O objetivo desta dissertação é apresentar um estudo sobre o emprego de demonstrações em uma narrativa contada em língua de sinais brasileira (libras). Este trabalho parte da ideia de que a copresença de participantes na interação tem influência na maneira como a narrativa é expressa e interpretada (McCleary 2011; McCleary & Viotti 2014). A investigação dos elementos envolvidos na elaboração das demonstrações é feita a partir de uma perspectiva multimodal. Para tratar da multimodalidade em interações presenciais, tomei como base as propostas de Clark (1996) e Hutchins (2010), segundo as quais os participantes de uma interação precisam se coordenar para a realização de atividades conjuntas. Nessa perspectiva, demonstrar envolve crucialmente dois aspectos: i) a habilidade de se coordenar na imaginação conjunta dos elementos da narrativa, como os cenários, as personagens e suas ações, os eventos etc; e ii) o uso do corpo, de gestos de diferentes tipos e do espaço como ferramentas nesse ato imaginativo situado na interação. Para estudar demonstrações especificamente em discursos sinalizados, tomei como base Liddell (2003) e Dudis (2007), que analisaram demonstrações em discursos em língua de sinais americana (ASL), e McCleary & Viotti (2010, 2011, 2014), que analisaram narrativas em libras. A partir da aproximação entre os trabalhos desses autores, foi possível observar: i) as ocorrências de demonstrações em diferentes níveis discursivos; e ii) a relevância de considerar o nível do narrador nas análises de ocorrências de demonstração. A narrativa analisada, intitulada \"Bolinha de Ping Pong\", foi transcrita no software ELAN seguindo o modelo de transcrição proposto por McCleary, Viotti & Leite (2010). A análise trouxe evidências de que a demonstração é uma estratégia discursiva central em narrativas sinalizadas, e que contar histórias fluentemente envolve, em grande medida, a habilidade no uso de demonstração, integrada com outras estratégias narrativas. / This thesis aims at presenting a study of uses of demonstration in a narrative in Brazilian Sign Language (Libras). The study is based on the idea that the co-presence of the participants involved in any interaction influences the way in which the narrative is expressed and interpreted (McCleary 2011; McCleary & Viotti 2014). The analysis of the elements involved in the use of demonstrations along the narrative is based on a multimodal approach. In order to describe the multimodality in face-to-face interaction, I have used models put forward by Clark (1996) and Hutchins (2010). These authors suggest that participants in any interaction coordinate themselves to perform joint activities. According to that approach, demonstration crucially involves two aspects: i) the ability to coordinate the joint imagination of elements of the narrative such as scenarios, characters, actions, events, etc; and ii) the use of the body, different kinds of gesture and the space as tools during that construct the imaginative act in the interaction. For the study of demonstrations in signed discourses, specifically, I based my analyses on Liddell (2003) and Dudis (2007), who described the use of demonstration in discourses in American Sign Language, and on McCleary & Viotti (2010, 2011, 2014), who analyzed narratives in Libras. Based on these authors, it was possible to observe: i) the uses of demonstrations are different levels of narrative discourse; ii) the relevance of taking into account the narrator level to the analysis of uses of demonstration. The analyzed narrative entitled Ping Pong ball was transcribed using the software ELAN, following the model proposed in McCleary, Viotti & Leite (2010). The analysis implies that demonstration is a central discursive strategy in signed narratives, and that, to a great extent, fluency in signing stories involves the ability to use demonstrations, integrated with other narrative strategies.

