Dexterous robotic manipulation of non-rigid objects is a challenging problem but necessary to explore as robots are increasingly interacting with more complex environments in which such objects are frequently present. In particular, common manipulation tasks such as molding clay to a target shape or picking fruits and vegetables for use in the kitchen, require a high-level understanding of the scene and objects. Commonly, the behavior of non-rigid objects is described by a model. Although, well-established modeling techniques are difficult to apply in robotic tasks since objects and their properties are unknown in such unstructured environments. This work proposes a sensing and modeling framework to measure the 3D shape deformation of non-rigid objects. Unlike traditional methods, this framework explores data-driven learning techniques focused on shape representation and deformation dynamics prediction using a graph-based approach. The proposal is validated experimentally, analyzing the performance of the representation model to capture the current state of the non-rigid object shape. In addition, the performance of the prediction model is analyzed in terms of its ability to produce future states of the non-rigid object shape due to the manipulation actions of the robotic system. The results suggest that the representation model is able to produce graphs that closely capture the deformation behavior of the non-rigid object. Whereas, the prediction model produces visually plausible graphs when short-term predictions are required.

abstract: Humans are capable of transferring learning for anticipatory control of dexterous object manipulation despite changes in degrees-of-freedom (DoF), i.e., switching from lifting an object with two fingers to lifting the same object with three fingers. However, the role that tactile information plays in this transfer of learning is unknown. In this study, subjects lifted an L-shaped object with two fingers (2-DoF), and then lifted the object with three fingers (3-DoF). The subjects were divided into two groups--one group performed the task wearing a glove (to reduce tactile sensibility) upon the switch to 3-DoF (glove group), while the other group did not wear the glove (control group). Compensatory moment (torque) was used as a measure to determine how well the subject could minimize the tilt of the object following the switch from 2-DoF to 3-DoF. Upon the switch to 3-DoF, subjects wearing the glove generated a compensatory moment (Mcom) that had a significantly higher error than the average of the last five trials at the end of the 3-DoF block (p = 0.012), while the control subjects did not demonstrate a significant difference in Mcom. Additional effects of the reduction in tactile sensibility were: (1) the grip force for the group of subjects wearing the glove was significantly higher in the 3-DoF trials compared to the 2-DoF trials (p = 0.014), while the grip force of the control subjects was not significantly different; (2) the difference in centers of pressure between the thumb and fingers (ΔCoP) significantly increased in the 3-DoF block for the group of subjects wearing the glove, while the ΔCoP of the control subjects was not significantly different; (3) lastly, the control subjects demonstrated a greater increase in lift force than the group of subjects wearing the glove (though results were not significant). Combined together, these results suggest different force modulation strategies are used depending on the amount of tactile feedback that is available to the subject. Therefore, reduction of tactile sensibility has important effects on subjects' ability to transfer learned manipulation across different DoF contexts. / Dissertation/Thesis / M.S. Bioengineering 2014

This thesis presents several new techniques for developing safe and flexible human-robot interaction tasks where human operators cooperate with robotic manipulators. The contributions of this thesis are divided in two fields: the development of safety strategies which modify the normal behavior of the robotic manipulator when the human operator is near the robot and the development of dexterous manipulation tasks for in-hand manipulation of objects with a multi-fingered robotic hand installed at the end-effector of a robotic manipulator. / Valencian Government by the research project "Infraestructura 05/053". Spanish Ministry of Education and Science by the pre-doctoral grant AP2005-1458 and the research projects DPI2005-06222 and DPI2008-02647, which constitute the research framework of this thesis.

