Orientador: Jacques Wainer / Tese (doutorado) - Universidade Estadual de Campinas, Instituto de Computação / Made available in DSpace on 2018-08-24T02:06:24Z (GMT). No. of bitstreams: 1 Pinheiro_PauloGurgel_D.pdf: 41476694 bytes, checksum: f3d5b1e2aa32aa6f00ef7ac689a261e2 (MD5) Previous issue date: 2013 / Resumo: Localização de robôs móveis é uma das áreas mais exploradas da robótica devido a sua importância para a resolução de problemas, como: navegação, mapeamento e SLAM. Muitos trabalhos apresentaram soluções envolvendo cooperação, comunicação e exploração do ambiente, onde em geral a localização é obtida através de ações randômicas ou puramente orientadas pelo estado de crença. Nesta tese, é apresentado um modelo de planejamento para localização utilizando POMDP e Localização de Markov, que indicaria a melhor ação que o robô deve efetuar em cada momento, com o objetivo de diminuir a quantidade de passos. O foco está principalmente em: i) problemas de difícil localização: onde não há landmark ou informação extra no ambiente que auxilie o robô, ii) situações de performance crítica: onde o robô deve evitar passos randômicos e o gasto de energia e, por último, iii) situações com múltiplas missões. Sabendo que um robô é projetado para desempenhar missões, será proposto, neste trabalho, um modelo onde essas missões são consideradas em paralelo com a localização. Planejar para cenários com múltiplos ambientes é um desafio devido a grande quantidade de estados que deve ser tratada. Para esse tipo de problema, será apresentado um modelo de compressão de mapas que utiliza padrões arquiteturais e de design, como: quantidade de portas, paredes ou área total de um ambiente, para condensar informações que possam ser redundantes. O modelo baseia-se na similaridade das características de desing para agrupar ambientes similares e combiná-los, gerando um único mapa representante que possui uma quantidade de estados menor que a soma total de todos os estados dos ambientes do grupo. Planos em POMDP são gerados apenas para os representantes e não para todo o mapa. Finalmente, será apresentado o modelo hierárquico onde a localização é executada em duas camadas. Na camada superior, o robô utiliza os planos POMDP e os mapas compactos para estimar a grossa estimativa de sua localização e, na camada inferior, utiliza POMDP ou Localização de Markov para a obtenção da postura mais precisa. O modelo hierárquico foi demonstrado com experimentos utilizando o simulador V-REP, e o robô Pioneer 3-DX. Resultados comparativos mostraram que o robô utilizando o modelo proposto, foi capaz de realizar o processo de localização em cenários com múltiplos ambientes e cumprir a missão, mantendo a precisão com uma significativa redução na quantidade de passos efetuados / Abstract: Mobile Robot localization is one of the most explored areas in robotics due to its importance for solving problems, such as navigation, mapping and SLAM. In this work, we are interested in solving global localization problems, where the initial pose of the robot is completely unknown. Several works have proposed solutions for localization focusing on robot cooperation, communication or environment exploration, where the robot's pose is often found by a certain amount of random actions or state belief oriented actions. In order to decrease the total steps performed, we will introduce a model of planning for localization using POMDPs and Markov Localization that indicates the optimal action to be taken by the robot for each decision time. Our focus is on i) hard localization problems, where there are no special landmarks or extra features over the environment to help the robot, ii) critical performance situation, where the robot is required to avoid random actions and the waste of energy roaming over the environment, and iii) multiple missions situations. Aware the robot is designed to perform missions, we have proposed a model that runs missions and the localization process, simultaneously. Also, since the robot can have different missions, the model computes the planning for localization as an offline process, but loading the missions at runtime. Planning for multiple environments is a challenge due to the amount of states we must consider. Thus, we also proposed a solution to compress the original map, creating a smaller topological representation that is easier and cheaper to get plans done. The map compression takes advantage of the similarity of rooms found especially in offices and residential environments. Similar rooms have similar architectural design features that can be shared. To deal with the compressed map, we proposed a hierarchical approach that uses light POMDP plans and the compressed map on the higher layer to find the gross pose, and on the lower layer, decomposed maps to find the precise pose. We have demonstrated the hierarchical approach with the map compression using both V-REP Simulator and a Pioneer 3-DX robot. Comparing to other active localization models, the results show that our approach allowed the robot to perform both localization and the mission in a multiple room environment with a significant reduction on the number of steps while keeping the pose accuracy / Doutorado / Ciência da Computação / Doutor em Ciência da Computação

