Este trabalho apresenta o desenvolvimento, implementação e análise de técnicas de controle H 'INFINITO' não lineares para robôs manipuladores subatuados, sujeitos a incertezas paramétricas e distúrbios externos. Na primeira parte, duas abordagens são consideradas para robôs manipuladores individuais subatuados. A primeira abordagem consiste em representar robôs manipuladores como um sistema não linear na forma quase-linear com parâmetros variantes e utilizar técnicas de controle H 'INFINITO' para sistemas lineares a parâmetros variantes baseadas em desigualdades matriciais lineares. Na segunda abordagem, uma solução explícita do problema de controle H 'INFINITO' não linear para robôs manipuladores é encontrada via teoria dos jogos diferenciais. Com este mesmo procedimento, implementam-se também os controles misto H'IND.2'/H 'INFINITO' não linear, adaptativo H 'INFINITO' não linear e adaptativo H 'INFINITO' não linear com redes neurais para robôs manipuladores. Também é desenvolvido um sistema tolerante a falhas para robôs manipuladores baseado em sistemas Markovianos e em controladores Markovianos H'IND.2', H 'INFINITO' e H'IND.2'/ H 'INFINITO'. Na segunda parte, o modelo dinâmico de robôs manipuladores cooperativos subatuados é representado na forma de espaço de estados, possibilitando a aplicação dos controladores H 'INFINITO' não lineares para controle de posição, juntamente com controle das forças de esmagamento, de um objeto. / This work presents the development, implementation and analysis of nonlinear H‡ control techniques applied to underactuated manipulators, under parametric uncertainties and external disturbances. At the first part, two approaches are considered for underactuated individual manipulators. The first approach consists in representing manipulators as nonlinear systems in the quasi-linear parameter varying form and in controlling them via H‡ control for linear parameter varying systems based on linear matrix inequalities. At the second approach, an explicit solution to the nonlinear H‡ control problem for manipulators is found via differential game theory. With this procedure, it is also implemented the nonlinear mixed H2/H‡, nonlinear adaptive H‡, and nonlinear adaptive H‡ with neural networks controls. Also is developed a fault tolerant system for manipulators based on Markovian systems and Markovian H2, H‡, and H2/H‡ controls. At the second part, the dynamic model of underactuated cooperative manipulators is represented in the state space form in order to apply the nonlinear H‡ controls to position control, plus the squeeze force control, of an object.

Este trabalho apresenta o desenvolvimento, implementação e análise de técnicas de controle H 'INFINITO' não lineares para robôs manipuladores subatuados, sujeitos a incertezas paramétricas e distúrbios externos. Na primeira parte, duas abordagens são consideradas para robôs manipuladores individuais subatuados. A primeira abordagem consiste em representar robôs manipuladores como um sistema não linear na forma quase-linear com parâmetros variantes e utilizar técnicas de controle H 'INFINITO' para sistemas lineares a parâmetros variantes baseadas em desigualdades matriciais lineares. Na segunda abordagem, uma solução explícita do problema de controle H 'INFINITO' não linear para robôs manipuladores é encontrada via teoria dos jogos diferenciais. Com este mesmo procedimento, implementam-se também os controles misto H'IND.2'/H 'INFINITO' não linear, adaptativo H 'INFINITO' não linear e adaptativo H 'INFINITO' não linear com redes neurais para robôs manipuladores. Também é desenvolvido um sistema tolerante a falhas para robôs manipuladores baseado em sistemas Markovianos e em controladores Markovianos H'IND.2', H 'INFINITO' e H'IND.2'/ H 'INFINITO'. Na segunda parte, o modelo dinâmico de robôs manipuladores cooperativos subatuados é representado na forma de espaço de estados, possibilitando a aplicação dos controladores H 'INFINITO' não lineares para controle de posição, juntamente com controle das forças de esmagamento, de um objeto. / This work presents the development, implementation and analysis of nonlinear H‡ control techniques applied to underactuated manipulators, under parametric uncertainties and external disturbances. At the first part, two approaches are considered for underactuated individual manipulators. The first approach consists in representing manipulators as nonlinear systems in the quasi-linear parameter varying form and in controlling them via H‡ control for linear parameter varying systems based on linear matrix inequalities. At the second approach, an explicit solution to the nonlinear H‡ control problem for manipulators is found via differential game theory. With this procedure, it is also implemented the nonlinear mixed H2/H‡, nonlinear adaptive H‡, and nonlinear adaptive H‡ with neural networks controls. Also is developed a fault tolerant system for manipulators based on Markovian systems and Markovian H2, H‡, and H2/H‡ controls. At the second part, the dynamic model of underactuated cooperative manipulators is represented in the state space form in order to apply the nonlinear H‡ controls to position control, plus the squeeze force control, of an object.

