir a demanda total de memória no sistema. Esses mecanismos “fundem” páginas idênticas encontradas nas várias máquinas virtuais em um mesmo quadro de memória física, usando uma abordagem copy-on-write, de forma transparente para os sistemas convidados. O objetivo deste estudo é apresentar uma visão geral desses mecanismos e também avaliar seu desempenho e efetividade. São apresentados resultados de experimentos realizados com dois hipervisores populares (VMware e KVM), usando sistemas operacionais convidados distintos (Linux e Windows) e cargas de trabalho diversas (sintéticas e reais). Os resultados obtidos evidenciam diferenças significativas de desempenho entre os hipervisores em função dos sistemas convidados, das cargas de trabalho e do tempo. / The growing demand for large-scale virtualization environments, such as the ones used in cloud computing, has led to a need for efficient management of computing resources. RAM memory is the one of the most required resources in these environments, and is usually the main factor limiting the number of virtual machines that can run on the physical host. Recently, hypervisors have brought mechanisms for transparent memory sharing between virtual machines in order to reduce the total demand for system memory. These mechanisms “merge” similar pages detected in multiple virtual machines into the same physical memory, using a copy-on-write mechanism in a manner that is transparent to the guest systems. The objective of this study is to present an overview of these mechanisms and also evaluate their performance and effectiveness. The results of two popular hypervisors (VMware and KVM) using different guest operating systems (Linux and Windows) and different workloads (synthetic and real) are presented herein. The results show significant performance differences between hypervisors according to the guest system workloads and execution time.

Com a popularização do uso da biometria, determinar a identidade de um indivíduo é uma atividade cada vez mais comum em diversos contextos: controle de acesso físico e lógico, controle de fronteiras, identificações criminais e forenses, pagamentos. Sendo assim, existe uma demanda crescente por Sistemas Automáticos de Identificação Biométrica (ABIS) cada vez mais rápidos, com elevada acurácia e que possam operar com um grande volume de dados. Este trabalho apresenta uma abordagem de implementação de uma central de confrontos para um ABIS de grande escala utilizando um framework de computação em memória. Foram realizados experimentos em uma base de dados real com mais de 50 milhões de impressões digitais em um cluster com até 16 nós. Os resultados mostraram a escalabilidade da solução proposta e a capacidade de operar em grandes bases de dados. / With the popularization of biometrics, personal identification is an increasingly common activity in several contexts: physical and logical access control, border control, criminal and forensic identification, payments. Thus, there is a growing demand for faster and accurate Automatic Biometric Identification Systems (ABIS) capable to handle a large volume of biometric data. This work presents an approach to implement a scalable cluster-based matching platform for a large-scale ABIS using an in-memory computing framework. We have conducted some experiments that involved a database with more than 50 million captured fingerprints, in a cluster up to 16 nodes. The results have shown the scalability of the proposed solution and the capability to handle a large biometric database.

No description available.