The objective of this dissertation is to incorporate the concept of optimality to multi-modal control and apply the theoretical results to obtain successful navigation strategies for autonomous mobile robots. The main idea in multi-modal control is to breakup a complex control task into simpler tasks. In particular, number of control modes are constructed, each with respect to a particular task, and these modes are combined according to some supervisory control logic in order to complete the overall control task. This way of modularizing the control task lends itself particularly well to the control of autonomous mobile robot, as evidenced by the success of behavior-based robotics. Many challenging and interesting research issues arise when employing multi-modal control. This thesis aims to address these issues within an optimal control framework. In particular, the contributions of this dissertation are as follows: We first addressed the problem of inferring global behaviors from a collection of local rules (i.e., feedback control laws). Next, we addressed the issue of adaptively varying the multi-modal control system to further improve performance. Inspired by adaptive multi-modal control, we presented a constructivist framework for the learning from example problem. This framework was applied to the DARPA sponsored Learning Applied to Ground Robots (LAGR) project. Next, we addressed the optimal control of multi-modal systems with infinite dimensional constraints. These constraints are formulated as multi-modal, multi-dimensional (M3D) systems, where the dimensions of the state and control spaces change between modes to account for the constraints, to ease the computational burdens associated with traditional methods. Finally, we used multi-modal control strategies to develop effective navigation strategies for autonomous mobile robots. The theoretical results presented in this thesis are verified by conducting simulated experiments using Matlab and actual experiments using the Magellan Pro robot platform and the LAGR robot. In closing, the main strength of multi-modal control lies in breaking up complex control task into simpler tasks. This divide-and-conquer approach helps modularize the control system. This has the same effect on complex control systems that object-oriented programming has for large-scale computer programs, namely it allows greater simplicity, flexibility, and adaptability.

Incremental smoothing and mapping (iSAM) is presented, a novel approach to the simultaneous localization and mapping (SLAM) problem. SLAM is the problem of estimating an observer's position from local measurements only, while creating a consistent map of the environment. The problem is difficult because even very small errors in the local measurements accumulate over time and lead to large global errors. iSAM provides an exact and efficient solution to the SLAM estimation problem while also addressing data association. For the estimation problem, iSAM provides an exact solution by performing smoothing, which keeps all previous poses as part of the estimation problem, and therefore avoids linearization errors. iSAM uses methods from sparse linear algebra to provide an efficient incremental solution. In particular, iSAM deploys a direct equation solver based on QR matrix factorization of the naturally sparse smoothing information matrix. Instead of refactoring the matrix whenever new measurements arrive, only the entries of the factor matrix that actually change are calculated. iSAM is efficient even for robot trajectories with many loops as it performs periodic variable reordering to avoid unnecessary fill-in in the factor matrix. For the data association problem, I present state of the art data association techniques in the context of iSAM and present an efficient algorithm to obtain the necessary estimation uncertainties in real-time based on the factored information matrix. I systematically evaluate the components of iSAM as well as the overall algorithm using various simulated and real-world data sets.

We present automatic tools for configuring and deploying multi-robot networks of decentralized, mobile robots. These methods are tailored to the decentralized nature of the multi-robot network and the limited information available to each robot. We present methods for determining if user-defined network tasks are feasible or infeasible for the network, considering the limited range of its sensors. To this end, we define rigid and persistent feasibility and present necessary and sufficient conditions (along with corresponding algorithms) for determining the feasibility of arbitrary, user-defined deployments. Control laws for moving multi-robot networks in acyclic, persistent formations are defined. We also present novel Embedded Graph Grammar Systems (EGGs) for coordinating and deploying the network. These methods exploit graph representations of the network, as well as graph-based rules that dictate how robots coordinate their control. Automatic systems are defined that allow the robots to assemble arbitrary, user-defined formations without any reliance on localization. Further, this system is augmented to deploy these formations at the user-defined, global location in the environment, despite limited localization of the network. The culmination of this research is an intuitive software program with a Graphical User Interface (GUI) and a satellite image map which allows users to enter the desired locations of sensors. The automatic tools presented here automatically configure an actual multi-robot network to deploy and execute user-defined network tasks.

