Dimensional variation analysis (DVA) is a computer based simulation process used to identify potential assembly process issues due the effects of component part and assembly variation during manufacture. The sponsoring company has over a number of years developed a DVA process to simulate the variation behaviour of a wide range of static mechanical systems. This project considers whether the current DVA process used by the sponsoring company is suitable for the simulation of complex kinematic systems. The project, which consists of three case studies, identifies several issues that became apparent with the current DVA process when applied to three types of complex kinematic systems. The project goes on to develop solutions to the issues raised in the case studies in the form of new or enhanced methods of information acquisition, simulation modelling and the interpretation and presentation of the simulation output Development of these methods has enabled the sponsoring company to expand the range of system types that can be successfully simulated and significantly enhances the information flow between the DVA process and the wider product development process.

Les accidents vasculaires cérébraux (AVC) constituent la 1ère cause de mortalité chez les femmes et la 3ème chez les hommes ainsi que la 1ère cause de dépendance chez les personnes qui survivent puisqu'ils gardent des séquelles neurologiques majeures. Les déficiences sont majoritairement fonctionnelles et seulement 5 à 25 % des patients vont récupérer un membre supérieur fonctionnel. La récupération fonctionnelle spontanée, maximale dans les 3 mois après la survenue de l'AVC est principalement sous-tendue par des mécanismes de plasticité cérébrale. Mécanismes qui peuvent être boostés par les différentes techniques de réhabilitation parmi lesquelles les substances pharmacologiques ou les techniques de stimulation cérébrale qui sont développées. Quantifier les capacités de chaque patient est indispensable pour déterminer les fonctions perdues et ainsi proposer des protocoles de rééducation adéquats. De nombreuses techniques ont démontré ce potentiel d'évaluation. Notamment, l'imagerie par résonnance magnétique fonctionnelle (IRMf) d'activation (i.e. lors de la réalisation de tâches motrices) de patients AVC, qui a permis de mieux préciser le décours temporel de la réorganisation cérébrale associée à la récupération motrice et des mesures obtenues en IRMf sont proposées comme facteurs pronostics. L'IRMf au repos est de plus en plus utilisée pour étudier les changements de connectivité cérébrale induits par la lésion ou par une intervention donnée. Les techniques de stimulation cérébrale non invasives (SCNI) dont font parties la Stimulation Magnétique Transcrânienne (TMS) et la Paired Associative Stimulation (PAS), en plus d'être utilisées comme outil de rééducation, sont également utilisées comme outils d'évaluation pour étudier la récupération et l'efficacité de thérapeutiques. De plus, depuis quelques années, de plus en plus d'auteurs recommandent l'utilisation de mesures cinématiques pour une évaluation plus objective et automatique des patients après un AVC. Cette évaluation cinématique qui peut être réalisée grâce-à des robots de rééducation permettrait également de différencier les stratégies mises en place lors de la récupération post-AVC. Afin que les changements mesurés par TMS ou par les outils robotiques puissent correctement révéler cette efficacité, il faut que les paramètres obtenus soient reproductibles, valides et fiables. Ainsi, l'objectif de ma thèse est d'étudier l'efficacité de stratégies thérapeutiques sur la plasticité cérébrale et dans la récupération motrice du membre supérieur après un AVC en parallèle de trouver les mesures adaptées pour évaluer au mieux ces effets. Au travers de données obtenues dans le cadre de six protocoles cliniques impliquant des sujets sains ou des patients AVC, il a été possible de mettre en évidence le potentiel d'un outil de mesure et/ou d'une technique interventionnelle à mesurer voire à stimuler les mécanismes à l'origine de la récupération motrice dans l'AVC. / Stroke is the leading cause of death in women and the 3rd in men, as well as the leading cause of dependency in surviving people because they have major neurological sequelae. The deficiencies are mainly functional and only 5 to 25% of patients will recover a functional upper limb. Spontaneous functional recovery, which is maximal within 3 months after the onset of stroke, is mainly underpinned by brain plasticity mechanisms. Mechanisms that can be boosted by different rehabilitation techniques, including pharmacological substances or brain stimulation techniques that are