Motion planning is an increasingly important field of research. Factory automation is becoming more prevalent and at the same time, production runs are shortening in the name of customisation. With computer controlled equipment becoming cheaper and more modular, setting up near-fully automated production lines is becoming fast and easy. This means that the actual programming of the robots and assembly system is becoming the rate determining step. Automated motion planning is a possible solution to this—but only if it can run fast enough. Many heuristic planners have been created in an attempt to achieve the necessary speeds in off-line (or more ambitiously, on-line) processing. This thesis aims to show that different types of heuristic planners can be designed to take advantage of specialised environments or robot characteristics. To show this, three distinct classes of heuristic planners are put forward for discussion. The first of these classes, addressed in Chapter 2, is of very generic planners which will work in virtually all situations (ie. almost any combination of robot and environment). This generality is obviously useful when lacking more specific domain knowledge. However these methods do suffer performance-wise in comparison with more specialised planners when there are characteristics of the problem which can be targeted. Chapter 3 moves to planners which are designed to specifically address certain peculiarities of the environment. Particular focus is given to environments whose corresponding configuration-spaces contain narrow gaps and passages. Finally Chapter 4 addresses a third class of planners: those which are designed for specific types of robots and movements. The particular focus is on locomotion for legged vehicles. For each of these three classes, some discussion is made of existing planners which can be so characterised. In addition, a novel algorithm is introduced in each as an example for particular consideration.

Η παρούσα διπλωματική εργασία ασχολείται με τον συνεργατικό έλεγχο μιας ομάδας αυτόνομων ρομπότ. Θεωρώντας ότι μόνο ένα ρομπότ έχει πλήρη γνώση για το μονοπάτι που επιθυμούμε να ακολουθήσουν τα ρομπότ, θα πρέπει τα υπόλοιπα παίρνοντας μετρήσεις από διάφορους αισθητήρες και ανταλλάσοντας πληροφορίες μεταξύ τους, να σχεδιάζουν και να ακολουθούν τέτοιες τροχιές ώστε να πετύχουν τον στόχο τους. Από τη στιγμή που ένα μοναδικό ρομπότ γνωρίζει το επιθυμητό μονοπάτι, πρόκειται για ένα leader-follower σχηματισμό όπου κάθε ρομπότ καλείται να ακολουθήσει αυτό που προηγείται. Προκειμένου τα ρομπότ να μπορούν να ακολουθήσουν μία τροχιά αναπτύχθηκε σε περιβάλλον LabVIEW ένας ελεγκτής παρακολούθησης τροχιάς. Για την ανταλλαγή πληροφοριών και μηνυμάτων συντονισμού μεταξύ των ρομπότ, αναπτύχθηκε ασύρματο δίκτυο ZigBee. Το πρόβλημα γνώσης σχετικά με το που βρίσκεται ο leader κάθε ρομπότ λύθηκε με τη χρήση αλγορίθμων που επεξεργάζονται τις μετρήσεις των ενσωματωμένων Sonar. Επίσης το πρόβλημα γνώσης του πραγματικού προσανατολισμού κάθε ρομπότ, αντιμετωπίστηκε με την κατασκευή ενός ψηφιακού μαγνητομέτρου. Για κάθε ρομπότ ξεχωριστά, αναπτύχθηκε ο συνεργατικός αλγόριθμος ο οποίος εξασφαλίζει ότι η ομάδα θα πετυχαίνει τον στόχο που έχει αρχικά τεθεί. Τέλος παρουσιάζονται πειραματικά αποτελέσματα για ομάδες δύο και τριών ρομπότ. / The present thesis elaborates on the cooperative control of mobile robots. Assuming that one robot has complete knowledge of the desired path the robotic platoon must follow, a coordination control scheme must be created, based on sensor measurements and platoon communication, that generates desired trajectories for each of the members of the platoon. From the moment a single robot has complete knowledge of the path, the coordination scheme is based on leader-follower formation control. A trajectory tracking controller was developed in LabVIEW, while a ZigBee based wireless network was implemented for the platoon communication. To find the relative position of the leader for each robot, a sonar based localization algorithm was created, with position measurements through the robot’s encoders and orientation given from a magnetometer. For each robot seperately, the coordination algorithm was developed, that ensures that the platoon will achieve the original goal.

