Robots can use a collection of auditory, visual, or haptic interfaces to convey information to human collaborators. The way these interfaces select signals typically depends on the task that the human is trying to complete: for instance, a haptic wristband may vibrate when the human is moving quickly and stop when the user is stationary. But people interpret the same signals in different ways, so what one user finds intuitive another user may not understand. In the absence of task knowledge, conveying signals is even more difficult: without knowing what the human wants to do, how should the robot select signals that helps them accomplish their task? When paired with the seemingly infinite ways that humans can interpret signals, designing an optimal interface for all users seems impossible. This thesis presents an information-theoretic approach to communication in task-agnostic settings: a unified algorithmic formalism for learning co-adaptive interfaces from scratch without task knowledge. The resulting approach is user-specific and not tied to any interface modality. This method is further improved by introducing symmetrical properties using priors on communication. Although we cannot anticipate how a human will interpret signals, we can anticipate interface properties that humans may like. By integrating these functional priors in the aforementioned learning scheme, we achieve performance far better than baselines that have access to task knowledge. The results presented here indicate that users subjectively prefer interfaces generated from the presented learning scheme while enabling better performance and more efficient interactions. / Master of Science / This thesis presents a novel interface for robot-to-human communication that personalizes to the current user without either task-knowledge nor an interpretative model of the human. Suppose that you are trying to find the location of buried treasure in a sandbox. You don't know the location of the treasure, but a robotic assistant does. Unfortunately, the only way the assistant can communicate the position of the treasure to you is through two LEDs of varying intensity --- and neither you nor the robot have a mutually understood interpretation of those signals. Without knowing the robot's convention for communication, how should you interpret the robot's signals? There are infinitely many viable interpretations: perhaps a brighter signal means that the treasure is towards the center of the sandbox -- or something else entirely. The robot has a similar problem: how should it interpret your behavior? Without knowing what you want to do with the hidden information (i.e., your task) or how you behave (i.e., your interpretative model), there is an infinite number pairs for either that fit your behavior. This work presents an interface optimizer that maximizes the correlation between the human's behavior and the hidden information. Testing with real humans indicates that this learning scheme can produce useful communicative mappings --- without knowing the users' tasks or their interpretative models. Furthermore, we recognize that humans have common biases in their interpretation of the world (leading to biases in their interpretations of robot communication). Although we cannot assume how a specific user will interpret an interface's signal, we can assume user-friendly interface designs that most humans find intuitive. We leverage these biases to further improve the aforementioned learning scheme across several user studies. As such, the findings presented in this thesis have a direct impact on human-robot co-adaptation in task-agnostic settings.

Les interfaces mobiles actuelles permettent aux utilisateurs de manipuler directement les objets affichés à l’écran avec de simples gestes, par exemple cliquer sur des boutons ou des menus ou pincer pour zoomer. Pour accéder à un espace de commande plus large, les utilisateurs sont souvent forcés de passer par de nombreuses étapes, rendant l’interaction inefficace et laborieuse. Des gestes plus complexes sont un moyen puissant d’accéder rapidement à l’information ainsi que d’exécuter des commandes plus efficacement [5]. Ils sont en revanche plus difficiles à apprendre et à contrôler. Le “Gesture Typing” (saisie de texte par geste) est une alternative intéressante au texte tapé: il permet aux utilisateurs de dessiner un geste sur leur clavier virtuel pour entrer du texte, depuis la première jusqu’à la dernière lettre d’un mot. Dans cette thèse, j’augmente le pouvoir d’expression de l’interaction mobile en tirant profit de la forme et la dynamique du geste et de l’espace d'écran, pour invoquer des commandes ainsi que pour faciliter l’appropriation dans différents contextes d’usage. Je conçois "Expressive Keyboard", qui transforme la variation du geste en un résultat riche et je démontre plusieurs applications dans le contexte de la communication textuelle. Et plus, je propose "CommandBoard", un clavier gestuel qui permet aux utilisateurs de sélectionner efficacement des commandes parmi un vaste choix tout en supportant la transition entre les novices et les experts. Je démontre plusieurs applications de "CommandBoard", dont chacune offre aux utilisateurs un choix basé sur leurs compétences cognitives et moteur, ainsi que différentes tailles et organisations de l’ensemble des