"We address the problem of performing complex tasks for a robot operating in changing environments. We propose two approaches to the following problem: 1) define task-specific cost functions for motion planning that represent path quality by learning from an expert's preferences and 2) using constraint-based representation of the task inside learning from demonstration paradigm. In the first approach, we generate a set of paths for a given task using a motion planner and collect data about their features (path length, distance from obstacles, etc.). We provide these paths to an expert as a set of pairwise comparisons. We then form a ranking of the paths from the expert's comparisons. This ranking is used as training data for learning algorithms, which attempt to produce a cost function that maps path feature values to a cost that is consistent with the expert's ranking. We test our method on two simulated car-maintenance tasks with the PR2 robot: removing a tire and extracting an oil filter. We found that learning methods which produce non-linear combinations of the features are better able to capture expert preferences for the tasks than methods which produce linear combinations. This result suggests that the linear combinations used in previous work on this topic may be too simple to capture the preferences of experts for complex tasks. In the second approach, we propose to introduce a constraint-based description of the task that can be used together with the motion planner to produce the trajectories. The description is automatically created from the demonstration by performing segmentation and extracting constraints from the motion. The constraints are represented with the Task Space Regions (TSR) that are extracted from the demonstration and used to produce a desired motion. To account for the parts of the motion where constraints are different a segmentation of the demonstrated motion is performed using TSRs. The proposed approach allows performing tasks on robot from human demonstration in changing environments, where obstacle distribution or poses of the objects could change between demonstration and execution. The experimental evaluation on two example motions was performed to estimate the ability of our approach to produce the desired motion and recover a demonstrated trajectory."

"This thesis presents learning and interaction algorithms to support a human teaching hierarchical task models to a robot using a single or multiple examples in the context of a mixed-initiative interaction with bi-directional communication. Our first contribution is an approach for learning a high level task from a single example using the bottom-up style. In particular, we have identified and implemented two important heuristics for suggesting task groupings and repetitions based on the data flow between tasks and on the physical structure of the manipulated artifact. We have evaluated our heuristics with users in a simulated environment and shown that the suggestions significantly improve the learning and interaction. For our second contribution, we extended this interaction by enabling users to teaching tasks using the top-down teaching style in addition to the bottom-up teaching style. Results obtained in a pilot study show that users utilize both the bottom-up and the top-down teaching styles to teach tasks. Our third contribution is an algorithm that merges multiple examples when there are alternative ways of doing a task. The merging algorithm is still under evaluation. "

A partir do estudo de eventos de protesto relacionados ao tema do transporte público, tendo as mobilizações de 2013 e seus desdobramentos como objeto empírico central, busca-se analisar como, através de diferentes performances que conformam tais eventos, são produzidas inovações nos repertórios de contestação. Para responder a este problema, foram estudadas as performances desenvolvidas nas manifestações que ocorreram em Porto Alegre nas últimas décadas e que tiveram como principal reivindicação a tarifa do transporte público. A literatura sobre as manifestações no Brasil em 2013 não desenvolve análises mais específicas sobre as inovações nos repertórios e nas performances identificadas. A literatura sobre “repertório” e sobre “performance” traz alguns elementos importantes para a compreensão do processo de inovação, porém tende a não apresentar uma sistematização ou um modelo explicativo para abordar tal processo. A proposta desta pesquisa é, portanto, avançar na formulação de um modelo que sistematize o processo de inovação nos repertórios de contestação, tendo como objeto as transformações nas manifestações sobre o transporte público em Porto Alegre. Como resultados, a pesquisa apresenta três mecanismos importantes no processo de inovação: adaptação, experimentação interativa (para explicação da emergência das performances) e rotinização (para explicação da incorporação no repertório). A entrada de novos atores no processo de mobilização foi uma característica central em 2013. O mecanismo de adaptação se daria através da trajetória desses atores e de dinâmicas de difusão. O mecanismo de experimentação interativa diz respeito a improvisações não previstas na interação. A rotinização das performances acontece através das noções de eficácia, de identificação, pela importância das organizações e pelo diálogo que estabelecem com o público. Algumas dimensões importantes no processo de inovação, que não foram previstas, apareceram no decorrer da pesquisa: o contexto político institucional, as disputas interpretativas sobre o confronto e os impactos das Tecnologias de Informação e Comunicação (TICs), influenciando o processo de inovação nos repertórios de contestação. / This dissertation draws on previous studies on protest events related to the topic of public transport and particularly on studies that have as their main empirical object the