Thesis (MSc)--Stellenbosch University, 2012. / ENGLISH ABSTRACT: In this thesis we consider the problem of building a dense and consistent map of a mobile robot’s environment that is updated as the robot moves. Such maps are vital for safe and collision-free navigation. Measurements obtained from a range sensor mounted on the robot provide information on the structure of the environment, but are typically corrupted by noise. These measurements are also relative to the robot’s unknown pose (location and orientation) and, in order to combine them into a world-centric map, pose estimation is necessary at every time step. A SLAM system can be used for this task. However, since landmark measurements and robot motion are inherently noisy, the pose estimates are typically characterized by uncertainty. When building a map it is essential to deal with the uncertainties in range measurements and pose estimates in a principled manner to avoid overconfidence in the map. A literature review of robotic mapping algorithms reveals that the occupancy grid mapping algorithm is well suited for our goal. This algorithm divides the area to be mapped into a regular lattice of cells (squares for 2D maps or cubes for 3D maps) and maintains an occupancy probability for each cell. Although an inverse sensor model is often employed to incorporate measurement uncertainty into such a map, many authors merely state or depict their sensor models. We derive our model analytically and discuss ways to tailor it for sensor-specific uncertainty. One of the shortcomings of the original occupancy grid algorithm is its inability to convey uncertainty in the robot’s pose to the map. We address this problem by altering the occupancy grid update equation to include weighted samples from the pose uncertainty distribution (provided by the SLAM system). The occupancy grid algorithm has been criticized for its high memory requirements. Techniques have been proposed to represent the map as a region tree, allowing cells to have different sizes depending on the information received for them. Such an approach necessitates a set of rules for determining when a cell should be split (for higher resolution in a local region) and when groups of cells should be merged (for lower resolution). We identify some inconsistencies that can arise from existing rules, and adapt those rules so that such errors are avoided. We test our proposed adaptive occupancy grid algorithm, that incorporates both measurement and pose uncertainty, on simulated and real-world data. The results indicate that these uncertainties are included effectively, to provide a more informative map, without a loss in accuracy. Furthermore, our adaptive maps need far fewer cells than their regular counterparts, and our new set of rules for deciding when to split or merge cells significantly improves the ability of the adaptive grid map to mimic its regular counterpart. / AFRIKAANSE OPSOMMING: In hierdie tesis beskou ons die probleem om ’n digte en konsekwente kaart van ’n mobiele robot se omgewing te bou, wat opgedateer word soos die robot beweeg. Sulke kaarte is van kardinale belang vir veilige, botsingvrye navigasie. Metings verkry vanaf ’n sensor wat op die robot gemonteer is, verskaf inligting rakende die struktuur van die omgewing, maar word tipies deur ruis vervorm. Hierdie metings is ook relatief tot die robot se onbekende postuur (posisie en oriëntasie) en, om hulle saam te voeg in ’n wêreldsentriese kaart, is postuurafskatting nodig op elke tydstap. ’n SLAM stelsel kan vir hierdie doeleinde gebruik word. Aangesien landmerkmetings en die beweging van die robot inherent ruiserig is, word die postuurskattings gekarakteriseer deur onsekerheid. Met die bou van ’n kaart moet hierdie onsekerhede in afstandmetings en postuurskattings op ’n beginselvaste manier hanteer word om te verhoed dat te veel vertroue in die kaart geplaas word. ’n Literatuurstudie