commandes. Ensemble, ces techniques donnent un plus grand pouvoir expressif aux utilisateurs en tirant profit de leur contrôle moteur et de leur capacité à apprendre, à contrôler et à s’approprier. / Current mobile interfaces let users directly manipulate the objects displayed on the screen with simple stroke gestures, e.g. tap on soft buttons or menus or pinch to zoom. To access a larger command space, the users are often forced to go through long steps, making the interaction cumbersome and inefficient. More complex gestures offer a powerful way to access information quickly and to perform a command more efficiently [5]. However, they are more difficult to learn and control. Gesture typing [78] is an interesting alternative to input text: it lets users draw a gesture on soft keyboards to enter text, from the first until the final letter in a word. In this thesis, I increase the expressive power of mobile interaction by leveraging the gesture’s shape and dynamics and the screen space to produce rich output, to invoke commands, and to facilitate appropriation in different contexts of use. I design "Expressive Keyboard" that transforms the gesture variations into rich output, and demonstrate several applications in the context of textbased communication. As well, I propose "CommandBoard", a gesture keyboard that lets users efficiently select commands from a large command space while supporting the transition from novices to experts. I demonstrate different applications of "CommandBoard", each offers users a choice, based on their cognitive and motor skills, as well as the size and organization of the current command set. Altogether, these techniques give users more expressive power by leveraging human’s motor control and cognitive ability to learn, to control, and to appropriate.

L’utilisation de plusieurs applications de communication au lieu d’une est de plus en plus commune. En particulier, ces applications permettent à l’utilisateur de communiquer de diverses façons avec son partenaire, les membres de sa famille ou ses amis proches. En plus d’observer comment chaque application influence la communication, il est nécessaire de comprendre comment les gens communiquent au travers d’un écosystème d’applications. Dans la première partie, je décris comment les pratiques de communication d’un utilisateur via une application sont influencées par les contacts et fonctionnalités d'autres applications de l’écosystème. La première étude montre que les utilisateurs isolent leurs contacts dans différentes applications : ils créent des “lieux de communication”, ayant des règles uniques d’adhésion, des buts perçus et des connotations émotionnelles. Ces lieux de communication sont rompus lorsqu’un utilisateur ajoute un contact qui brise des règles d’adhésion d’une application, par exemple pour utiliser une fonctionnalité présente uniquement dans cette application. La deuxième étude montre que les utilisateurs personnalisent leurs applications de communication pour mieux exprimer leur identité et leurs liens avec d’autres personnes. Au-delà de ces personnalisations, les fonctionnalités d’une application influencent la manière dont l’utilisateur s’exprime dans d’autres applications. Pour cette raison, les fonctionnalités exclusives à une application empêchent les utilisateurs de s’exprimer de manière cohérente dans leur écosystème d’applications. Je propose quatre pistes pour explorer comment améliorer la communication via un écosystème d’applications : permettre la création de plusieurs lieux de communication dans une même application, soutenir une manière de communiquer propre à une relation à travers l’écosystème d’applications, accéder aux fonctionnalités d’une application depuis les autres applications, et de permettre des outils de communication qui appartiennent aux utilisateurs et non exclusivement aux applications. Dans la seconde partie, j’explore ces pistes/directions en réutilisant des mécanismes déjà disponibles dans les systèmes d’exploitation des téléphones mobiles : des notifications, des commandes gestuelles, et des claviers virtuels. Je réutilise les notifications comme un affichage périphérique d’alerte pour construire Lifelines, un canal de communication dédié aux couples qui partage des chronologies graphiques d’informations contextuelles, comme la proximité du lieu de vie, le niveau de batterie et le nombre de pas. Je réutilise les commandes gestuelles comme raccourcis personnels pour diverses fonctionnalités que les utilisateurs peuvent exécuter dans n’importe quelle application. Je présente une vision de conception et étudie les stratégies des utilisateurs pour créer des gestes personnels dans une étude comparative avec Fieldward et Pathward, deux techniques d’interaction pour créer des gestes qui sont faciles à mémoriser pour l’utilisateur et facile à reconnaître pour le système. Enfin, je réutilise les claviers virtuels comme des boîtes à outils de communication que les utilisateurs peuvent transporter d’une application à une autre. Je présente une vision de conception et explore sa faisabilité grâce à deux prototypes : La Shared Emoji Toolbox, qui permet de partager des collections de raccourcis pour des emojis, et CommandBoard, qui combine de la saisie gestuelle de texte avec des raccourcis gestuels pour accéder à un vaste ensemble de commandes. En conclusion, je soutiens