Robots are now moving from their conventional confined habitats such as factory floors to human environments where they assist and physically interact with people. The requirement for inherent mechanical safety is overarching in such human-robot interaction systems. We propose a dual actuator called Parallel Force/Velocity Actuator (PFVA) that combines a Force Actuator (FA) (low velocity input) and a Velocity Actuator (VA) (high velocity input) using a differential gear train. In this arrangement mechanical safety can be achieved by limiting the torque on the FA and thus making it a backdriveable input. In addition, the kinematic redundancy in the drive can be used to control output velocity while satisfying secondary operational objectives. Our research focus was on three areas: (i) scalable parametric design of the PFVA, (ii) analytical modeling of the PFVA and experimental testing on a single-joint prototype, and (iii) generalized model formulation for PFVA-driven serial robot manipulators. In our analysis, the ratio of velocity ratios between the FA and the VA, called the relative scale factor, emerged as a purely geometric and dominant design parameter. Based on a dimensionless parametric design of PFVAs using power-flow and load distributions between the inputs, a prototype was designed and built using commercial-off-the-shelf components. Using controlled experiments, two performance-limiting phenomena in our prototype, friction and dynamic coupling between the two inputs, were identified. Two other experiments were conducted to characterize the operational performance of the actuator in velocity-mode and in what we call ‘torque-limited’ mode (i.e. when the FA input can be backdriven). Our theoretical and experimental results showed that the PFVA can be mechanical safe to both slow collisions and impacts due to the backdriveability of the FA. Also, we show that its kinematic redundancy can be effectively utilized to mitigate low-velocity friction and backlash in geared mechanisms. The implication at the system level of our actuator level analytical and experimental work was studied using a generalized dynamic modeling framework based on kinematic influence coefficients. Based on this dynamic model, three design case studies for a PFVA-driven serial planar 3R manipulator were presented. The major contributions of this research include (i) mathematical models and physical understanding for over six fundamental design and operational parameters of the PFVA, based on which approximately ten design and five operational guidelines were laid out, (ii) analytical and experimental proof-of-concept for the mechanical safety feature of the PFVA and the effective utilization of its kinematic redundancy, (iii) an experimental methodology to characterize the dynamic coupling between the inputs in a differential-summing mechanism, and (iv) a generalized dynamic model formulation for PFVA-driven serial robot manipulators with emphasis on distribution of output loads between the FA and VA input-sets. / text

La problématique traitée dans cette thèse concerne la commande de robots manipulateurs à articulations flexibles. Les méthodes développées visent à satisfaire les spécifications de performance et de robustesse en suivi de trajectoire, ainsi qu’à assurer un niveau de sécurité compatible avec un scénario de fonctionnement interactif dans lequel l’homme et le robot partagent un même espace de travail. Seules les mesures moteur sont utilisées dans un contexte d’instrumentation réduite. Le premier objectif de performance de la commande de mouvement est atteint grâce à l’identification expérimentale d’un modèle flexible représentatif du système, et l’usage de ce modèle pour la synthèse de lois de commande avancées intégrées au sein d’une structure cascade. Deux approches complémentaires fondées d’une part sur la commande prédictive de type GPC (Generalized Predictive Control), et d’autre part sur la commande Hinfini, sont considérées pour la synthèse de lois de commande à deux degrés de liberté, prédictives et robustes. Les performances de ces deux approches sont analysées et évaluées expérimentalement. Le deuxième objectif de sécurité est abordé à travers un algorithme de détection de collisions du robot avec son environnement, sans capteur d’effort et en présence d’incertitudes de modélisation. Afin de séparer efficacement les effets dynamiques des collisions de ceux des erreurs de modélisation, une stratégie adaptative de filtrage et de décision tenant compte de l’état du système est proposée. La validation expérimentale montre une très bonne sensibilité de détection, compatible avec les normes et les recommandations de sécurité relatives à la robotique collaborative. / The present thesis addresses the problem of motion control of flexible-joint robot manipulators using motor sensors only. The global objective is to guarantee tracking performance and robustness with respect to modeling uncertainties, together with safe human-robot interaction in a collaborative scenario where the robot and the human operator share the same workspace. The first objective of performance is achieved through the experimental identification of a flexible model of the system and the use of this model for the design of advanced control laws implemented in a cascade structure. Two complementary approaches, based either on predictive (Generalized Predictive Control, GPC) or Hinfinity control frameworks, are considered to design predictive and robust two degrees-of-freedom controllers. Experimental evaluation and analysis of the proposed strategies is provided. The second objective of safety is addressed by a novel algorithm for human-robot collision detection, without force sensors and in the presence of modeling uncertainties. In order to efficiently separate the dynamic effects of the collisions from the effects due to modeling errors, the proposed approach includes adaptive filtering and uses dynamic thresholds depending on the robot state. Experimental evaluation demonstrates a good detection sensitivity which is consistent with safety standards and recommendations for collaborative robotics.