Brazilian mobilizations in 2013 and their aftermath. It seeks to analyze how different performances that took place in such events produced innovations in contentious repertoires. To address this problem, we studied performances developed in demonstrations that took place in Porto Alegre in recent decades and that had as their main object the public transport fare. The literature on demonstrations in Brazil in 2013 does not develop more specific analysis of innovations in repertoires and performances. The literature on "repertoire" and "performance" indicates important elements for understanding the innovation process, but in general does not present a theoretical systematization or an explanatory model to explain this process. Therefore, the purpose of this research is to contribute for the construction of a model that systematizes the innovation process in protest repertoires. For this purpose, we define as our object the transformations in demonstrations focused on public transportation in Porto Alegre. Our results suggest that three important mechanisms affect innovation processes: adaptation, interactive experimentation (for explaining the emergence of performances), and routinization (for explaining the incorporation of repertoires). The arrival of new actors was a central feature in the mobilization process of 2013. The adaptation mechanism took place through the trajectory of these actors and diffusion dynamics. The interactive experimentation mechanism is about the unforeseen improvisations in interactions. The routinization of perfomances is mediated by: the notions of effectiveness and identification; the importance of organizations; and the dialogue established with the public. Some important dimensions of the innovation process, which were not foreseen, were identified during the research: the political and institutional context, the interpretative disputes over confrontation and the impact of Information and Communication Technologies.

O objetivo desta dissertação é apresentar um estudo sobre o emprego de demonstrações em uma narrativa contada em língua de sinais brasileira (libras). Este trabalho parte da ideia de que a copresença de participantes na interação tem influência na maneira como a narrativa é expressa e interpretada (McCleary 2011; McCleary & Viotti 2014). A investigação dos elementos envolvidos na elaboração das demonstrações é feita a partir de uma perspectiva multimodal. Para tratar da multimodalidade em interações presenciais, tomei como base as propostas de Clark (1996) e Hutchins (2010), segundo as quais os participantes de uma interação precisam se coordenar para a realização de atividades conjuntas. Nessa perspectiva, demonstrar envolve crucialmente dois aspectos: i) a habilidade de se coordenar na imaginação conjunta dos elementos da narrativa, como os cenários, as personagens e suas ações, os eventos etc; e ii) o uso do corpo, de gestos de diferentes tipos e do espaço como ferramentas nesse ato imaginativo situado na interação. Para estudar demonstrações especificamente em discursos sinalizados, tomei como base Liddell (2003) e Dudis (2007), que analisaram demonstrações em discursos em língua de sinais americana (ASL), e McCleary & Viotti (2010, 2011, 2014), que analisaram narrativas em libras. A partir da aproximação entre os trabalhos desses autores, foi possível observar: i) as ocorrências de demonstrações em diferentes níveis discursivos; e ii) a relevância de considerar o nível do narrador nas análises de ocorrências de demonstração. A narrativa analisada, intitulada \"Bolinha de Ping Pong\", foi transcrita no software ELAN seguindo o modelo de transcrição proposto por McCleary, Viotti & Leite (2010). A análise trouxe evidências de que a demonstração é uma estratégia discursiva central em narrativas sinalizadas, e que contar histórias fluentemente envolve, em grande medida, a habilidade no uso de demonstração, integrada com outras estratégias narrativas. / This thesis aims at presenting a study of uses of demonstration in a narrative in Brazilian Sign Language (Libras). The study is based on the idea that the co-presence of the participants involved in any interaction influences the way in which the narrative is expressed and interpreted (McCleary 2011; McCleary & Viotti 2014). The analysis of the elements involved in the use of demonstrations along the narrative is based on a multimodal approach. In order to describe the multimodality in face-to-face interaction, I have used models put forward by Clark (1996) and Hutchins (2010). These authors suggest that participants in any interaction coordinate themselves to perform joint activities. According to that approach, demonstration crucially involves two aspects: i) the ability to coordinate the joint imagination of elements of the narrative such as scenarios, characters, actions, events, etc; and ii) the use of the body, different kinds of gesture and the space as tools during that construct the imaginative act in the interaction. For the study of demonstrations in signed discourses, specifically, I based my analyses on Liddell (2003) and Dudis (2007), who described the use of demonstration in discourses in American Sign Language, and on McCleary & Viotti (2010, 2011, 2014), who analyzed narratives in Libras. Based on these authors, it was possible to observe: i) the uses of demonstrations are different levels of narrative discourse; ii) the relevance of taking into account the narrator level to the analysis of uses of demonstration. The analyzed narrative entitled Ping Pong ball was transcribed using the software ELAN, following the model proposed in McCleary, Viotti & Leite (2010). The analysis implies that demonstration is a central discursive strategy in signed narratives, and that, to a great extent, fluency in signing stories involves the ability to use demonstrations, integrated with other narrative strategies.

