Orientador: Eleri Cardozo / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação / Made available in DSpace on 2018-08-26T06:38:13Z (GMT). No. of bitstreams: 1 Olivi_LeonardoRocha_D.pdf: 17712423 bytes, checksum: a9b0ec521405e93735844f7ce8b3b968 (MD5) Previous issue date: 2014 / Resumo: A mobilidade é uma competência humana extremamente valiosa, e pode ser perdida por diversas razões, tais como traumas na coluna vertebral, acidentes vasculares cerebrais, dentre outras. Algumas tecnologias desenvolvidas para as áreas médicas, como eletroencefalografia e eletromiografia, podem ser empregadas no desenvolvimento de Interfaces Humano-Máquina (Human-Machine Interface, ou, HMI) com o objetivo de permitir que pessoas com capacidades motoras severamente comprometidas possam comunicar e operar os mais diversos equipamentos. Assim, esses usuários podem conduzir robôs móveis por meio de uma HMI adaptada para suas capacidades motoras e cognitivas. Entretanto, essas interfaces apresentam erros na identificação da ação pretendida pelo usuário, os quais comprometem sobremaneira a experiência desse usuário na interação com o mundo exterior. Para o caso dos robôs móveis assistivos, o conceito de controle compartilhado (shared control) foi concebido para compensar as taxas de erros das HMIs, e a responsabilidade pela navegação do robô é dividida entre o usuário e um sistema supervisor automático. As abordagens mais populares na literatura comutam poucos modos de comportamentos específicos para situações individualizadas, como passagens estreitas e desvio de obstáculos. Esta tese propõe uma nova técnica de controle compartilhado, denominada de navegação assistida por campos vetoriais, a qual não possui modos de operação chaveados, evitando quaisquer instabilidades e abrangendo as configurações do ambiente de maneira ilimitada, além de minimizar significativamente os erros gerados pelas HMIs e facilitar a navegação do usuário. Os campos vetoriais oferecem as navegações manual e autônoma, ampliando a interação do usuário com o ambiente. Nessa nova técnica, o sistema de controle embarcado irá executar a ação identificada pela HMI em função dos estados do robô e do ambiente, com o objetivo de maximizar a segurança e capacidade de controle do usuário. Os resultados mostrados neste trabalho evidenciam uma nova forma de tratar o problema, obtendo ganhos significativos com relação ao estado da arte, com baixa complexidade computacional, alta flexibilidade a ambientes e usuários e otimização efetiva de erros, contribuindo para a recuperação da mobilidade dos usuários / Abstract: Human mobility is an extremely valued skill that can be lost due to various reasons, such as spinal cord injuries, strokes, amputations, among others. Technologies developed for the medical areas, such as electroencephalography and electromyography, can be employed in the development of Human-Machine Interfaces (HMI) with the objective of allowing people with severe mobility impairments to communicate and to operate a wide range of equipments. Therefore, these users are allowed to conduct assistive robots allow through a HMI designed according to the user's mobility and cognitive skills. However, these interfaces still present low performance in the correct identification of the intended action by the user, which severely compromise the experience of the user when interacting with external world. In the specific case of assistive mobile robots, a shared control concept was developed in order to compensate the high error rates produced by the HMIs, where the responsibility for the navigation of the mobile robot is shared among the user and an autonomous supervisor system. The mainly approaches shown in literature switch a few modes of specific behavior for individualized situations, such as narrow corridors and obstacle avoidance. This work presents a novel shared control technique, named assistive navigation by vector fields, which does not employ switching modes, avoiding any instabilities and covering the unlimited environment configurations, significantly minimizing the HMI errors, facilitating the user's navigation. The vector fields offer both manual and autonomous navigation, increasing the user's interaction with the environment. In this novel technique, the embedded control system incorporates the HMI command with the robot and environment states, aiming the maximization of user's security and control capabilities. Results shown a novel manner for treating the problem, obtaining substantial gains over the state of art works, with low computational complexity, high flexibility concerning environments and users and effective optimization of errors, contributing for the user's mobility retrieval / Doutorado / Automação / Doutor em Engenharia Elétrica