developed for this purpose. Quantifying the capacities of each patient is essential to determine the lost functions and thus propose appropriate rehabilitation protocols. Many techniques have demonstrated their evaluation potential. In particular, functional magnetic resonance imaging (fMRI) of stroke patients performing a motor task, has made it possible to better define the temporal course of brain reorganization associated with motor recovery. In fact, fMRI measures are proposed as prognostic factors. Resting-state fMRI is increasingly being used to study changes in brain connectivity induced by the lesion or by a given therapeutic intervention. Non-invasive brain stimulation (NIBS) techniques, including Transcranial Magnetic Stimulation (TMS) and Paired Associative Stimulation (PAS), in addition to being used as a rehabilitation tool, are also used as assessment tools to study the recovery and efficacy of therapies. In addition, in recent years, more and more authors have recommended the use of kinematic measurements for more objective and automatic patient assessment after stroke. This kinematic assessment, which can be carried out using rehabilitation robots, would also make it possible to differentiate the strategies implemented during post-stroke recovery. In order for the changes measured by TMS or robotic tools to correctly reveal this effectiveness, the parameters obtained must be reproducible, valid and reliable. Thus, the objective of my thesis is to study the effectiveness of therapeutic strategies on brain plasticity and in the motor recovery of the upper limb after a stroke in parallel to find the appropriate measures to best evaluate these effects. Through data obtained from six clinical protocols involving healthy subjects or stroke patients, it was possible to highlight the potential of a measurement tool and/or intervention technique to measure or even stimulate the mechanisms underlying motor recovery in stroke.

Les machines parallèles (MPs) existent depuis plus d'un demi-siècle et ils ont fait l'objet d'études intensives. Par opposition avec leurs homologues de structure série, ces mécanismes sont constitués de plusieurs chaînes cinématiques qui relient la base fixe à la plateforme mobile. L'intérêt de ces architectures s'explique par les nombreux avantages qu'elles offrent, parmi lesquels: une rigidité élevée, un rapport important charge/poids global, des capacités dynamiques élevées en raison des masses en mouvement réduites (en particulier lorsque les actionneurs sont sur ou près de la base), une meilleure précision, des fréquences propres plus élevées, etc. Néanmoins, leur exploitation comme machines-outils reste timide et limitée, et le plus souvent elles ne dépassent pas le stade d'étude et de prototype de laboratoires universitaires ou de fabricants de machines-outils. Les principaux inconvénients qui entravent la généralisation des MPs dans l'industrie sont les suivants: un espace de travail limité, des débattements angulaires réduits, la présence de configurations singulières, la complexité de conception, les difficultés d'étalonnage, les problèmes causés par les collisions, la complexité du contrôle/commande (en particulier dans le cas de redondance à actionnement), etc. De plus, si les MPs ont rencontré un grand succès dans les applications de pick-and-place grâce à leur rapidité (capacité d'accélération), leur précision reste inférieure à ce qui a été prévu initialement. Par ailleurs, on trouve également des MPs de très précision, mais malheureusement avec de faibles performances dynamiques. En partant du constat précédant, cette thèse se concentre sur l'obtention de MPs avec un bon compromis entre rapidité et précision. Nous commençons par donner un aperçu de la bibliographie disponible concernant MPs et les avancées majeures dans ce domaine, tout en soulignant les limites de performance des MPs, ainsi que les limites des outils de conception classique. En outre, nous insistons sur les outils d'évaluation des performances, et montrons leurs limites dès qu'il s'agit de traiter le cas de la redondance ou l'hétérogénéité des degrés de liberté (ddl). En effet, si la synthèse architecturale est un point dur de la conception de MPs, la synthèse dimensionnelle reposant sur des indices de performances réellement significatifs l'est également. Par conséquent, de nouveaux indices de performance sont proposés pour évaluer la précision, les capacités cinétostatiques et dynamiques des manipulateurs de manière générale qui apportent des solutions aux difficultés évoquées ci-dessus. Par la suite, plusieurs nouvelles architectures 3T-2R et 3T-1R (T: signifie ddl en translation et R signifie un ddl de rotation) sont