The dynamic performance of a programmable manipulator depends on both the motion profile to be followed and the feedback control method used. To improve this performance the manipulator trajectory requires planning at an advanced level and an efficient control method has to be used. The purpose of this study is to investigate high-level trajectory planning and trajectory tracing problems. It is shown that conventional trajectory planning methods where the motion curves are generated using standard mathematical functions are ineffective for general application especially when velocity and acceleration conditions are included. Polynomial functions are shown to be the most versatile for these applications but these can give curves with unexpected oscillations, commonly called meandering. In this study, a new method using polynomials is developed to overcome this disadvantage. A general motion design computer program (MOTDES) is developed which enables the user to produce motion curves for general body motion in a plane. The program is fully interactive and operates within a graphics environment. A planar manipulator is designed and 'constructed to investigate the practical problems of trajectory control particularly when operating at high speeds. Different trajectories are planned using MOTDES and implemented to the manipulator. The precise tracing of a trajectory requires the use of advanced control methods such as adaptive control or learning. In learning control, the inputs of the current cycle are calculated using the experience of the previous circle. The main advantage of learning control over adaptive control is its simplicity. It can be applied more easily in real time for high-speed systems. However, learning algorithms may cause saturation of the driving servo motors after a few learning cycles due to discontinuities being introduced into the command curve. To prevent this saturation problem a new approach involving the filtering of the input command is developed and tested.

L’assistance à la mobilité est un enjeu majeur, compte tenu de son importance dans l’augmentation de l’autonomie des personnes. Dans le cadre de cette thèse, nous nous intéressons au moyen d’assistance le plus commun qu’est la canne. Nous avons, dans un premier temps, analysé l’apport des cannes conventionnelles dans l’assistance à des marches perturbées.Cette analyse repose sur des caractérisations expérimentales ainsi que sur la modéli- sation et la simulation de la marche assistée. Nous avons ainsi mis en évidence l’aide au support du poids, au freinage et à la propulsion permises par le point d’appui supplé- mentaire fourni par la canne.Nous avons ensuite introduit le développement d’une canne robotique. Contrairement aux cannes robotiques existantes, notre prototype ne repose pas sur une base mobile statiquement stable. Pour plus de compacité et pour garder les attributs d’une canne conventionnelle, elle est composée d’un axe télescopique et d’une roue à son extrémité, tous deux motorisés. La commande de ce prototype est ensuite décrite. Elle permet de synchroniser les mouvements de la canne robotique avec le cycle de la marche. La canne suit ainsi activement le mouvement de la jambe “invalide” durant la phase de balancement et offre un point d’appui stable pendant la phase d’appui. / Mobility assistance is major challenge since its importance in people autonomy enhancement.In this thesis, we focus on one of the most used assisting device which is the cane.Firstly, we have analyzed the supply provided by conventional canes during impaired gaits.This analyse is based on cane-assisted gait experimental characterizations and simulation.We have highlighted the weight-bearing, braking and propulsion assistance allowed by the additional contact point represented by the cane.Then, the development of a robotized cane is introduced. Unlike existing robotized canes, our prototype does not remain on a stable mobile platform.In order to reduce its volume and keep the shape of a conventional cane, the cane is composed of a telescopic shaft and a wheel at its tip, both motorized.The control of this cane is described. It allows to synchronize the cane motion with its user’s gait. The active cane follows the weakest leg during its swinging phase and offers a stable contact point during the stance phase.

This dissertation presents the results of a series of studies on dynamic stabilisation of CoMan, which is actuated by series elastic actuators. The main goal of this dissertation is to dynamically stabilise the humanoid robot on the floor by the simplest multivariate feedback control for the purpose of walking. The multivariable scheme is chosen to take into account the joints' interactions, as well as providing a systematic way of designing the feedback system to improve the bandwidth and tracking performance of CoMan's existing PID control. A detailed model is derived which includes all the motors and joints state variables and their multibody interactions which are often ignored in the previous studies on bipedal robots in the literature. The derived dynamic model is then used to design multivariable optimal control feedback and observers with a mathematical proof for the relative stability and robustness of the closed loop system in face of model uncertainties and disturbances. In addition, two decentralized optimal feedback design algorithms are presented that explicitly take the compliant dynamics and the multibody interactions into account while providing the mathematical proof for the stability of the overall system. The purpose of the proposed decentralized control methods is to provide a systematic model based PDPID design to replace the existing PID controllers which are derived by a trial and error process. Moreover, the challenging constrained and compliant motion of the robot in double support is studied where a novel constrained feedback design is proposed which directly takes the compliance dynamics, interactions and the constraints into account to provide a closed loop feedback tracking system that drives the robot inside the constrained subspace. This method of control is particularly interesting since most control methods applied to closed kinematic chains (such as the double support phase) are over complicated for implementation purposes or have an ad-hoc approach to controller design. In terms of walking trajectory generation, an extension to the ZMP walking trajectory generation is proposed to utilise the CoMan's upper body to tackle the non-minimum phase behaviour that is faced in trajectory generation. Simple inverted pendulum models of walking are then used to study the maximum feasible walking speed and step size where parameters of CoMan are used to provide numerical upperbounds on the step size and walking speed. Use of straight knee and toe push-off during walking is shown to be beneficial for taking larger step lengths and hence achieving faster walking speeds. Subsequently, the designed tracking systems are then applied to a dynamic walking simulator which is developed during this PhD project to accurately model the compliant walking behaviour of the CoMan. A walking gait is simulated and visualized to show the effectiveness of the developed walking simulator. Moreover, the experimental results and challenges faced during the implementation of the designed tracking control systems are discussed where it is shown that the LQR feedback results in 50% less control effort and tracking errors in comparison with CoMan's existing independent PID control. This advantage directly affects the feasible walking speed. In addition, a set of standard and repeatable tests for CoMan are designed to quantify and compare the performance of various control system designs. Finally, the conclusions and future directions are pointed out.