Ce travail s’inscrit dans le cadre de la conception et réalisation d’un robot d’assistance à la personne. Dans cette thèse, nous nous intéressons particulièrement à la conception, à la modélisation et à la commande d’un robot manipulateur mobile. La conception mécanique couplée à un outil de simulation dynamique multi-corps nous a permis d’obtenir un modèle virtuel très réaliste. Le modèle cinématique du système a été obtenu en utilisant la méthode D-H modifiée. L’approche Bond graph et la méthode de Lagrange ont permis de construire le modèle dynamique. Un algorithme hybride qui combine la pseudoinverse du jacobien et la méthode RRT a été proposé pour la planification de mouvement d’un manipulateur redondant et rechercher de configurations continues, stables et sans collision. Un contrôleur basé sur les réseaux de neurones a été introduit pour la commande coordonnée d’un manipulateur mobile. Cette méthode ne nécessite pas un modèle précis du robot. Les paramètres inconnus sont identifiés et compensés en utilisant des réseaux de neurones RBF. Un algorithme de contrôle similaire est présenté pour la commande force/position d’un manipulateur mobile qui est soumis à des contraintes holonomes et nonholonomes. L’étude de la main robotique a été effectuée séparément avant d’être couplée au reste du système. Les modèles cinématique et dynamique du système main-objet ont été obtenus en utilisant les approches mathématiques et bond graph. Un algorithme est proposé afin d’assurer une prise ferme, éviter les dérapages et suivre les mouvements désirés. Les validations des modèles et des différentes lois de commande ont été effectuées grâce à la co-simulation Matlab/modèle virtuel / The purpose of this thesis is to design, model and control of a personal assistant robot used for domestic tasks. In order to make the robot’s design more efficient, a virtual simulation system is built using dynamic simulation software. The kinematic model is set up based on modified D-H principle. The dynamic model is built using the Lagrange theorem and elaborated in Matlab. We also employ an energy-based approach for modeling and its bond graph notation ensures encapsulation of functionality, extendibility and reusability of each element of the model. A hybrid algorithm of combining the Jacobian pseudoinverse algorithm with Rapidly-Exploring Random Tree method is presented for collision-free path planning of a redundant manipulator. An intelligent robust controller based on neural network is introduced for the coordinated control of a mobile manipulator. This method does not require an accurate model of the robot. Unknown dynamic parameters of the mobile platform and the manipulator are identified and compensated in closed-loop control using RBF neural network. A similar control algorithm is presented for coordinated force/motion control of a mobile manipulator suffering both holonomic and nonholonomic constraints. Kinematics and dynamics of a dexterous hand manipulating an object with known shape by rolling contacts are derived. A computed torque control algorithm is presented to ensure firm grip, avoid slippage and well track a given motion imposed to the object. The validation of models and different control laws were made by the co-simulation Matlab / virtual model

Les travaux de cette thèse portent sur la saisie et la manipulation dextre et ont pour dénominateur commun la robustesse vis-à-vis d'un environnement incertain (méconnaissance de la géométrie de l'objet ou du préhenseur, initialisation imparfaite du système, etc). La mesure de qualité de prise permet d'évaluer la stabilité d'une saisie. Nos travaux proposent une nouvelle mesure de qualité de prise, dont le principe reste dans la continuité des méthodes les plus connues qui consistent à déterminer l'espace des torseurs dynamiques applicables sur l'objet par le préhenseur. Notre mesure cherche à déterminer cet espace quelle que soit l'incertitude qui affecte le système préhenseur/objet. On appelle cet ensemble le Reachable Wrench Space under Uncertainties (RWSU). Deux algorithmes sont proposés afin de déterminer un majorant et un minorant du RWSU. La deuxième contribution concerne l'application d'algorithmes de commande robuste aux incertitudes de modèle pour la manipulation dextre. La première méthode de commande que nous proposons est un retour d'état, permettant de répondre à la consigne de manipulation, auquel on ajoute une action dynamique, permettant de répondre aux contraintes de saisie. Le retour d'état est synthétisé suivant un problème d'optimisation avec contraintes LMI. Les contraintes LMI permettent de définir la réponse dynamique du système bouclé, et d'assurer la robustesse aux incertitudes de modèle. Une seconde méthode de commande est proposée afin d'améliorer les performances de suivi de trajectoire pour ce système MIMO en découplant le mouvement à suivre des mouvements perturbateurs résultant des couplages dynamiques entre les axes. / This thesis deals with grasping and dexterous manipulation with multifingered hands, with the robustness to uncertain environments as a common denominator.The first contribution of the present work is a new measure of the grasp quality, which is used to evaluate the stability of a grasp. In the footsteps of the most known methods which consist in determining the reachable wrench space, our measure aims to evaluate this space whatever the uncertainty which affects the gripper/object system. This new space is called Reachable Wrench Space under Uncertainty (RWSU). Two algorithms are proposed to find respectively an upper and a lower bound of the RWSU.The second contribution concerns the application of robust control algorithms for dexterous manipulation. The first control method is composed of as a state-space feedback, which enables a manipulation task, and of an additional dynamic action, allowing to respect the grasp constraints. The state-space feedback is designed for a robust regional pole placement by the resolution of an optimization problem under LMI constraints. The LMI constraints define the dynamic response of the system in closed loop, and ensure the robustness with respect to model uncertainties. A second control method, based on eigenstructure assignment, is proposed to improve the trajectory tracking for the MIMO system. The eigenstructure assignment decouples the movement of the task from the disturbing movements resulting from the dynamic coupling between the axes.