présentées, à savoir MachLin5, ARROW V1, et ARROW V2 et ses versions dérivées ARROW V2 M1 et M2. En outre, la synthèse dimensionnelle d'ARROW V2 M2 est réalisée, et les performances de la machine sont évaluées. Finalement, des améliorations futures concernant la précision sont proposées au regard de premiers résultats obtenus sur le prototype. / Parallel manipulators (PMs) have been there for more than half a century and they have been subject of intensive research. In comparison with their serial counterparts, PMs consist of several kinematic chains that connect the fixed base to the moving platform. The interest in such architectures is due to the several advantages they offer, among which we mention: high rigidity and payload-to-weight ratio, elevated dynamical capabilities due to reduced moving masses (especially when the actuators are at or near the base), better precision, higher proper frequencies, etc. Nevertheless, despite of the aforementioned merits, their exploitation as machine tools is still timid and limited, in which they most often do not exceed the research and prototyping stages at university laboratories and machine tool manufacturers. The main drawbacks that hinder the widespread of parallel kinematic machines (PKMs) are the following: limited operational workspace and tilting capacity, presence of singular configurations, design complexities, calibration difficulties, collision-related problems, sophistication of control (especially in the case of actuation redundancy), etc. Besides, though PMs have met a great success in pick-and-place applications, thanks to their rapidity (acceleration capacity), still their precision is less than what has been initially anticipated. On the other hand, extremely precise PMs exist, but unfortunately with poor dynamic performance. Starting from the aforementioned problematics, the current thesis focuses on obtaining PKMs with a good compromise between rapidity and precision. We begin by providing a survey of the available literature regarding PKMs and the major advancements in this field, while emphasizing the shortcomings on the level of design as well as performance. Moreover, an overview on the state of the art regarding performance evaluation is presented and the inadequacies of classical measures, when dealing with redundancy and heterogeneity predicaments, are highlighted. In fact, if finding the proper architectures is one of the prominent issues hindering PKMs' widespread, the performance evaluation and the criteria upon which these PKMs are dimensionally synthesized are of an equal importance. Therefore, novel performance indices are proposed to assess precision, kinetostatic and dynamic capabilities of general manipulators, while overcoming the aforementioned dilemmas. Subsequently, several novel architectures with 3T-2R and 3T-1R degrees of freedom (T and R signify translational and rotational degrees of freedom), namely MachLin5, ARROW V1, and ARROW V2 with its mutated versions ARROW V2 M1/M2, are presented. Furthermore, the dimensional synthesis of the executed PKM, namely ARROW V2 M2, is discussed with its preliminary performances and possible future enhancements, particularly regarding precision amelioration.

This article describes the development of a component-based technology robot system that can be rapidly configured to perform a specific manufacturing task. The system is conceived with standard and inter-operable components including actuator modules, rigid link connectors and tools that can be assembled into robots with arbitrary geometry and degrees of freedom. The reconfigurable "plug-and-play" robot kinematic and dynamic modeling algorithms are developed. These algorithms are the basis for the control and simulation of reconfigurable robots. The concept of robot configuration optimization is introduced for the effective use of the rapidly reconfigurable robots. Control and communications of the workcell components are facilitated by a workcell-wide TCP/IP network and device level CAN-bus networks. An object-oriented simulation and visualization software for the reconfigurable robot is developed based on Windows NT. Prototypes of the robot systems configured to perform 3D contour following task and the positioning task are constructed and demonstrated. Applications of such systems for biomedical tissue scaffold fabrication are considered. / Singapore-MIT Alliance (SMA)