van karteringsalgoritmes openbaar die besettingsroosteralgoritme as geskik vir ons doel. Die algoritme verdeel die gebied wat gekarteer moet word in ’n reëlmatige rooster van selle (vierkante vir 2D kaarte of kubusse vir 3D kaarte) en onderhou ’n besettingswaarskynlikheid vir elke sel. Alhoewel ’n inverse sensormodel tipies gebruik word om metingsonsekerheid in so ’n kaart te inkorporeer, noem of wys baie outeurs slegs hulle model. Ons herlei ons model analities en beskryf maniere om sensorspesifieke metingsonsekerheid daarby in te sluit. Een van die tekortkominge van die besettingsroosteralgoritme is sy onvermoë om onsekerheid in die postuur van die robot na die kaart oor te dra. Ons spreek hierdie probleem aan deur die opdateringsvergelyking van die oorspronklike besettingsroosteralgoritme aan te pas, om geweegde monsters van die postuuronsekerheidsverdeling (verskaf deur die SLAM stelsel) in te sluit. Die besettingsroosteralgoritme word soms gekritiseer vir sy hoë verbruik van geheue. Tegnieke is voorgestel om die kaart as ’n gebiedsboom voor te stel, wat selle toelaat om verskillende groottes te hê, afhangende van die inligting wat vir hulle verkry is. So ’n benadering noodsaak ’n stel reëls wat spesifiseer wanneer ’n sel verdeel (vir ’n hoër resolusie in ’n plaaslike gebied) en wanneer ’n groep selle saamgevoeg (vir ’n laer resolusie) word. Ons identifiseer teenstrydighede wat kan voorkom as die huidige reëls gevolg word, en pas hierdie reëls aan sodat sulke foute vermy word. Ons toets ons voorgestelde aanpasbare besettingsroosteralgoritme, wat beide metings- en postuuronsekerheid insluit, op gesimuleerde en werklike data. Die resultate dui daarop dat hierdie onsekerhede op ’n effektiewe wyse na die kaart oorgedra word sonder om akkuraatheid prys te gee. Wat meer is, ons aanpasbare kaarte benodig heelwat minder selle as hul reëlmatige eweknieë. Ons nuwe stel reëls om te besluit wanneer selle verdeel of saamgevoeg word, veroorsaak ook ’n merkwaardige verbetering in die vermoë van die aanpasbare roosterkaart om sy reëlmatige eweknie na te boots.

This thesis describes novel machine learning techniques specifically designed for use in real-time embedded systems. The techniques directly address three major requirements of such learning systems. Firstly, learning must be capable of being achieved incrementally, since many applications do not have a representative training set available at the outset. Secondly, to guarantee real-time performance, the techniques must be able to operate within a deterministic and limited time bound. Thirdly, the memory requirement must be limited and known a priori to ensure the limited memory available to hold data in embedded systems will not be exceeded. The work described here has three principal contributions. The frequency table is a data structure specifically designed to reduce the memory requirements of incremental learning in embedded systems. The frequency table facilitates a compact representation of received data that is sufficient for decision tree generation. The frequency table decision tree (FTDT) learning method provides classification performance similar to existing decision tree approaches, but extends these to incremental learning while substantially reducing memory usage for practical problems. The incremental decision path (IDP) method is able to efficiently induce, from the frequency table of observations, the path through a decision tree that is necessary for the classification of a single instance. The classification performance of IDP is equivalent to that of existing decision tree algorithms, but since IDP allows the maximum number of partial decision tree nodes to be determined prior to the generation of the path, both the memory requirement and the execution time are deterministic. In this work, the viability of the techniques is demonstrated through application to realtime mobile robot navigation.