The field of swarm robotics has been growing fast over the last few years. Using a swarm of simple and cheap robots has advantages in various tasks. Apart from performance gains on tasks that allow for parallel execution, simple robots can also be smaller, enabling them to reach areas that can not be accessed by a larger, more complex robot. Their ability to cooperate means they can execute complex tasks while offering self-organised adaptation to changing environments and robustness due to redundancy. In order to keep individual robots simple, a control algorithm has to keep expensive communication to a minimum and has to be able to act on little information to keep the amount of sensors down. The number of sensors and actuators can be reduced even more when necessary capabilities are spread out over different agents that then combine them by cooperating. Self-organised differentiation within these heterogeneous groups has to take the individual abilities of agents into account to improve group performance. In this thesis it is shown that a homogeneous group of versatile agents can not be easily replaced by a heterogeneous group, by separating the abilities of the versatile agents into several specialists. It is shown that no composition of those specialists produces the same outcome as a homogeneous group on a clustering task. In the second part of this work, an adaptation mechanism for a group of foragers introduced by Labella et al. (2004) is analysed in more detail. It does not require communication and needs only the information on individual success or failure. The algorithm leads to self-organised regulation of group activity depending on object availability in the environment by adjusting resting times in a base. A possible variation of this algorithm is introduced which replaces the probabilistic mechanism with which agents determine to leave the base. It is demonstrated that a direct calculation of the resting times does not lead to differences in terms of differentiation and speed of adaptation. After investigating effects of different parameters on the system, it is shown that there is no efficiency increase in static environments with constant object density when using a homogeneous group of agents. Efficiency gains can nevertheless be achieved in dynamic environments. The algorithm was also reported to lead to higher activity of agents which have higher performance. It is shown that this leads to efficiency gains in heterogeneous groups in static and dynamic environments.

Ce travail concerne une catégorie de robot mobile manipulateur omnidirectionnel bionique, en l’occurence le RobotinoXT qui dispose d’un manipulateur bionique CBHA monté sur un robot mobile omnidirectionnel, nommée Robotino. D’abord, nous avons proposé un modèle cinématique direct de ce système en utilisant une méthode nommée « arc geometry ». Celle-ci a été validée grâce à un banc d’expérimentation en ayant recours à une technique de trilatération. Pour atteindre cet objectif, les capteurs ont été calibrés. Concernant le problème de calibration des paramètres constants du modèle, il a été résolu en développant un algorithme d’optimisation incluant une méthode nommée SQP, validée via un manipulateur industriel à bras rigides. Puis, nous avons proposé un modèle cinématique inverse du CBHA en supposant que chaque vertèbre de celui-ci est assimilée à un robot parallèle. Ce modèle a été validé par les mesures réelles obtenues avec le manipulateur industriel, permettant la définition de l’espace de travail du robot. Enfin, nous avons implémenté une boîte à outils qui englobe les modèles développés dans ce travail. / This work deals with a particular class of mobile omnidrive-bionic manipulator robot namely RobotinoXT. It contains a bionic manipulator Compact Bionic Handling Assistant (CBHA) mounted over a mobile omnidrive robot Robotino. We first proposed a forward kinematic model of such a system by using an “arc geometry” method which was validated through a test bench using a trilateration technique. To achieve this purpose, the sensors were calibrated. For the model’s constant parameters calibration problem, this latter was resolved by developing an optimization algorithm which incorporates a SQP method, validated via an industrial manipulator with rigid links. Then, we proposed an inverse kinematic model of the CBHA by assuming that each backbone of it, is assimilated to a parallel robot. This model was validated by real measurements obtained with the industrial manipulator, allowing defining the robot workspace. Finally, we implemented a toolbox which encompasses the models developed in this work.