This thesis addresses the kinematic analysis of mechanisms, in particular, the position analysis of kinematic chains, or linkages, that is, mechanisms with rigid bodies (links) interconnected by kinematic pairs (joints). This problem, of completely geometrical nature, consists in finding the feasible assembly modes that a kinematic chain can adopt. An assembly mode is a possible relative transformation between the links of a kinematic chain. When an assignment of positions and orientations is made for all links with respect to a given reference frame, an assembly mode is called a configuration. The methods reported in the literature for solving the position analysis of kinematic chains can be classified as graphical, analytical, or numerical. The graphical approaches are mostly geometrical and designed to solve particular problems. The analytical and numerical methods deal, in general, with kinematic chains of any topology and translate the original geometric problem into a system of kinematic analysis of all the Assur kinematic chains resulting from replacing some of its revolute joints by slider joints. Thus, it is concluded that the polynomials of all fully-parallel planar robots can be derived directly from that of the widely known 3-RPR robot. In addition to these results, this thesis also presents an efficient procedure, based on distance and oriented area constraints, and geometrical arguments, to trace coupler curves of pin-jointed Gr¨ubler kinematic chains. All these techniques and results together are contributions to theoretical kinematics of mechanisms, robot kinematics, and distance plane geometry. equations that defines the location of each link based, mainly, on independent loop equations. In the analytical approaches, the system of kinematic equations is reduced to a polynomial, known as the characteristic polynomial of the linkage, using different elimination methods —e.g., Gr¨obner bases or resultant techniques. In the numerical approaches, the system of kinematic equations is solved using, for instance, polynomial continuation or interval-based procedures. In any case, the use of independent loop equations to solve the position analysis of kinematic chains, almost a standard in kinematics of mechanisms, has seldom been questioned despite the resulting system of kinematic equations becomes quite involved even for simple linkages. Moreover, stating the position analysis of kinematic chains directly in terms of poses, with or without using independent loop equations, introduces two major disadvantages: arbitrary reference frames has to be included, and all formulas involve translations and rotations simultaneously. This thesis departs from this standard approach by, instead of directly computing Cartesian locations, expressing the original position problem as a system of distance-based constraints that are then solved using analytical and numerical procedures adapted to their particularities. In favor of developing the basics and theory of the proposed approach, this thesis focuses on the study of the most fundamental planar kinematic chains, namely, Baranov trusses, Assur kinematic chains, and pin-jointed Gr¨ubler kinematic chains. The results obtained have shown that the novel developed techniques are promising tools for the position analysis of kinematic chains and related problems. For example, using these techniques, the characteristic polynomials of most of the cataloged Baranov trusses can be obtained without relying on variable eliminations or trigonometric substitutions and using no other tools than elementary algebra. An outcome in clear contrast with the complex variable eliminations require when independent loop equations are used to tackle the problem. The impact of the above result is actually greater because it is shown that the characteristic polynomial of a Baranov truss, derived using the proposed distance-based techniques, contains all the necessary and sufficient information for solving the position / Esta tesis aborda el problema de análisis de posición de cadenas cinemáticas, mecanismos con cuerpos rígidos (enlaces) interconectados por pares cinemáticos (articulaciones). Este problema, de naturaleza geométrica, consiste en encontrar los modos de ensamblaje factibles que una cadena cinemática puede adoptar. Un modo de ensamblaje es una transformación relativa posible entre los enlaces de una cadena cinemática. Los métodos reportados en la literatura para la solución del análisis de posición de cadenas cinemáticas se pueden clasificar como gráficos, analíticos o numéricos. Los enfoques gráficos son geométricos y se diseñan para resolver problemas particulares. Los métodos analíticos y numéricos tratan con cadenas cinemáticas de cualquier topología y traducen el problema geométrico original en un sistema de ecuaciones cinemáticas que define la ubicación de cada enlace, basado generalmente en ecuaciones de bucle independientes. En los enfoques analíticos, el sistema de ecuaciones cinemáticas