Nous nous intéressons à la conception des mains mécaniques anthropomorphes destinées à manipuler des objets dans un environnement humain. Via l'analyse du mouvement de sujets humains lors d'une tâche de manipulation de référence, nous proposons une méthode pour évaluer la capacité des mains robotiques à manipuler les objets. Nous montrons comment les rapports de couplage angulaires entre les articulations et les limites articulaires, influent sur l'aptitude à manipuler dynamiquement des objets. Nous montrons également l'impact du poignet sur les tâches de manipulation rapides. Nous proposons une stratégie pour calculer les forces de manipulation en bout de doigts et dimensionner les moteurs d'un tel préhenseur. La méthode proposée est dépendante de la tâche visée et s'adapte à tout type de mouvement dès lors qu'il peut être capturé et analysé. Dans une deuxième partie, consacrée aux robots manipulateurs, nous élaborons des algorithmes de commande optimale. En considérant l'énergie cinétique du robot comme une métrique, le modèle dynamique est formulé sous forme tensorielle dans le cadre de la géométrie Riemannienne. La discrétisation temporelle est basée sur les Éléments Finis d'Hermite. Nous intégrons les équations de Lagrange du mouvement par une méthode de perturbation. Des exemples de simulation illustrent la superconvergence de la technique d'Hermite. Le critère de contrôle est choisi indépendant des paramètres de configuration. Les équations de la commande associées aux équations du mouvement se révèlent covariantes. La méthode de commande optimale proposée consiste à minimiser la fonction objective correspondant au critère invariant sélectionné. / We focus on the design of anthropomorphous mechanical hands destined to manipulate objects in a human environment. Via the motion analysis of a reference manipulation task performed by human subjects, we propose a method to evaluate a robotic hand manipulation capacities. We demonstrate how the angular coupling between the fingers joints and the angular limits affect the hands potential to manipulate objects. We also show the influence of the wrist motions on the manipulation task. We propose a strategy to calculate the fingertip manipulation forces and dimension the fingers motors. In a second part devoted to articulated robots, we elaborate optimal control algorithms. Regarding the kinetic energy of the robot as a metric, the dynamic model is formulated tensorially in the framework of Riemannian geometry. The time discretization is based on the Hermite Finite Elements.A time integration algorithm is designed by implementing a perturbation method of the Lagrange's motion equations. Simulation examples illustrate the superconvergence of the Hermite's technique. The control criterion is selected to be coordinate free. The control equations associated with the motion equations reveal to be covariant. The suggested control method consists in minimizing the objective function corresponding to the selected invariant criterion.