La génération de mouvements de locomotion en robotique mobile est étudiée dans le monde académique depuis plusieurs décennies. La théorie concernant la modélisation et le contrôle des robots à roues est largement mature. Cependant, la mise en œuvre effective de ces modèles dans des conditions réelles demande des études complémentaires. Dans cette thèse, nous présentons trois projets mettant en œuvre trois différents types de robots mobiles. Nous débutons dans chaque cas par une analyse sur les qualités recherchées d’un mouvement dans un contexte particulier, qu’il soit artistique ou industriel, et terminons par la présentation des architectures algorithmiques et logicielles mises en œuvre, notamment dans le cadre d’expositions de plusieurs mois, où le public est invité à partager l’espace d’évolution de robots. La réalisation de ces projets montre que certains choix technologiques semblant insignifiants au moment de la conception des robots sont déterminants dans les dernières étapes de la production. On peut extrapoler cette remarque depuis ces robots mobile à deux ou trois degrés de liberté vers des robots humanoïdes pouvant en avoir plusieurs dizaines. La stratégie classique qui consiste à concevoir, dans un premier temps, l’architecture mécatronique des robots humanoïdes, pour se poser ensuite la question de leur contrôle, atteint ses limites, comme le montrent par exemple la consommation énergétique et la difficulté d’obtenir des mouvements de marche dynamique sur ces robots, pourtant conçus dans le but de marcher. Dans une perspective globale de conception des robots marcheurs, nous proposons un système de codesign, où il est possible d’optimiser simultanément la conception mécanique et les contrôleurs d’un robot.. / Generation of locomotion motions in mobile robotics has been studied in the academic world for several decades. The theory concerning the modeling and control of wheeled robots is largely mature. However, the actual implementation of these models in real conditions requires further studies. In this thesis, we present three projects using three different types of mobile robots. In each case, we begin with an analysis of the required qualities of a motion in a particular context, whether artistic or industrial, and end with the presentation of the algorithmic and software architectures implemented, particularly in the context of exhibitions of several months, where the public is invited to share the space of evolution of robots. The realization of these projects shows that some technological choices seem insignificant at the time of the design of the robots are decisive in the final stages of production. One can extrapolate this remark from these mobile robots with two or three degrees of freedom towards humanoid robots which can have several tens. The classical strategy of first designing the mechatronic architecture of humanoid robots and then raising the question of their control has reached its limits, as illustrated, for example, by their energy consumption and the difficulty to obtain dynamic walking motions on these robots, yet designed for the purpose of walking. From a global perspective of robot design, we propose a system of codesign, where it is possible to simultaneously optimize the mechanical design and the controllers of a robot. This system is firstly tested by various examples as proof of concept. It is then applied to the comparison of rigid and elastic actuators on different biped robots, then to the study of the impact of the stabilization of the head on the general stabilization of the body and finally to the design of a prototype of semi-passive walker.