se reduce a un polinomio, conocido como el polinomio característico de la cadena cinemática, utilizando diferentes métodos de eliminación. En los métodos numéricos, el sistema se resuelve utilizando, por ejemplo, la continuación polinomial o procedimientos basados en intervalos. En cualquier caso, el uso de ecuaciones de bucle independientes, un estándar en cinemática de mecanismos, rara vez ha sido cuestionado a pesar de que el sistema resultante de ecuaciones es bastante complicado, incluso para cadenas simples. Por otra parte, establecer el análisis de la posición de cadenas cinemáticas directamente en términos de poses, con o sin el uso de ecuaciones de bucle independientes, presenta dos inconvenientes: sistemas de referencia arbitrarios deben ser introducidos, y todas las fórmulas implican traslaciones y rotaciones de forma simultánea. Esta tesis se aparta de este enfoque estándar expresando el problema de posición original como un sistema de restricciones basadas en distancias, en lugar de directamente calcular posiciones cartesianas. Estas restricciones son posteriormente resueltas con procedimientos analíticos y numéricos adaptados a sus particularidades. Con el propósito de desarrollar los conceptos básicos y la teoría del enfoque propuesto, esta tesis se centra en el estudio de las cadenas cinemáticas planas más fundamentales, a saber, estructuras de Baranov, cadenas cinemáticas de Assur, y cadenas cinemáticas de Grübler. Los resultados obtenidos han demostrado que las técnicas desarrolladas son herramientas prometedoras para el análisis de posición de cadenas cinemáticas y problemas relacionados. Por ejemplo, usando dichas técnicas, los polinomios característicos de la mayoría de las estructuras de Baranov catalogadas se puede obtener sin realizar eliminaciones de variables o sustituciones trigonométricas, y utilizando solo álgebra elemental. Un resultado en claro contraste con las complejas eliminaciones de variables que se requieren cuando se utilizan ecuaciones de bucle independientes. El impacto del resultado anterior es mayor porque se demuestra que el polinomio característico de una estructura de Baranov, derivado con las técnicas propuestas, contiene toda la información necesaria y suficiente para resolver el análisis de posición de las cadenas cinemáticas de Assur que resultan de la sustitución de algunas de sus articulaciones de revolución por articulaciones prismáticas. De esta forma, se concluye que los polinomios de todos los robots planares totalmente paralelos se pueden derivar directamente del polinomio característico del conocido robot 3-RPR. Adicionalmente, se presenta un procedimiento eficaz, basado en restricciones de distancias y áreas orientadas, y argumentos geométricos, para trazar curvas de acoplador de cadenas cinemáticas de Grübler. En conjunto, todas estas técnicas y resultados constituyen contribuciones a la cinemática teórica de mecanismos, la cinemática de robots, y la geometría plana de distancias. Barcelona 13-

People use both their hands together cooperatively in many everyday activities. The modern computer interface fails to take advantage of this basic human ability, with the exception of the keyboard. However, the keyboard is limited in that it does not afford continuous spatial input. The computer mouse is perfectly suited for the point and click tasks that are the major method of manipulation within graphical user interfaces, but standard computers have a single mouse. A single mouse does not afford spatial coordination between the two hands within the graphical user interface. Although the advent of the Universal Serial Bus has made it possible to easily plug in many peripheral devices, including a second mouse, modern operating systems work on the assumption of a single spatial input stream. Thus, if a second mouse is plugged into a Macintosh computer, a Windows computer or a UNIX computer, the two mice control the same cursor. <br /><br /> Previous work in two-handed or bimanual interaction techniques has often followed the asymmetric interaction guidelines set out by Yves Guiard's Kinematic Chain Model. In asymmetric interaction, the hands are assigned different tasks, based on hand dominance. I show that there is an interesting class of desktop user interface tasks which can be classified as symmetric. A symmetric task is one in which the two hands contribute equally to the completion of a unified task. I show that dual-mouse symmetric interaction techniques outperform traditional single-mouse techniques as well as dual-mouse asymmetric techniques for these symmetric tasks. I also show that users prefer the symmetric interaction techniques for these naturally symmetric tasks.