Contrairement aux robots manipulateurs industriels qui sont de taille énorme et de prix élevé, beaucoup de robots manipulateurs à bas prix sont déjà entrés dans le marché, avec une petite taille, un poids léger, ce type de robots est plus accessible pour les particuliers. Cependant, limité par le coût de revient, des accessoires (matériaux, actuateurs, contrôleurs, etc) adoptés sont aussi limités, cela conduit souvent à la performance moins robuste au niveau de contrôle. Cette thèses se concentre sur la conception de contrôleur pour améliorer la performance des robots manipulateurs à bas prix. D'abord, pour des robots manipulateurs rigides, la modélisation dynamique en lien avec le système d'actualisation est établie, qui forme une équation différentielle avec paramètres constants et perturbation. Une méthode d'identification des paramètres en utilisant des observateurs et une commande adaptative sont proposées, et des résultats de simulation et d'expérimentation sont donnés. Ensuite, pour le cas d'articulation flexibles, pour simplifier, le modèle 1DOF est pris en compte. Premièrement, avec la mesure de la vitesse de lien, une méthode d'identification et une loi deux-étages adaptative sont proposées à condition que la position statique de lien puisse également être mesurée, des résultats de simulation sont donnés. Deuxièmement, en utilisant des mesures d'accélération de lien, une méthode d'identification et la même loi deux-étages adaptative sont proposées, cette idée est généralisée à l'identification et au contrôle de systèmes linéaires avec mesures de dérivées d'ordre élevé, des résultat de simulation sont présentés. Pour la mise en œuvre, des capteurs inertiels (gyroscopes et accéléromètres) sont utilisés et des résultats expérimentaux sont présentés. / Unlike industrial robot manipulators which are huge in size and of high price, many low-cost robot manipulators have already entered the market, with small size and light weight, this type of robots are more accessible to the public. However, limited by the cost, the components adopted (materials, actuators, controllers, etc.) are also limited, this often leads to less robust control performance. This thesis focuses on the controller design to improve the performance for such kind low-cost robot manipulators. To start with, for rigid case, dynamic modeling considering the actuator system is established, which forms a differential equation with constant parameters and disturbance, a method to identify the model parameters using observers and then an adaptive controller are proposed, simulation and experimental results are given. Then, in case of flexible joints, for simplicity, a single-link case model is considered. Firstly, link velocity measurement is assumed to provide link information, and an identification method and a two-stage adaptive control low are proposed provided that the static link position can also be measured, simulation result is given. Secondly, by using link acceleration measurement, an identification method and the same two-stage adaptive control low areproposed, this idea is generalized to identification and control of linear system using high-order derivative measurements, simulation result is presented. For implementation, inertial sensors (gyro and accelerometer) are used and experimental result is presented.