La thématique de la navigation autonome constitue l’un des principaux axes de recherche dans le domaine des véhicules intelligents et des robots mobiles. Dans ce contexte, on cherche à doter le robot d’algorithmes et de méthodes lui permettant d’évoluer dans un environnement complexe et dynamique, en toute sécurité et en parfaite autonomie. Dans ce contexte, les algorithmes de localisation et de cartographie occupent une place importante. En effet, sans informations suffisantes sur la position du robot (localisation) et sur la nature de son environnement (cartographie), les autres algorithmes (génération de trajectoire, évitement d’obstacles ...) ne peuvent pas fonctionner correctement. Nous avons centré notre travail de thèse sur une problématique précise : développer un algorithme de SLAM simple, rapide, léger et limitant les erreurs de localisation et de cartographie au maximum sans fermeture de boucle. Au cœur de notre approche, on trouve un algorithme d’IML : Incremental Maximum Likelihood. Ce type d’algorithmes se base sur une estimation itérative de la localisation et de la cartographie. Il est ainsi naturellement divergent. Le choix de l’IML est justifié essentiellement par sa simplicité et sa légèreté. La particularité des travaux réalisés durant cette thèse réside dans les différents outils et algorithmes utilisés afin de limiter la divergence de l’IML au maximum, tout en conservant ses avantages. / One of the main areas of research in the field of intelligent vehicles and mobile robots is Autonomous navigation. In this field, we seek to create algorithms and methods that give robots the ability to move safely and autonomously in a complex and dynamic environment. In this field, localization and mapping algorithms have an important place. Indeed,without reliable information about the robot position (localization) and the nature of its environment (mapping), the other algorithms (trajectory generation, obstacle avoidance ...) cannot achieve their tasks properly. We focused our work during this thesis on a specific problem: to develop a simple, fast and lightweight SLAM algorithm that can minimize localization errors without loop closing. At the center of our approach, there is an IML algorithm: Incremental Maximum Likelihood. This kind of algorithms is based on an iterative estimation of the localization and the mapping. It contains thus naturally a growing error in the localization process. The choice of IML isjustified mainly by its simplicity and lightness. The main idea of our work is built around thedifferent tools and algorithms used to minimize the localization error of IML, while keeping its advantages.

This dissertation aims to present a structured overview of the state-of-the-art in cooperative robotics research. As we illustrate in this dissertation, there are several interesting aspects that draws attention to the field, among which 'Life Sciences' and 'Applied AI' are emphasized. We analyse the key concepts and main research issues within the field, and discuss its relations to other disciplines, including cognitive science, biology, artificial life and engineering. In particular it can be noted that the study of collective robot behaviour has drawn much inspiration from studies of animal behaviour. In this dissertation we also analyse one of the most attractive research areas within cooperative robotics today, namely RoboCup. Finally, we present a hierarchy of levels and mechanisms of cooperation in robots and animals, which we illustrate with examples and discussions.

Wireless sensor networks (WSN) increasingly permeate modern societies nowadays. But in spite of their plethora of successful applications, WSN are often unable to surmount many operational challenges that unexpectedly arise during their lifetime. Fortunately, robotic agents can now assist a WSN in various ways. This thesis illustrates how mobile robots which are able to carry a limited number of sensors can help the network react to sensor faults, either during or after its deployment in the monitoring region. Two scenarios are envisioned. In the first one, carrier robots surround a point of interest with multiple sensor layers (focused coverage formation). We put forward the first known algorithm of its kind in literature. It is energy-efficient, fault-reactive and aware of the bounded robot cargo capacity. The second one is that of replacing damaged sensing units with spare, functional ones (coverage repair), which gives rise to the formulation of two novel combinatorial optimization problems. Three nature-inspired metaheuristic approaches that run at a centralized location are proposed. They are able to find good-quality solutions in a short time. Two frameworks for the identification of the damaged nodes are considered. The first one leans upon diagnosable systems, i.e. existing distributed detection models in which individual units perform tests upon each other. Two swarm intelligence algorithms are designed to quickly and reliably spot faulty sensors in this context. The second one is an evolving risk management framework for WSNs that is entirely formulated in this thesis.

Starting a few decades ago, the unmanned wheeled vehicle research has drawn lately more attention, especially for off-road environment. As the demand to use electric vehicles increased, the need to conceptualize the use of electrically driven vehicles in autonomous operations became a target. That is because in addition to the fact that they are more environmentally friendly, they are also easier to control. This also gives another reason to enhance further the energy economy of those unmanned electric vehicles. Off-road vehicles research was always challenging, but in the present work the nature of the off-road land is utilized to benefit from in order to enhance the energy consumption of those vehicles. An algorithm for energy consumption optimization for electrically driven unmanned wheeled vehicles is presented. The algorithm idea is based on the fact that in off-road conditions, when the vehicle passes a ditch or a hole, the kinetic energy gained while moving downhill could be utilized to reduce the energy consumption for moving uphill if the dimensions of the ditch/hole were known a distance ahead. Two manipulated variables are evaluated: the wheels DC motors supply voltage and the DC armature current. The developed algorithm is analysed and compared to the PID speed iii controller and to the open-loop control of DC motors. The developed predictive controller achieved encouraging results compared to the PID speed control and also compared to the open-loop control. Also, the use of the DC armature current as a manipulated variable showed more noticeable improvement over using the DC input voltage. Experimental work was carried out to validate the predictive control algorithm. A mobile robot with two DC motor driven wheels was deployed to overcome a ditch-like hindrance. The experimental results verified the simulation results. A parametric study for the predictive control is conducted. The effect of changing the downhill angle and the uphill angle as well as the size of the prediction horizon on the consumed electric energy by the DC motors is addressed. The simulation results showed that, when using the proposed approach, the larger the prediction horizon, the lower the energy consumption is.