This report presents optimal control methods integrated with hierarchical control framework to realize real-time collision-free optimal trajectories for motion control in kinematic chain manipulator (KCM) robot systems under dynamic environments. Recently, they have been increasingly used in applications where manipulators are required to interact with random objects and humans. As a result, more complex trajectory planning schemes are required. The main objective of this research is to develop new motion control strategies that can enable such robots to operate efficiently and optimally in such unknown and dynamic environments. Two direct optimal control methods: The direct collocation method and discrete mechanics for optimal control methods are investigated for solving the related constrained optimal control problem and the results are compared. Using the receding horizon control structure, open-loop sub-optimal trajectories are generated as real-time input to the controller as opposed to the predefined trajectory over the entire time duration. This, in essence, captures the dynamic nature of the obstacles. The closed-loop position controller is then engaged to span the robot end-effector along this desired optimal path by computing appropriate torque commands for the joint actuators. Employing a two-degree of freedom technique, collision-free trajectories and robot environment information are transmitted in real-time by the aid of a bidirectional connectionless datagram transfer. A hierarchical network control platform is designed to condition triggering of precedent activities between a dedicated machine computing the optimal trajectory and the real-time computer running a low-level controller. Experimental results on a 2-link planar robot are presented to validate the main ideas. Real-time implementation of collision-free workspace trajectory control is achieved for cases where obstacles are arbitrarily changing in the robot workspace.

People use both their hands together cooperatively in many everyday activities. The modern computer interface fails to take advantage of this basic human ability, with the exception of the keyboard. However, the keyboard is limited in that it does not afford continuous spatial input. The computer mouse is perfectly suited for the point and click tasks that are the major method of manipulation within graphical user interfaces, but standard computers have a single mouse. A single mouse does not afford spatial coordination between the two hands within the graphical user interface. Although the advent of the Universal Serial Bus has made it possible to easily plug in many peripheral devices, including a second mouse, modern operating systems work on the assumption of a single spatial input stream. Thus, if a second mouse is plugged into a Macintosh computer, a Windows computer or a UNIX computer, the two mice control the same cursor. <br /><br /> Previous work in two-handed or bimanual interaction techniques has often followed the asymmetric interaction guidelines set out by Yves Guiard's Kinematic Chain Model. In asymmetric interaction, the hands are assigned different tasks, based on hand dominance. I show that there is an interesting class of desktop user interface tasks which can be classified as symmetric. A symmetric task is one in which the two hands contribute equally to the completion of a unified task. I show that dual-mouse symmetric interaction techniques outperform traditional single-mouse techniques as well as dual-mouse asymmetric techniques for these symmetric tasks. I also show that users prefer the symmetric interaction techniques for these naturally symmetric tasks.

In this thesis, the measurement uncertainty of Coordinate Measuring Machine (CMM) is analysed and software is designed to simulate this. Analysis begins with the inspection of the measurement process and structure of the CMMs. After that, error sources are defined with respect to their effects on the measurement and then an error model is constructed to compensate these effects. In other words, systematic part of geometric, kinematic and thermal errors are compensated with error modelling. Kinematic and geometric error model is specific for the structure of CMM under inspection. Also, a common orthogonal kinematic model is formed and with using the laser error data of the CMM and error maps of the machine volume is obtained. Afterwards, the models are compared with each other by taking the difference and ratio. The definition and compensation of the systematic errors leave the uncertainty of measurements for analysing. Measurement uncertainty consists of the uncompensated systematic errors and random errors. The other aim of the thesis is to quantify these uncertainties with using the different methods and to inspect the success of these methods. Uncertainty budgeting, comparison, statistical evaluation by designing an experiments and simulation methods are examined and applied to the CMM under inspection. In addition, Virtual CMM software is designed to simulate the task specific measurement uncertainty of circle, sphere and plane without using the repeated measurements. Finally, the performance of the software, highly depending on the mathematical modelling of machine volume, is tested by using actual measurements.

This thesis comprises a study on the kinematic and force analyses of the overconstrained mechanisms. The scope of the overconstrained mechanisms is too wide and difficult to handle. Therefore, the study is restricted to the planar overconstrained mechanisms. Although the study involves only the planar overconstrained mechanisms, the investigated methods and approaches could be extended to the spatial overconstrained mechanisms as well. In this thesis, kinematic analysis is performed in order to investigate how an overconstrained mechanism can be constructed. Four methods are used. These are the analytical method, the method of cognates, the method of combining identical modules and the method of extending an overconstrained mechanism with extra links. This thesis also involves the force analysis of the overconstrained mechanisms. A method is introduced in order to eliminate the force indeterminacy encountered in the overconstrained mechanisms. The results are design based and directly associated with the assembly phase of the mechanism.