The major problem of robotics research today is that there is a barrier to entry into robotics research. Robot system software is complex and a researcher that wishes to concentrate on one particular problem often needs to learn about details, dependencies and intricacies of the complete system. This is because a robot system needs several different modules that need to communicate and execute in parallel. Today there is not much controlled comparisons of algorithms and solutions for a given task, which is the standard scientific method of other sciences. There is also very little sharing between groups and projects, requiring code to be written from scratch over and over again. This thesis proposes a general framework for robotics. By examining successful systems and architectures of past and present, yields a number of key properties. Some of these are ease of use, modularity, portability and efficiency. Even though there is much consensus on that the hybrid deliberate/reactive is the best architectural model that the community has produced so far, a framework should not stipulate a specific architecture. Instead the framework should enable the building of different architectures. Such a scheme implies that the modules are seen as common peers and not divided into clients and servers or forced into a set layering. Using a standardized middleware such as CORBA, efficient communication can be carried out between different platforms and languages. Middleware also provides network transparency which is valuable in distributed systems. Component-based Software Engineering (CBSE) is an approach that could solve many of the aforementioned problems. It enforces modularity which helps to manage complexity. Components can be developed in isolation, since algorithms are encapsulated in components where only the interfaces need to be known by other users. A complete system can be created by assembling components from different sources. Comparisons and sharing can greatly benefit from CBSE. A component-based framework called ORCA has been implemented with the following characteristics. All communication is carried out be either of three communication patterns, query, send and push. Communication is done using CORBA, although most of the CORBA code is hidden for the developer and can in the future be replaced by other mechanisms. Objects are transported between components in the form of the CORBA valuetype. A component model is specified that among other things include support for a state-machine. This also handles initialization and sets up communication. Configuration is achieved by the presence of an XML-file per component. A hardware abstraction scheme is specified that basically route the communication patterns right down to the hardware level. The framework has been verified by the implementation of a number of working systems.

[pt] O uso de robôs móveis para monitorar locais inacessíveis vem se tornando cada vez mais comum. Essas operações podem ser autônomas ou tripuladas e quando são feitas em terrenos irregulares é preciso garantir segurança na missão, pois muitas das vezes o resgate se torna inviável. O robô estudado nesta dissertação terá dificuldades para locomoção em certas localidades, como por exemplo: derrapagem em regiões alagadas, vencer atoleiro em regiões pantanosas e de brejos e capotagem nas regiões que apresentam aclives e declives. Diante deste quadro de problemas apresentados, garantir a estabilidade nas regiões de ladeiras é de grande valor nas operações, sejam elas tele-operadas ou autônomas. Visando contribuir para o sucesso da locomoção do robô, esta dissertação apresenta uma técnica de controle de estabilidade de um robô móvel para sensoreamento remoto em terrenos irregulares, incluindo projeto, simulação e construção de um protótipo funcional. Este controle visa garantir que as rodas do veículo não descolem do terreno, através da atuação nas forças de atrito entre as rodas e o solo variando os torques nos seus motores. / [en] The use of mobile robots to monitor non-accessible environments has become increasingly common in the recent years. These tasks can be either autonomous, remote-controlled, or passenger-operated. When performed in rough terrain, it is necessary to guarantee mission safety, since many times it is impossible to send a rescue party for recovery. The hybrid environmental robot presented in this thesis is a mobile robot that will face very challenging conditions, avoiding e.g. slippage in wet terrain, becoming trapped in muddy soil, and tipping over in regions with high slopes. Therefore, it is a challenging task to guarantee robot stability under such circumstances, either in autonomous or operated tasks. This thesis presents a stability control methodology for a mobile robot to perform remote sensing tasks in rough terrain. The model-based technique guarantees wheel-ground contact at all times, acting individually at the wheel motors to control the traction/friction forces. This work also addresses the design, simulation and construction aspects of a functional prototype of a mobile robot to validate the proposed approach.