Approximately 25% of households in the U.S. treat their wastewater onsite using conventional onsite wastewater treatment systems (OWTS). These systems typically include a septic tank or a series of septic tanks followed by a soil absorption system. They effectively remove biochemical oxygen demand (BOD), total suspended solids (TSS), fats and grease but are not designed to remove significant amounts of nitrogen. High nitrogen loading to coastal and ground waters can be dangerous to aquatic life and public health. Hence, there is a need for advanced onsite wastewater treatment systems that can effectively remove nitrogen. Making enhanced nitrogen removal for OWTS as our primary goal, a laboratory scale Hybrid Adsorption and Biological Treatment Systems (HABiTS) was developed and upon observation of its effective nitrogen removal capacity, a pilot demonstration study with two side-by-side HABiTS, one with recirculation and one without recirculation (only forward flow) were constructed and tested at the Northwest Regional Water Reclamation Facility in Hillsborough County (Florida). HABiTS employ biological nitrogen removal and ion exchange for effective nitrogen removal. HABiTS is a two-stage process which uses nitrification for the oxidation of ammonium to nitrate and ion exchange for ammonium adsorption that helps buffer transient loading and also acts as a biofilm carrier in its stage 1 biofilter and it uses tire-sulfur hybrid adsorption denitrification (T-SHAD) in its stage 2 biofilter. These sulfur pellets help promote sulfur oxidation denitrification (SOD) and tire chips are used for nitrate adsorption during transient loading conditions, as biofilm carriers for denitrifying bacteria, and can also be used as organic carbon source to promote heterotrophic denitrification because they leach organic carbon. For this research, HABiTS without recirculation is considered as the control system and the performance of HABiTS with recirculation was tested for its ability to further enhance nitrogen removal from HABiTS. Nitrified effluent recirculation is a common strategy employed in wastewater treatment for enhanced nitrogen removal. It is the reintroduction of semi-treated wastewater to pass through an anoxic pre-treatment chamber to achieve better quality effluent. Recirculation is said to improve and consistently remove nitrogen at any hydraulic loading rate and/or nitrogen concentration. This is because of the dilution of high BOD septic tank effluent with nitrified effluent which lowers COD:TKN ratio and also improves mass transfer of substrates in the stage 1 biofilter. Recirculation also provides some pre-denitrification in the pre-treatment chamber, thereby reducing nitrogen load on the system. The HABiTS with recirculation (R) was run at 1:1 ratio of nitrified effluent recirculation rate to the influent flow rate for 50 days, and at 3:1 ratio for the remaining period of this research (200 days). The forward flow system (FF) was run under constant conditions throughout the research and comparisons between the two systems were made for different water quality parameters (pH, DO, conductivity, alkalinity, TSS, chemical oxygen demand (COD), total nitrogen (TN), total phosphorus (TP) and various nitrogen species). The final effluent ammonium results showed that the system with recirculation removed consistently > 80% NH4+-N during 1:1 and 3:1 recirculation ratios whereas the forward flow system achieved 57% removal. Further, an average of 81% total inorganic nitrogen (TIN) removal from the system influent was seen in the recirculation system’s final effluent when compared to an average of 55% in forward flow system’s final effluent. This research explains in detail, the impact of nitrified effluent recirculation on enhanced nitrogen removal in onsite systems and the results presented in this thesis proved that nitrified effluent recirculation provides promising enhanced nitrogen removal in an onsite wastewater treatment system.