Orientador: Ricardo Ribeiro Gudwin / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação / Made available in DSpace on 2018-08-27T01:31:15Z (GMT). No. of bitstreams: 1 Raizer_Klaus_D.pdf: 4879759 bytes, checksum: 77716297b6419a3ee55bdf97ac67493d (MD5) Previous issue date: 2015 / Resumo: O objetivo deste trabalho é o desenvolvimento de funções executivas para uma arquitetura cognitiva bioinspirada baseada em codelets. Um desafio que toda criatura (seja ela artificial ou biológica) enfrenta é definir qual a próxima ação a ser tomada, a cada instante de tempo, em função da percepção de um determinado ambiente. Essa decisão pode ser definida por um algoritmo que sempre repete as mesmas decisões em função de uma determinada situação, ou pode ser uma decisão adaptativa, que utiliza de mecanismos de aprendizagem para assumir decisões distintas, em função das experiências em situações passadas. Neste trabalho, buscou-se a integração dos processos de tomada de decisão deliberativos e mecanismos de aprendizado por reforço em um mesmo framework. Estas funções são conhecidas na literatura de ciências cognitivas como funções executivas. A solução aqui proposta insere-se dentro do contexto de nosso grupo de pesquisa, onde se busca o desenvolvimento de uma arquitetura cognitiva baseada em codelets. Nesta perspectiva, uma das contribuições deste trabalho é desenvolver algoritmos e implementações computacionais dotando a arquitetura cognitiva desenvolvida pelo grupo de funções executivas diversas, que poderão ser utilizadas para implementar soluções complexas com granularidade arbitrária. As funções de tomada de decisão deliberativa foram implementada na forma de uma rede de comportamentos modificada, enquanto que o componente de aprendizado foi desenvolvido na forma de um novo algoritmo (GLAS - Gated-Learning Action Selection) baseado em stimulus gating e inspirado em modelos de neurociência computacional conhecidos da literatura. Este framework foi validado em problemas de robótica móvel e de seleção de ação por aprendizado por reforço. A arquitetura cognitiva sendo desenvolvida, incrementada com as contribuições deste trabalho, tem o potencial de servir de base para futuros trabalhos de pesquisa nas áreas de inteligência artificial, robótica e cognição artificial / Abstract: This work¿s goal is the development of executive functions for a codelet-based bio-inspired cognitive architecture. One of the major challenges every creature faces, being biological or artificial, is to define the next action to be taken, at each time step, as a function of how it perceives its surrounding environment. This decision can be made by a reactive algorithm, which always repeats the same decisions for a given situation, or by an adaptive process, which is able to make use of learning mechanisms in order to make distinct decisions based on past experience. In this work, deliberative decision-making and reinforcement learning mechanisms have been integrated into a single framework. In cognitive science literature, these functions are known as executive functions. The solution proposed here is part of our group¿s central line of research, which is the investigation and development of a codelet-based cognitive architecture. In this context, a central contribution made by this work is the development and implementation of algorithms capable of providing this cognitive architecture with a group of executive functions, which in turn can be used to implement complex solutions with arbitrary granularity. Functions for deliberative decision-making have been implemented in the form of a modified behavior network, while the learning component was developed in the form of a new algorithm called GLAS (Gated-Learning Action Selection), based on stimulus gating and known computational neuroscience models. This framework has been validated with problems in mobile robotics and in action selection by reinforcement learning. The cognitive architecture under development, when incremented by the contributions presented in this work, has the potential to serve as a base for future work and research in the fields of artificial intelligence, robotics and artificial cognition / Doutorado / Engenharia de Computação / Doutor em Engenharia Elétrica

Monocular vision robots use a single camera to process information about its environment. By analyzing this scene, the robot can determine the best navigation direction. Many modern approaches to robot hallway navigation involve using a plethora of sensors to detect certain features in the environment. This can be laser range finders, inertial measurement units, motor encoders, and cameras. By combining all these sensors, there is unused data which could be useful for navigation. To draw back and develop a baseline approach, this thesis explores the reliability and capability of solely using a camera for navigation. The basic navigation structure begins by taking frames from the camera and breaking them down to find the most prominent lines. The location where these lines intersect determine the forward direction to drive the robot. To improve the accuracy of navigation, algorithm improvements and additional features from the camera frames are used. This includes line intersection weighting to reduce noise from extraneous lines, floor segmentation to improve rotational stability, and person detection.