que chaque application affecte la manière dont l'utilisateur communique dans les autres applications de l'écosystème. Nous devrions cesser de concevoir uniquement des applications isolées mais concevoir des mécanismes qui aident les utilisateurs à préserver leurs lieux de communication et à exprimer leur identité et leur intimité avec leurs proches de manière cohérente dans leur écosystème d'applications. / More and more, people communicate via not one, but a complex mix of apps. In particular, couples, close friends and families use multiple apps to express caring in diverse ways throughout the day. This calls for a new focus of study: besides observing how each app shapes communication, I argue that we need a deeper understanding of how people communicate via ecosystems of communication apps. In Part One, I show that users' communication practices in one app are not only influenced by its contacts and features but also by the contacts and features in their other apps. A first study shows that the contacts in an app affect the conversations with other contacts. To control this phenomenon, people isolate contacts in different apps: they create communication places, each with its own membership rules, perceived purpose, and emotional connotations. As relationships change, people move contacts in and out of their apps, driving communication places to redefine each other. People may break their places by bringing outsiders when the functionality they need exists only in one app.A second study shows that people customize their communication apps to better express their identities, culture and intimate bonds with others. Beyond customizations, the features of each app nurture expression habits that transfer to other apps, thus influencing how users express themselves across their entire app ecosystem. App-exclusive features prevent consistent identity expressions across apps and interfere with relationship-specific communication styles. Based on these insights, I propose four design directions for supporting ecosystems of communication apps: allow multiple communication places within the same app, support relationships across apps, provide access to functionality from other apps, and enable user-owned---rather than app-exclusive---communication tools. In Part Two, I explore those design directions by repurposing three mechanisms currently available in mobile operating systems: notifications, which users can overlay on top of any open app; gesture commands, which users could perform on any app that recognizes gestures; and soft keyboards, which appear in any app that accepts text input. I repurpose notifications as peripheral awareness displays to build Lifelines, a dedicated communication channel for couples which shares visual timelines of contextual information, e.g. closeness to home, battery level, and steps. A longitudinal study with nine couples shows how each couple leveraged Lifelines in unique ways, finding opportunities for coordinating implicitly, starting conversations, and being more understanding with each other. I repurpose gesture commands as personal gesture shortcuts to diverse functionality which users can perform in any app. I present a design envisionment and study user strategies for creating personal gestures in a comparative study of Fieldward and Pathward, two interaction techniques for creating gestures that are easy to remember and easy for the system to recognize. The results show large individual differences in users' gesture-creation strategies, reflecting their culture, their intimate bonds with special contacts and technology usage habits. Last, I repurpose soft keyboards as communication toolboxes that users can carry with them from app to app. I present a design envisionment and explore its feasibility by building two prototypes: The Shared Emoji Toolbox, which allows sharing collections of emoji shortcuts, and Command Board, which combines gesture typing with gesture shortcuts to access rich sets of commands. In conclusion, I argue that when people communicate via multiple apps, each app shapes how communication happens in others. We should shift from building isolated apps to designing mechanisms that help users preserve their communication places and express their identities and intimate bonds with others consistently across their apps.

Littorals in the in the Exuma Cays, Bahamas and the Bath Plantation, Barbados are comparative in many ways. These edges of the sea have provided critical services to local people during the time of slavery and since. More than food and medicine, the littoral is the nightly sea bath, where children are instructed, and the last ecosystem effectively used by the elderly. Independence and self- respect derive from use and protection of these littoral by individuals and communities. Local patterns of conservation and use are argued to be essential in the ecological structure and functions of the littoral. Development projects and marine protected areas alike are seen as potentially breaking local ties with the littoral causing trophic skew and damaging local society. If development occurs, mitigation solutions potentially derive from legally recognizing local people as partners in the co-management of their traditional littoral. Included with this article is a presentation prepared by Drs. Brent and Richard Stoffle.

