La robótica de enjambre es un campo de investigación dentro del área de la robótica que estudia la coordinación de un gran número de robots simples. Este campo de investigación se inspira en el comportamiento observado en los insectos sociales, los cuales son grandes ejemplos de cómo un gran número de individuos simples pueden interactuar para crear sistemas inteligentes colectivos. En estos sistemas el comportamiento colectivo emerge de forma auto-organizada a partir de las interacciones entre los individuos y entre los individuos y el entorno. De la misma manera, en los sistemas de enjambre artificiales, la inteligencia es una propiedad emergente a partir del comportamiento global del enjambre. En estos sistemas, el enjambre es capaz de llevar a cabo tareas, de manera global, que están fuera de las capacidades de un robot individual. Los sistemas robóticos de enjambre deben cumplir una serie de características que los diferencian de otros sistemas multi-robóticos. Algunas de estas características son compartidas con los sistemas multi-agente, por lo tanto, éstos pueden ser una alternativa para la construcción de este tipo de sistemas. En este trabajo se presentan un conjunto de comportamientos colectivos para enjambres artificiales. Para dotar de un marco de implantación a estos comportamientos, se define un modelo de arquitectura híbrida para el control de un enjambre de robots basada en un sistema multi-agente. Una de las características básicas de este modelo es su división en capas, que permite utilizar la capa inferior, de enjambre puro, de manera aislada, para tareas donde el comportamiento colectivo del enjambre emerge únicamente a partir de las interacciones entre los agentes y el entorno. En primer lugar, se analizan tres comportamientos básicos de robótica de enjambre: agregación, movimiento coordinado (flocking) y dispersión. Posteriormente, se definen dos comportamientos concretos para un enjambre de robots. El primero de éstos muestra un comportamiento para la localización de recursos en el entorno, donde los robots son capaces de localizar la fuente de recursos más prometedora en entornos desconocidos, con ruido y con diversas fuentes de recursos. El segundo caso define un comportamiento capaz de detectar, monitorizar, cubrir y marcar el perímetro de un vertido petrolífero marítimo.

The World Health Organization (WHO) stated in February 2021 at the Seventy- Third World Health Assembly that, globally, at least 2.2 billion people have a near or distance vision impairment. They also denoted the severe impact vision impairment has on the quality of life of the individual suffering from this condition, how it affects the social well-being and their economic independence in society, becoming in some cases an additional burden to also people in their immediate surroundings. In order to minimize the costs and intrusiveness of the applications and maximize the autonomy of the individual life, the natural solution is using systems that rely on computer vision algorithms. The systems improving the quality of life of the visually impaired need to solve different problems such as: localization, path recognition, obstacle detection, environment description, navigation, etc. Each of these topics involves an additional set of problems that have to be solved to address it. For example, for the task of object detection, there is the need of depth prediction to know the distance to the object, path recognition to know if the user is on the road or on a pedestrian path, alarm system to provide notifications of danger for the user, trajectory prediction of the approaching obstacle, and those are only the main key points. Taking a closer look at all of these topics, they have one key component in common: depth estimation/prediction. All of these topics are in need of a correct estimation of the depth in the scenario. In this thesis, our main focus relies on addressing depth estimation in indoor and outdoor environments. Traditional depth estimation methods, like structure from motion and stereo matching, are built on feature correspondences from multiple viewpoints. Despite the effectiveness of these approaches, they need a specific type of data for their proper performance. Since our main goal is to provide systems with minimal costs and intrusiveness that are also easy to handle we decided to infer the depth from single images: monocular depth estimation. Estimating depth of a scene from a single image is a simple task for humans, but it is notoriously more difficult for computational models to be able to achieve high accuracy and low resource requirements. Monocular Depth Estimation is this very task of estimating depth from a single RGB image. Since there is only a need of one image, this approach is used in applications such as autonomous driving, scene understanding or 3D modeling where other type of information is not available. This thesis presents contributions towards solving this task using deep learning as the main tool. The four main contributions of this thesis are: first, we carry out an extensive review of the state-of-the-art in monocular depth estimation; secondly, we introduce a novel large scale high resolution outdoor stereo dataset able to provide enough image information to solve various common computer vision problems; thirdly, we show a set of architectures able to predict monocular depth effectively; and, at last, we propose two real life applications of those architectures, addressing the topic of enhancing the perception for the visually impaired using low-cost wearable sensors.