Les fauteuils roulants « intelligents » dotés de facultés de navigation autonomes visent à soulager les personnes handicapés moteurs ayant des difficultés à conduire un fauteuil électrique standard. Depuis les années 80, de nombreuses études ont été menées pour réaliser de tels prototypes mais très peu ont abouti à de réels progrès pour les utilisateurs et ce uniquement pour des fonctionnalités simples. Cela peut s'expliquer par plusieurs facteurs, notamment la sous-estimation des contraintes d'acceptabilité et d'adaptabilité. Dans ce contexte, notre objectif est de concevoir une assistance à la conduite sous respect de ces contraintes. Pour cela, nous réalisons un système léger, composé de trois capteurs laser, d'une caméra et d'un micro-ordinateur. Il est susceptible de s'adapter sur tout type de fauteuil électrique et nous a permis de développer les fonctionnalités de franchissement de passage étroit et de suivi de mur ainsi qu'une interface humain-machine ergonomique. Celle-ci est établie par une commande « déictique » qui consiste à fournir un aperçu de l'environnement sur lequel l'utilisateur désigne la tâche autonome choisie par son lieu d'application, le fauteuil réalisant automatiquement l'action correspondante. Tout contact sur le joystick rend immédiatement le contrôle du fauteuil à l'utilisateur afin qu'il ne se sente pas prisonnier de l'assistance, ainsi le pilotage se compose d'une alternance de commandes manuelles et d'indications sur l'interface. Ensuite, l'évaluation de notre assistance à la conduite s'est déroulée en trois étapes. La première, qualitative, a consisté à présenter et à faire essayer le système à des utilisateurs potentiels. Dans un second temps nous avons comparé quantitativement les performances de la conduite assistée avec celles d'une conduite manuelle sur un panel de sujets valides (temps de parcours, nombre d'actions réalisées, indice de confort...). La troisième étape a consisté à évaluer la charge cognitive des utilisateurs dans les deux modes de conduite en mesurant la charge attentionnelle et la capacité décisionnelle à partir d'une méthode de douche tâche / The Smart Wheelchairs, that are equipped for autonomous navigation functionalities, aim to relieve people with disabilities who have difficulty to drive a standard electric wheelchair. Since the 80s, many studies have been conducted to design such prototypes but very few have led to real progress for users and only for simple functionalities. This can be explained by several factors, including the underestimation of the constraints of acceptability and adaptability. In this context, our goal is to provide a driving assistance in respect of these constraints. For this, we design a lightweight system, consisting of three laser sensors, a camera and a computer. It is adaptable to any type of electric wheelchair and allowed us to develop the functionalities of passing through of the narrow passages and wall following, as well as an ergonomic human-machine interface. This latter is established by a deictic command which consists in to provide an overview of the environment where the user indicates the chosen autonomous task by his application location, and then the wheelchair performs the corresponding action automatically. Moreover, all contact with the joystick gives back the control to the user so that he does not feel a prisoner of assistance; well the driving is composed of alternating manual controls and indications on the interface. Finally, the evaluation of our assistance in the conduct took place in three stages. The first, qualitative, is to present and to do try the system to potential users. In a second step, we compare quantitatively the performance of the assistance driving with those of a manual driving on a panel of valid persons (travel time, number of actions, discomfort index?). The third step is to assess the cognitive load of users in both driving modes by measuring the attentional load and decision-making capacity from a dual task method

La capacité à se localiser est d’une importance cruciale pour la navigation des robots. Cette importance a permis le développement de plusieurs techniques de localisation de grande précision. Notre contribution consiste à proposer un passage de la technique de localisation visuelle inertielle du cas individuel, au cas multi collaboratif. Ce travail a pour objectif d’aboutir à une localisation collaborative aussi rapide, robuste et précise que la technique individuelle de départ. Notre approche se base sur le filtrage en couplage serré Multi State Constraint Kalman Filter (MSCKF) pour la fusion de données. Les caractéristiques de ce filtrage sont d’abord étudiées dans le cas individuel pour tester la robustesse et la précision dans différentes conditions et avec différents modèles d’observation. Les résultats de cette étude nous ont orienté vers la structure la mieux adaptée à une augmentation au cas de localisation collaborative. L’algorithme collaboratif proposé, est basé sur un processus hiérarchique en trois étapes. Une localisation collaborative est initialisée sur la base des mesures relatives de distances Ultra Large Bande (ULB). Une localisation collaborative améliorée se base ensuite sur le chevauchement des images en utilisant un modèle de mesure adapté, et une structure de fusion de données qui absorbe l’excédent en temps de calcul provoqué par le traitement collaboratif. Enfin, pour augmenter la précision, une extraction des contraintes de structure environnement, suivie d’une intégration à l’aide d’une troncature dans le filtre sont proposées. / Localization is of crucial importance for robots navigation. This importance has allowed the emergence of several precise localization techniques. Our contribution consists of proposing a transition from an individual inertial visual localization technique to the multi-robots collaborative localization case. This work aims to achieve a collaborative localization as fast, robust and accurate as the individual starting technique. We adopt a tightly coupled MSCKF (Multi State Constraint Kalman Filter) approach to achieve the data fusion. The characteristics of this data fusion are first studied in the individual case to test the robustness and the precision under different conditions and with different observation models. The results of this study directed us towards the best structure adapted to an augmentation to the collaborative localization case. The proposed collaborative algorithm is a hierarchical process of three stages. A collaborative localization is initialized based on the relative distance measurements using Ultra-Wide Band (ULB) sensors. Then, a collaborative localization based on images overlapping using a suitable measurement model, and a data fusion structure that absorbs the computation time excess caused by the collaboration is achieved. Finally, to increase precision, an extraction of the environment constraints, followed by an integration using a truncation in the filter are proposed.