Cette thèse étudie comment l'intégration du calcul dans les démonstrations peut les simplifier. Nous nous intéressons pour cela à la déduction modulo et à la surdéduction, deux formalismes proches dans lesquels le calcul est incorporé dans les démonstrations via un système de réécriture. Pour améliorer la recherche mécanisée de démonstration, nous considérons trois critères de simplicité.<br /><br />L'admissibilité des coupures permet de restreindre l'espace de recherche des démonstrations, mais elle n'est pas toujours assurée en déduction modulo. Nous définissons une procédure qui complète le système de réécriture pour, au final, admettre les coupures. Au passage, nous montrons comment transformer toute théorie pour l'intégrer à la partie calculatoire des démonstrations.<br /><br />Nous montrons ensuite comment la déduction modulo permet de réduire arbitrairement la taille des démonstrations, en transférant des étapes de déduction dans le calcul. En particulier, nous appliquons ceci à l'arithmétique d'ordre supérieur pour démontrer que les réductions de taille qui sont possibles en augmentant l'ordre dans lequel on se place disparaissent si on travaille en déduction modulo. <br /><br />Suite à ce dernier résultat, nous avons recherchés quels sont les systèmes d'ordre supérieur pouvant être simulés au premier ordre, en déduction modulo. Nous nous sommes intéressés aux systèmes de type purs et nous montrons comment ils peuvent être encodés en surdéduction, ce qui offre de nouvelles perspectives concernant leur normalisation et la recherche de démonstration dans ceux-ci. Nous développons également une méthodologie qui permet d'utiliser la surdéduction pour spécifier des systèmes de déduction.

Depuis les travaux de Martelli et Montanari en 1982, la resolution de problemes de E-unification s'effectue souvent par transformation de systemes d'equations. L'objectif de cette these est de presenter des nouvelles regles de transformations qui de- crivent de facon unifiee comment appliquer des axiomes a la ra- cine des termes. Les proprietes theoriques de ces regles sont etablies (correction, completude...). Nous prouvons egalement que cette approche, basee sur la notion de presentations strictement resolventes, est plus generale que des algorithmes tres connus (Root-Rewriting [J. Gallier & W. Snyder ], Mutation Syntaxique [C. Kirchner ]). Une analyse du comportement de ces regles per- met d'etablir l'inter^et de l'application d'axiomes a la racine et de definir le type de presentations strictement resolventes qui devraient fournir les meilleurs resultats. Ces presentations sont generees automatiquement. Pour ce faire, nous introduisons la notion de completion strictement resolvente. Elle permet de definir les proprietes theoriques des regles de completion donnees. Differentes strategies sont etu- diees, allant d'une strategie qui termine toujours a la strategie (parfois divergente) conduisant a une presentation tres efficace. Des recherches theoriques peuvent s'effectuer dans ce (nouveau) formalisme general. Elles s'appliquent aux algorithmes subsumes par cette approche. Par exemple, les avantages vis a vis du parallelisme sont etablis et conduisent a une presentation compacte de l'ensemble des unificateurs. Des optimisations theoriques plus complexes sont egalement etudiees. La detection des instanciations inutiles des variables lors de l'unification d'un terme avec les t^etes de regles est la plus importante. Elle permet d'etablir la completude des solutions donnees pour un probleme m^eme si la presentation n'est pas strictement resol- vente (completion divergente). Les resultats experimentaux mettent en valeur la simplicite et la generalite de cette nouvelle approche. Sa generalite permet egalement de comparer les differents algorithmes et de justifier l'utilisation de la strategie non complete de generation des re- gles. L'etude de cas particuliers montre que l'on peut ainsi ob- tenir des resultats tres interessants en un temps tout a fait raisonnable. On peut donc envisager d'utiliser ce module de E- unification au sein d'un demonstrateur.