In this thesis, a study of two blooming fields in the artificial intelligence topic is carried out. The first part of the present document is about 3D object recognition methods. Object recognition in general is about providing the ability to understand what objects appears in the input data of an intelligent system. Any robot, from industrial robots to social robots, could benefit of such capability to improve its performance and carry out high level tasks. In fact, this topic has been largely studied and some object recognition methods present in the state of the art outperform humans in terms of accuracy. Nonetheless, these methods are image-based, namely, they focus in recognizing visual features. This could be a problem in some contexts as there exist objects that look alike some other, different objects. For instance, a social robot that recognizes a face in a picture, or an intelligent car that recognizes a pedestrian in a billboard. A potential solution for this issue would be involving tridimensional data so that the systems would not focus on visual features but topological features. Thus, in this thesis, a study of 3D object recognition methods is carried out. The approaches proposed in this document, which take advantage of deep learning methods, take as an input point clouds and are able to provide the correct category. We evaluated the proposals with a range of public challenges, datasets and real life data with high success. The second part of the thesis is about hand pose estimation. This is also an interesting topic that focuses in providing the hand's kinematics. A range of systems, from human computer interaction and virtual reality to social robots could benefit of such capability. For instance to interface a computer and control it with seamless hand gestures or to interact with a social robot that is able to understand human non-verbal communication methods. Thus, in the present document, hand pose estimation approaches are proposed. It is worth noting that the proposals take as an input color images and are able to provide 2D and 3D hand pose in the image plane and euclidean coordinate frames. Specifically, the hand poses are encoded in a collection of points that represents the joints in a hand, so that they can be easily reconstructed in the full hand pose. The methods are evaluated on custom and public datasets, and integrated with a robotic hand teleoperation application with great success.

In this thesis, a study of two blooming fields in the artificial intelligence topic is carried out. The first part of the present document is about 3D object recognition methods. Object recognition in general is about providing the ability to understand what objects appears in the input data of an intelligent system. Any robot, from industrial robots to social robots, could benefit of such capability to improve its performance and carry out high level tasks. In fact, this topic has been largely studied and some object recognition methods present in the state of the art outperform humans in terms of accuracy. Nonetheless, these methods are image-based, namely, they focus in recognizing visual features. This could be a problem in some contexts as there exist objects that look alike some other, different objects. For instance, a social robot that recognizes a face in a picture, or an intelligent car that recognizes a pedestrian in a billboard. A potential solution for this issue would be involving tridimensional data so that the systems would not focus on visual features but topological features. Thus, in this thesis, a study of 3D object recognition methods is carried out. The approaches proposed in this document, which take advantage of deep learning methods, take as an input point clouds and are able to provide the correct category. We evaluated the proposals with a range of public challenges, datasets and real life data with high success. The second part of the thesis is about hand pose estimation. This is also an interesting topic that focuses in providing the hand's kinematics. A range of systems, from human computer interaction and virtual reality to social robots could benefit of such capability. For instance to interface a computer and control it with seamless hand gestures or to interact with a social robot that is able to understand human non-verbal communication methods. Thus, in the present document, hand pose estimation approaches are proposed. It is worth noting that the proposals take as an input color images and are able to provide 2D and 3D hand pose in the image plane and euclidean coordinate frames. Specifically, the hand poses are encoded in a collection of points that represents the joints in a hand, so that they can be easily reconstructed in the full hand pose. The methods are evaluated on custom and public datasets, and integrated with a robotic hand teleoperation application with great success.