[ES] Mucha gente en el mundo se ve afectada por la pérdida de una extremidad (las predicciones estiman que en 2050 habrá más de 3 millones de personas afectadas únicamente en los Estados Unidos de América). A pesar de la continua mejora en las técnicas de amputación y la prostética, vivir sin una extremidad sigue limitando las actividades de los afectados en su vida diaria, provocando una disminución en su calidad de vida. En este trabajo nos centramos en los casos de amputaciones de extremidades superiores, entendiendo por ello la pérdida de cualquier parte del brazo o antebrazo. Esta tesis trata sobre el control mioeléctrico (potenciales eléctricos superficiales generados por la contracción de los músculos) de prótesis de extremidades superiores. Los estudios en este campo han crecido exponencialmente en las últimas décadas intentando reducir el hueco entre la parte investigadora más dinámica y propensa a los cambios e innovación (por ejemplo, usando técnicas como la inteligencia artificial) y la industria prostética, con una gran inercia y poco propensa a introducir cambios en sus controladores y dispositivos. El principal objetivo de esta tesis es desarrollar un nuevo controlador implementable basado en filtros adaptativos que supere los principales problemas del estado del arte. Desde el punto de vista teórico, podríamos considerar dos contribuciones principales. Primero, proponemos un nuevo sistema para modelar la relación entre los patrones de la señales mioélectricas y los movimientos deseados; este nuevo modelo tiene en cuenta a la hora de estimar la posición actual el valor de los estados pasados generando una nueva sinergia entre máquina y ser humano. En segundo lugar, introducimos un nuevo paradigma de entrenamiento más eficiente y personalizado autónomamente, el cual puede aplicarse no sólo a nuestro nuevo controlador, sino a otros regresores disponibles en la literatura. Como consecuencia de este nuevo protocolo, la estructura humano-máquina difiere con respecto del actual estado del arte en dos características: el proceso de aprendizaje del controlador y la estrategia para la generación de las señales de entrada. Como consecuencia directa de todo esto, el diseño de la fase experimental resulta mucho más complejo que con los controladores tradicionales. La dependencia de la posición actual de la prótesis con respecto a estados pasados fuerza a la realización de todos los experimentos de validación del nuevo controlador en tiempo real, algo costoso en recursos tanto humanos como de tiempo. Por lo tanto, una gran parte de esta tesis está dedicada al trabajo de campo necesario para validar el nuevo modelo y estrategia de entrenamiento. Como el objetivo final es proveer un nuevo controlador implementable, la última parte de la tesis está destinada a testear los métodos propuestos en casos reales, tanto en entornos simulados para validar su robustez ante rutinas diarias, como su uso en dispositivos prostéticos comerciales. Como conclusión, este trabajo propone un nuevo paradigma de control mioélectrico para prótesis que puede ser implementado en una prótesis real. Una vez se ha demostrado la viabilidad del sistema, la tesis propone futuras líneas de investigación, mostrando algunos resultados iniciales. / [CA] Molta gent en el món es veu afectada per la pèrdua d'una extremitat (les prediccions estimen que en 2050 hi haurà més de 3 milions de persones afectades únicament als Estats Units d'Amèrica). Malgrat la contínua millora en les tècniques d'amputació i la prostètica, viure sense una extremitat continua limitant les activitats dels afectats en la seua vida diària, provocant una disminució en la seua qualitat de vida. En aquest treball ens centrem en els casos d'amputacions d'extremitats superiors, entenent per això la pèrdua de qualsevol part del braç o avantbraç. Aquesta tesi tracta sobre el control mioelèctric (potencials elèctrics superficials generats per la contracció dels músculs) de pròtesis d'extremitats superiors. Els estudis en aquest camp han crescut exponencialment en les últimes dècades intentant reduir el buit entre la part investigadora més dinàmica i propensa als canvis i innovació (per exemple, usant tècniques com la intel·ligència artificial) i la indústria prostètica, amb una gran inèrcia i poc propensa a introduir canvis en els seus controladors i dispositius. Aquesta tesi contribueix a la investigació des de diversos punts de vista. El principal objectiu és desenvolupar un nou controlador basat en filtres adaptatius que supere els principals problemes de l'estat de l'art. Des del punt de vista teòric, podríem considerar dues contribucions principals. Primer, proposem un nou sistema per a modelar la relació entre els patrons de la senyals mioelèctrics i els moviments desitjats; aquest nou model té en compte a l'hora d'estimar la posició actual el valor