La problématique traitée dans cette thèse vise à développer une stratégie efficace de détection et de classification des impacts en présence d'incertitudes de modélisation du robot et de son environnement et en utilisant un nombre minimal de capteurs, notamment en l'absence de capteur d’effort.La première partie de la thèse porte sur la détection d'un impact pouvant avoir lieu à n'importe quel endroit du bras robotique et à n'importe quel moment de sa trajectoire. Les méthodes de détection d’impacts sont généralement basées sur un modèle dynamique du système, ce qui les rend sujettes au compromis entre sensibilité de détection et robustesse aux incertitudes de modélisation. A cet égard, une méthodologie quantitative a d'abord été mise au point pour rendre explicite la contribution des erreurs induites par les incertitudes de modèle. Cette méthodologie a été appliquée à différentes stratégies de détection, basées soit sur une estimation directe du couple extérieur, soit sur l'utilisation d'observateurs de perturbation, dans le cas d’une modélisation parfaitement rigide ou à articulations flexibles. Une comparaison du type et de la structure des erreurs qui en découlent et de leurs conséquences sur la détection d'impacts en a été déduite. Dans une deuxième étape, de nouvelles stratégies de détection d'impacts ont été conçues: les effets dynamiques des impacts sont isolés en déterminant la marge d'erreur maximale due aux incertitudes de modèle à l’aide d’une approche stochastique.Une fois l'impact détecté et afin de déclencher la réaction post-impact du robot la plus appropriée, la deuxième partie de la thèse aborde l'étape de classification. En particulier, la distinction entre un contact intentionnel (l'opérateur interagit intentionnellement avec le robot, par exemple pour reconfigurer la tâche) et un contact non-désiré (un sujet humain heurte accidentellement le robot), ainsi que la localisation du contact sur le robot, est étudiée en utilisant des techniques d'apprentissage supervisé et plus spécifiquement des réseaux de neurones feedforward. La généralisation à plusieurs sujet humains et à différentes trajectoires du robot a été étudiée. / The present thesis aims to develop an efficient strategy for impact detection and classification in the presence of modeling uncertainties of the robot and its environment and using a minimum number of sensors, in particular in the absence of force/torque sensor.The first part of the thesis deals with the detection of an impact that can occur at any location along the robot arm and at any moment during the robot trajectory. Impact detection methods are commonly based on a dynamic model of the system, making them subject to the trade-off between sensitivity of detection and robustness to modeling uncertainties. In this respect, a quantitative methodology has first been developed to make explicit the contribution of the errors induced by model uncertainties. This methodology has been applied to various detection strategies, based either on a direct estimate of the external torque or using disturbance observers, in the perfectly rigid case or in the elastic-joint case. A comparison of the type and structure of the errors involved and their consequences on the impact detection has been deduced. In a second step, novel impact detection strategies have been designed: the dynamic effects of the impacts are isolated by determining the maximal error range due to modeling uncertainties using a stochastic approach.Once the impact has been detected and in order to trigger the most appropriate post-impact robot reaction, the second part of the thesis focuses on the classification step. In particular, the distinction between an intentional contact (the human operator intentionally interacts with the robot, for example to reconfigure the task) and an undesired contact (a human subject accidentally runs into the robot), as well as the localization of the contact on the robot, is investigated using supervised learning techniques and more specifically feedforward neural networks. The challenge of generalizing to several human subjects and robot trajectories has been investigated.

Ce travail présente une nouvelle approche de synthèse de la commande non linéaire en suivi de trajectoire de robots manipulateurs. Malgré la richesse de la littérature dans le domaine, le problème n'a pas encore été traité de manière adéquate : en raison de l'existence inévitable dans les applications pratiques de perturbations et incertitudes telles que les forces de frottement, des perturbations externes ou les variations des paramètres il est difficile d'assurer un suivi de trajectoire de haute précision. Afin de résoudre ce problème, nous proposons tout d'abord une méthode de commande prenant en compte la performance H∞ pour le suivi de trajectoire d'un robot manipulateur. Deuxièmement, nous proposons un nouveau cadre pour la synthèse de lois de commande combinant une action anticipatrice et un retour d'état basée sur une représentation sous forme Takagi-Sugeno descripteur de la dynamique du manipulateur. Un avantage de la représentation choisie est de pouvoir simultanément simplifier le calcul des gains de commande à l'aide de LMI de dimension réduite et de réduire la complexité du correcteur en agissant sur le nombre de règles du modèle de Takagi-Sugeno. Basé sur la théorie de la stabilité de Lyapunov, le réglage du correcteur est formulé comme un problème d'optimisation LMI (inégalité matricielle linéaire). Les résultats obtenus en simulation effectuée avec un modèle de manipulateur série développé dans l'environnement Simscape MultibodyTM de Matlab R démontrent clairement l'efficacité de la méthode proposée en comparaison avec le régulateur PID et la commande CTC (Computed Torque Control). / This work presents a new design approach for trajectory tracking control of robot manipulators. In spite of the rich literature in the field, the problem has not yet been addressed adequately due to the lack of an effective control design. In general, it is difficult to adopt design to achieve high-precision tracking control due to the uncertainties in practical applications, such as friction forces, external disturbances and parameter variations. In order to cope this problem, we propose first control with H∞ performance to reference trajectory tracking control of two degrees of freedom robot. Secondly, we propose a new design framework with parametric uncertainties and unknown disturbances by using the feedback and the feedforward controllers. Using the descriptor Takagi-Sugeno systems, the design goal is to achieve a guaranteed tracking performance while signicantly reducing the numerical complexity of the designed controller through a robust control scheme. Based on Lyapunov stability theory, the control design is formulated as an LMI (linear matrix inequality) optimization problem. Simulation results carried out with a high-fidelity serial manipulator model embedded in the Simscape MultibodyTM environment of MatlabR clearly demonstrate the effectiveness of the proposed method by comparing with PID controller and computed torque controller.