Desde hace años venimos contemplando cómo nuestra sociedad ha cambiado de la mano de la evolución de las Tecnologías de la Información (TI). Nos encontramos en un entorno que cambia constantemente, en el que la información se renueva continuamente, lo que nos lleva a un aprendizaje dinámico, continuo, y cuyas barreras están desapareciendo, hacia un aprendizaje global. Con esto, la educación está inmersa en un proceso de cambio, de una transformación que permita hacer frente a estas nuevas características y necesidades que presenta la sociedad en este nuevo entorno, una transformación digital. Se trata de una forma diferente de aprendizaje en la que, además, los espacios educativos se están deslocalizando. Y en este proceso de transformación, el potencial y el rápido crecimiento de las tecnologías de la información pueden tener un papel crucial y conformar la base para una verdadera evolución en este entorno digital. Sin embargo, la situación actual es que estas expectativas no se han cumplido, el uso de las TI en educación no está logrando el efecto que se esperaba, no están contribuyendo a una verdadera transformación del proceso de aprendizaje. Y es que, entre otras razones, el uso que se está haciendo de las TI es de meras herramientas complementarias, un uso superficial, cuando deberíamos emplearlas para poder profundizar en el proceso de aprendizaje. Para conseguir un cambio significativo, esa transformación que se pretende, debemos ir más allá. Por esta razón, proponemos un modelo de aprendizaje adaptativo y personalizado, que sirva de base para crear un sistema de aprendizaje que permita cubrir las necesidades detectadas en la sociedad digital sin descuidar los objetivos intencionales educativos. Un modelo que hemos llamado CALM, acrónimo de Customized Adaptive Learning Model. Se trata de un modelo que se adaptará a las características y al estado de cada aprendiz y que busca acrecentar su motivación, ofreciéndole autonomía en su propio proceso de aprendizaje, en un ciclo continuo de mejora. Todo ello diseñado y supervisado en todo momento por el docente, cuyo papel consideramos crucial en este proceso. En este modelo, el contenido está dividido en competencias, que serán los conocimientos, las habilidades y las aptitudes que los aprendices irán adquiriendo, dispuestas en forma de grafo dirigido o, como lo llamamos, mapa de competencias. Estas competencias serán desarrolladas a través de actividades que irán realizando, y será el propio sistema, a través de lo que llamamos el motor de selección, el que asigne a cada aprendiz en cada momento la actividad que considere más apropiada. Por su parte, el docente será el que diseñe todo el conjunto de aprendizaje, creando las competencias y configurando el mapa, y añadiendo las actividades. Después, podrá en todo momento supervisar el proceso de todos los aprendices, analizando su progreso y estado, tanto colectivo como individual, y gestionarlo a través de un factor clave que introducimos en el modelo: las estrategias instruccionales. A través de ellas, el docente podrá guiar al modelo en la selección de actividades, de modo que, a pesar de que este analiza de forma dinámica las características de cada aprendiz para asignarles una actividad, la estrategia docente marcará la decisión a tomar, según los criterios que el docente considere apropiados, individual o globalmente. Para comprobar que las características de nuestro modelo, hemos puesto en práctica el modelo a través de una prueba con una plataforma piloto, usada por estudiantes y docentes reales, obteniendo unas valoraciones muy positivas por ambas partes. Con CALM, hemos propuesto una base para construir un sistema de aprendizaje inteligente con el que cubrir las necesidades educativas que presenta nuestra sociedad actual, a través de un aprendizaje adaptativo y personalizado, teniendo siempre presentes los objetivos docentes.