O emprego de veículos autônomos é uma prática comumente adotada para a melhoria da produtividade no setor agrícola. No entanto, o custo computacional é um fator limitante na implementação desses dispositivos autônomos. A alternativa apresentada neste trabalho consistiu no desenvolvimento de um dispositivo de hardware dedicado para a navegação de robôs móveis agrícolas, o qual indica áreas navegáveis e não navegáveis, além do ângulo de inclinação do veículo em relação à linha de plantio. O desenvolvimento do projeto foi baseado em um método de extração de características visuais locais por meio do processamento de imagens coloridas obtidas por uma câmera de vídeo. O circuito foi implementado por meio de uma ferramenta de desenvolvimento baseado em um FPGA de baixo custo. O circuito consiste nas etapas de classificação, processamento morfológico e extração das linhas de navegação. Na primeira etapa, os pixels são classificados a partir do modelo de cores HSL em classes que representam as áreas passíveis e não passíveis de navegação. Posteriormente, a etapa de processamento morfológico realiza as tarefas de filtragem, agrupamento e extração de bordas. O processamento morfológico é realizado por meio de um arranjo de unidades de processamento dedicadas. Cada unidade pode realizar uma operação básica de morfologia matemática. O elemento estruturante utilizado na operação, bem como a operação realizada pela unidade, é configurado por meio de parâmetros do projeto. O processo de extração das linhas de orientação é realizado por meio do método de regressão linear por mínimos quadrados. A arquitetura proposta no projeto permitiu o processamento em tempo real de imagens para a aplicação de navegação autônoma de robôs móveis em ambientes agrícolas. / The use of autonomous vehicles is a generally adopted practice to improve the productivity in the agriculture sector. However, the computer requirements are a limiting factor for implementation of these autonomous devices. The alternative shown in this paper is the design of a dedicated hardware for the autonomous agricultural robot navigation. The project development was based on a local visual feature extraction method by processing digital images obtained from a color video camera. The circuit was implemented through a development tool based on a low cost FPGA. The circuit consists of stages of classification, morphological processing and guidance line extraction. In the first stage, the pixels are classified through HSL color model into classes that represent suitable and unsuitable area for navigation. Then, the morphological processing stage performs filtering, grouping and edge detection tasks. The morphological processing is carried out by an arrangement of dedicated processing units. Each unit can perform a basic operation of mathematical morphology. The structuring element used in the operation and the operation performed by the unit are configured through project parameters. The guidance line extraction process is performed through the linear regression method by least square. The architecture proposed in the design allowed the real-time image processing in autonomous robot navigation applications in agricultural environments.

O emprego de técnicas de controle para veículos submarinos, envolve muitas questões de interesse prático e teórico. Neste trabalho apresenta-se o desenvolvimento de um sistema de controle inteligente e adaptativo a ser aplicado na navegação de veículos submarinos semi-autônomos (VSSAs). Utiliza-se uma técnica baseada no controle nebuloso (fuzzy), visando gerenciar o veículo submarino no controle de velocidade, profundidade, orientação e na evasão de obstáculos. A operação de veículos submarinos usando a técnica proposta, exige a definição, análise e tratamento de um vasto conjunto de comandos complexos manipulados pelo controlador. A metodologia utilizada divide a ação de controle em 3 fases. A primeira trata do posicionamento inicial do veículo submarino, a segunda fase trata da sua navegação e a fase final de gerenciar o comportamento do veículo próximo da posição-objetivo. A implementação funcional do controlador e dividida em módulos. O primeiro módulo informa o comportamento do ambiente e do próprio veículo, fornecendo dados iniciais sobre seu posicionamento e sua profundidade; um segundo módulo trata da presença de obstáculos em diferentes direções com dados fornecidos por sonares e assim determina as ações para a evasão de obstáculos. A ação de controle e estabelecida usando conceitos da teoria nebulosa (fuzzy) no universo de discurso, através de variáveis lingüísticas e de regras de inferência definidas a partir do conhecimento de especialistas, que envolvem a imprecisão característica do comportamento humano. As informações no final do processo são concentradas, de forma que a ação de controle e determinada para que possa enviar sinais de controle aos atuadores. / The design of underwater vehicle involves a very large number of practical and theoretical problems. In this work, it is tackled the development of a intelligent and adaptive controller, to be used in the navigation of semi-autonomous underwater vehicles (SAUV). To achieve this goal, a technique based on the fuzzy theory was employed to control the vehicle movements, including the evasion of obstacles. The operation of underwater vehicles using this approach demands the definition and treatment of a vast set of complex commands, manipulated by the controller. The control action is subdivided into three stages, the first one deals the control action during the initial positioning of the vehicle, a second stage deals with the navigation itself and the final stage deals the control action when the vehicle is close to the objective-position. The functional development of the controller was also subdivided into modules. The first module deals with the management of input data such as environmental disturbances and initial vehicle position, such as depth of the vehicle. The second module deals the detection of obstacles in different directions and the optimal evasion action to avoid collisions. The control action during a mission is established using concepts of the Fuzzy theory in the universe of speech, through linguistic variables and rules of inference defined from the knowledge of specialists, involving the characteristic imprecision of human behavior. In the end of the process, the information is defuzzificated, so that control actions are determined, allowing a practical implementation.