<p>På Alfred Nobelmuseet i Karlskoga finns en avdelning som kallas Fiffiga huset. Där finns många experiment som besökare kan prova på, samt montrar som visar hur olika tekniska och maskinella apparater fungerar. Till utställningen och öppnandet sommaren 2009 ville de ha två nya demonstratorer. I rapporten kallas dem för reaktionstidsväggen och baklängesprataren. Reaktionstidsväggen skulle mäta besökarens reaktionstid genom att tryckknappar, med inbyggda dioder som slumpvis tänds, trycks ned i rätt ordning. Baklängesprataren skulle spela in besökarens tal och sedan spela upp det baklänges. Tanken var att besökaren skulle försöka prata baklänges och få det att låta rätt vid uppspelning. Arbetet gick ut på att bygga dessa demonstratorer och montera dem i Fiffiga huset. Reaktionstidsväggen styrdes med hjälp av en mikroprocessor ATMega16 från Atmel. Dess jobb var att läsa av knapptryckningar samt skicka information till dioder och display. Även baklängesprataren skulle styras med hjälp av mikroprocessorn ATMega16. En enkel mikrofonförstärkare samt filter till högtalarna skulle anslutas till processorn. Det inbyggda minnet skulle inte räcka till så ett extern minne behövde anslutas.Endast reaktionstidsväggen hann bli färdig inom projektets ramar eftersom all gammal elektronik som medföljde från Nobelmuseet var komplex, samt att många kortslutningar och avbrott gjorde felsökning till en tidskrävande procedur.</p>

The work presented in this dissertation investigates techniques for robot Learning from Demonstration (LFD). LFD is a well established approach where the robot is to learn from a set of demonstrations. The dissertation focuses on LFD where a human teacher demonstrates a behavior by controlling the robot via teleoperation. After demonstration, the robot should be able to reproduce the demonstrated behavior under varying conditions. In particular, the dissertation investigates techniques where previous behavioral knowledge is used as bias for generalization of demonstrations. The primary contribution of this work is the development and evaluation of a semi-reactive approach to LFD called Predictive Sequence Learning (PSL). PSL has many interesting properties applied as a learning algorithm for robots. Few assumptions are introduced and little task-specific configuration is needed. PSL can be seen as a variable-order Markov model that progressively builds up the ability to predict or simulate future sensory-motor events, given a history of past events. The knowledge base generated during learning can be used to control the robot, such that the demonstrated behavior is reproduced. The same knowledge base can also be used to recognize an on-going behavior by comparing predicted sensor states with actual observations. Behavior recognition is an important part of LFD, both as a way to communicate with the human user and as a technique that allows the robot to use previous knowledge as parts of new, more complex, controllers. In addition to the work on PSL, this dissertation provides a broad discussion on representation, recognition, and learning of robot behavior. LFD-related concepts such as demonstration, repetition, goal, and behavior are defined and analyzed, with focus on how bias is introduced by the use of behavior primitives. This analysis results in a formalism where LFD is described as transitions between information spaces. Assuming that the behavior recognition problem is partly solved, ways to deal with remaining ambiguities in the interpretation of a demonstration are proposed. The evaluation of PSL shows that the algorithm can efficiently learn and reproduce simple behaviors. The algorithm is able to generalize to previously unseen situations while maintaining the reactive properties of the system. As the complexity of the demonstrated behavior increases, knowledge of one part of the behavior sometimes interferes with knowledge of another parts. As a result, different situations with similar sensory-motor interactions are sometimes confused and the robot fails to reproduce the behavior. One way to handle these issues is to introduce a context layer that can support PSL by providing bias for predictions. Parts of the knowledge base that appear to fit the present context are highlighted, while other parts are inhibited. Which context should be active is continually re-evaluated using behavior recognition. This technique takes inspiration from several neurocomputational models that describe parts of the human brain as a hierarchical prediction system. With behavior recognition active, continually selecting the most suitable context for the present situation, the problem of knowledge interference is significantly reduced and the robot can successfully reproduce also more complex behaviors.