Metaphors are an important literary figure that is found in books or and daily use. Nowadays it is an essential task for Natural Language Processing (NLP), but the dependence of the context and the lack corpus in other languages make it a bottleneck for some tasks such as translation or interpretation of texts. We present a classification model using recurrent neural networks for metaphor identification in Spanish sentences. We tested our model and his variants on a new corpus in Spanish and compared it with the current baseline using an English corpus. Our best model reports an F-score of 52.5% for Spanish and 60.4% for English. / Trabajo académico

Proyecto DGSIC PB98-0977

No description available.

Along with the improvement of data acquisition techniques and the increasing computational capacity of computers, the dimensionality of the data grows higher. Pattern recognition methods have to deal with samples consisting of thousands of features and the reduction of their dimensionality becomes crucial to make them tractable. Feature selection is a technique for removing the irrelevant and noisy features and selecting a subset of features which describe better the samples and produce a better classification performance. It is becoming an essential part of most pattern recognition applications. / In this thesis we propose a feature selection method for supervised classification. The main contribution is the efficient use of information theory, which provides a solid theoretical framework for measuring the relation between the classes and the features. Mutual information is considered to be the best measure for such purpose. Traditionally it has been measured for ranking single features without taking into account the entire set of selected features. This is due to the computational complexity involved in estimating the mutual information. However, in most data sets the features are not independent and their combination provides much more information about the class, than the sum of their individual prediction power. / Methods based on density estimation can only be used for data sets with a very high number of samples and low number of features. Due to the curse of dimensionality, in a multi-dimensional feature space the amount of samples required for a reliable density estimation is very high. For this reason we analyse the use of different estimation methods which bypass the density estimation and estimate entropy directly from the set of samples. These methods allow us to efficiently evaluate sets of thousands of features. / For high-dimensional feature sets another problem is the search order of the feature space. All non-prohibitive computational cost algorithms search for a sub-optimal feature set. Greedy algorithms are the fastest and are the ones which incur less overfitting. We show that from the information theoretical perspective, a greedy backward selection algorithm conserves the amount of mutual information, even though the feature set is not the minimal one. / We also validate our method in several real-world applications. We apply feature selection to omnidirectional image classification through a novel approach. It is appearance-based and we select features from a bank of filters applied to different parts of the image. The context of the task is place recognition for mobile robotics. Another set of experiments are performed on microarrays from gene expression databases. The classification problem aims to predict the disease of a new patient. We present a comparison of the classification performance and the algorithms we present showed to outperform the existing ones. Finally, we succesfully apply feature selection to spectral graph classification. All the features we use are for unattributed graphs, which constitutes a contribution to the field. We also draw interesting conclusions about which spectral features matter most, under different experimental conditions. In the context of graph classification we also show important is the precise estimation of mutual information and we analyse its impact on the final classification results.

El documento de esta tesis por compendio de publicaciones se divide en dos partes: la síntesis donde se resume la fundamentación, resultados y conclusiones de esta tesis, y las propias publicaciones en su formato original, que se incluyen como apéndices. Dado que existen acuerdo de confidencialidad (véase "Derechos" más adelante) que impiden su publicación en formato electrónico de forma pública y abierta (como es el repositorio de la UA), y acorde con lo que se dictamina en el punto 6 del artículo 14 del RD 99/2011, de 28 de enero, no se incluyen estos apéndices en el documento electrónico que se presenta en cedé, pero se incluyen las referencias completas y sí se incluyen integramente en el ejemplar encuadernado. Si el CEDIP y el RUA así lo decidiesen más adelante, podría modificarse este documento electrónico para incluir los enlaces a los artículos originales. / Ministerio de Educación y Ciencia y los programas FEDER y FSE de la UE (proyecto ref. BEC2003-02391 y Programa de Personal Técnico de Apoyo en la modalidad de Proyectos de I+D, ref. solicitud PTA-2003-02-00178, 495); Ministerio de Fomento (proyecto ref. T 75/2006); Ministerio de Ciencia e Innovación y el programa FEDER de la UE (proyectos ref. SEJ2007-67767-C04-02 y ref. CSO2011-29943-C03-02); Universidad de Alicante.