dels estats passats generant una nova sinergia entre màquina i ésser humà. En segon lloc, introduïm un nou paradigma d'entrenament més eficient i personalitzat autònomament, el qual pot aplicar-se no sols al nostre nou controlador, sinó a uns altres regresors disponibles en la literatura. Com a conseqüència d'aquest nou protocol, l'estructura humà-màquina difereix respecte a l'actual estat de l'art en dues característiques: el procés d'aprenentatge del controlador i l'estratègia per a la generació dels senyals d'entrada. Com a conseqüència directa de tot això, el disseny de la fase experimental resulta molt més complex que amb els controladors tradicionals. La dependència de la posició actual de la pròtesi respecte a estats passats força a la realització de tots els experiments de validació del nou controlador en temps real, una cosa costosa en recursos tant humans com de temps. Per tant, una gran part d'aquesta tesi està dedicada al treball de camp necessari per a validar el nou model i estratègia d'entrenament. Com l'objectiu final és proveir un nou controlador implementable, l'última part de la tesi està destinada a testar els mètodes proposats en casos reals, tant en entorns simulats per a validar la seua robustesa davant rutines diàries, com el seu ús en dispositius prostètics comercials. Com a conclusió, aquest treball proposa un nou paradigma de control mioelèctric per a pròtesi que pot ser implementat en una pròtesi real. Una vegada s'ha demostrat la viabilitat del sistema, la tesi proposa futures línies d'investigació, mostrant alguns resultats inicials. / [EN] Many people in the world suffer from the loss of a limb (predictions estimate more than 3 million people by 2050 only in the USA). In spite of the continuous improvement in the amputation rehabilitation and prosthetic restoration, living without a limb keeps limiting the daily life activities leading to a lower quality of life. In this work, we focus in the upper limb amputation case, i.e., the removal of any part of the arm or forearm. This thesis is about upper limb prosthesis control using electromyographic signals (the superficial electric potentials generated during muscle contractions). Studies in this field have grown exponentially in the past decades trying to reduce the gap between a fast growing prosthetic research field, with the introduction of machine learning, and a slower prosthetic industry and limited manufacturing innovation. This thesis contributes to the field from different perspectives. The main goal is to provide and implementable new controller based on adaptive filtering that overcomes the most common state of the art concerns. From the theoretical point of view, there are two main contributions. First, we propose a new system to model the relationship between electromyographic signals and the desired prosthesis movements; this new model takes into account previous states for the estimation of the current position generating a new human-machine synergy. Second, we introduce a new and more efficient autonomously personalized training paradigm, which can benefit not only to our new proposed controller but also other state of the art regressors. As a consequence of this new protocol, the human-machine structure differs with respect to current state of the art in two features: the controller learning process and the input signal generation strategy. As a direct aftereffect of all of this, the experimental phase design results more complex than with traditional controllers. The current state dependency on past states forces the experimentation to be in real time, a very high demanding task in human and time resources. Therefore, a major part of this thesis is the associated fieldwork needed to validate the new model and training strategy. Since the final goal is to provide an implementable new controller, the last part of the thesis is devoted to test the proposed methods in real cases, not only analyzing the robustness and reliability of the controller in real life situations but in real prosthetic devices. As a conclusion, this work provides a new paradigm for the myoelectric prosthetic control that can be implemented in a real device. Once the thesis has proven the system's viability, future work should continue with the development of a physical device where all these ideas are deployed and used by final patients in a daily basis. / The work of Carles Igual Bañó to carry out this research and elaborate this dissertation has been supported by the Ministerio de Educación, Cultura y Deporte under the FPU Grant FPU15/02870. One visiting research fellowships (EST18/00544) was also funded by the Ministerio de Educación, Cultura y Deporte of Spain. / Igual Bañó, C. (2021). Co-adaptive myoelectric control for upper limb prostheses [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/168192 / TESIS