Dizertační práce se zabývá problematikou kinematického modelování a řízení mobliních kolových robotů. Přináší sumarizaci problematiky kinematického modelování mobilních robotů obecně a popis vlastností kolových mobilních robotů s několika řízenými koly. Použitý aparát z matematiky, fyziky je vysvětlován s důrazem na pohled teorie řízení. Dále je prezentován nový řídicí algoritmus pro mobilní kolové roboty s více řízenými koly, vhodný pro úlohu stabilizace v bodě i sledování trajektorie, tedy obě nejčastěji řešené úlohy pohybu mobilních robotů.

Dans cette thèse, des méthodes dites de Lyapunov sont proposées afin de résoudre des problèmes liés à la coordination distribuée des systèmes multiagent, plus précisément, un groupe de systèmes (agents) non-linéaires formés de robots mobiles non-holonomes est considéré. Pour ce groupe de systèmes, des lois de commande distribuée sont proposées dans le but de résoudre des problèmes de type leader-suiveur en formation et aussi des problèmes de type formation sans-leader par une approche de consensus, sous différentes hypothèses sur le graphe de communication et surtout sur les vitesses du leader.L'originalité de ce travail est dans l'approche proposée pour l'étude de stabilité de la boucle fermée, cette approche consiste à transformer les deux derniers problèmes en des problèmes de stabilisation globale asymptotique d'un ensemble invariant. L’analyse de stabilité est basée sur la construction de fonction de Lyapunov et de fonction de Lyapunov-Karasovskii strictes pour des classes de systèmes non-linéaires variant dans le temps présentant des retards bornés et variant dans le temps. / In this thesis, we propose a Lyapunov based approaches to address some distributedsolutions to multi-agent coordination problems, more precisely, we consider a groupof agents modeled as nonholonomic mobile robots, we provide a distributed controllaws in order to solve the leader-follower and the leaderless consensus problems under different assumptions on the communication graph topology and on the leader’strajectories. The originality of this work relies on the closed-loop analysis approach, that is, it consists on transforming the last two problems into a global stabilization problem of an invariant set. The stability analysis is mainly based on the construction of strict Lyapunov functions and strict Lyapunov-Krasovskii functionals for a classes of nonlinear time-varying and/or delayed systems.

Providing mobile robots with the ability to manipulate objects has, despite decades of research, remained a challenging problem. The problem is approachable in constrained environments where there is ample prior knowledge of the environment layout and manipulatable objects. The challenge is in building systems that scale beyond specific situational instances and gracefully operate in novel conditions. In the past, researchers used heuristic and simple rule-based strategies to accomplish tasks such as scene segmentation or reasoning about occlusion. These heuristic strategies work in constrained environments where a roboticist can make simplifying assumptions about everything from the geometries of the objects to be interacted with, level of clutter, camera position, lighting, and a myriad of other relevant variables. The work in this thesis will demonstrate how to build a system for robotic mobile manipulation that is robust to changes in these variables. This robustness will be enabled by recent simultaneous advances in the fields of big data, deep learning, and simulation. The ability of simulators to create realistic sensory data enables the generation of massive corpora of labeled training data for various grasping and navigation-based tasks. It is now possible to build systems that work in the real world trained using deep learning entirely on synthetic data. The ability to train and test on synthetic data allows for quick iterative development of new perception, planning and grasp execution algorithms that work in many environments. To build a robust system, this thesis introduces a novel multiple-view shape reconstruction architecture that leverages unregistered views of the object. To navigate to objects without localizing the agent, this thesis introduces a novel panoramic target goal architecture that takes previous views of the agent to inform a policy to navigate through an environment. Additionally, a novel next-best-view methodology is introduced to allow the agent to move around the object and refine its initial understanding of the object. The results show that this deep learned sim-to-real approach performs best when compared to heuristic-based methods in terms of reconstruction quality and success-weighted-by-path-length (SPL). This approach is also adaptable to the environment and robot chosen due to its modular design.