D.Ing. / The grasping strategy for a three dimensional object by a robotic gripper requires a geometrical reasoning and analysis of the physical gripper design, control and operation. The work addresses the problem of data acquisition and processing required for an object recognition and its application in the selection of grasping strategy for a given gripper. The system described in the thesis integrates the analyses of image data, object geometry and grasping operation in a systematic way. It is hierarchically constructed in several levels of analyses and processes including object recognition, grasping feature representation and classification, matching strategy for objects and the gripper and grasping description and operation. Object shape features are taken for recognition based on the image data collected through an infrared sensor. With a face relation graph proposed, an object model is built for describing the object geometrical properties and extracting its grasping features. A coding system based on group technology concepts is proposed for object classification. It describes object features relative to grasping operation. Gripping models are established and incorporated with the coding system for analysis of object gripping features. By means of the gripping models and the coding system, objects to be grasped are classified and grouped into specific families according to their similarities in gripping. The information transformation between the object and the gripper is made through a matrix representation. An object matrix describes the selection of gripping faces and object geometry for gripping , while a gripper matrix describes the fingers selection and its configuration in correspondence with the object to be grasped. The matching of the matrices is established through a knowledge-based reasoning approach. The grasping operation is controlled by a computer in terms of the commands generated by the gripper matrix through a gripper code. The design of the generic gripper for this application is described.

M.Ing. (Mechanical Engineering) / The research reported in this thesis describes the design and simulation of a neural controller for a three -degree-of-freedom robot leg for use as an hexapod leg . Biological systems are considered as a motivation to develop the neural control system for hexapod walking on a horizontal surface . Backpropagation training of multilayer perceptrons and a combination of heterogeneous neurons are used to implement several pattern generators with different behaviours. The artificial neurons are simulated and connected together with the pattern generators to form a complete control system . Previous work [48] shows the performance of a two -degree-of-freedom leg controller - this type of controller however cannot compensate for surface irregularities , The control system for the three degree-of-freedom leg is then further extended to compensate for surface irregularities that cannot be traversed by the two -degree-of-freedom leg.

One of the main challenges in the field of robotics is to build machines that can function intelligently in unstructured environments. Because of this, the field has witnessed a trend away from the sense-plan-act paradigm where the robot makes an attempt to model everything before planning and acting. Nevertheless, few approaches to robotic grasping and manipulation have been proposed that do not require detailed geometric models of the manipulation environment. One exception is the control-based approach where closed-loop controllers reactively generate grasping and manipulation behavior. This thesis develops and extends the control-based approach to grasping and manipulation and proposes a new framework for learning control-based skills based on generalized solutions. This thesis extends control-based approaches to grasping and manipulation in several ways. First, several new controllers relevant to reaching and grasping are proposed, including a grasp controller that slides contacts over the surface of an object toward good grasp configurations by using haptic feedback. The number of different grasps that can be generated using grasp controllers is expanded through the use of virtual contacts. In addition, a new approach to statically-stable dexterous manipulation is proposed whereby the robot navigates through a space of statically stable grasp configurations by executing closed-loop controllers. In a series of experiments, grasp controllers are shown to be a practical approach to synthesizing different grasps from a variety of different starting configurations. This thesis also proposes a new approach to learning control-based behaviors by applying a generalized solution in new situations. Instead of searching the entire space of all controller sequences and combinations, only variations of a generalized solution, encoded by an action schema, are considered. A new algorithm, known as schema structured learning, is proposed that learns how to apply the generalized solution in different problem contexts through a process of trial and error. This approach is applied to the grasp synthesis problem, enabling a robot to learn grasp skills with relatively little training experience. The algorithm learns to select an appropriate reach-grasp strategy based on coarse visual context. In an experiment where a dexterous humanoid robot grasps a range of grocery items it had no prior experience with, the learned grasp skills are shown to generalize well to new objects and object configurations.

No description available.

Note:

No description available.

No description available.

No description available.

Robots already impact the way we understand our world and live our lives. However, their impact and use is limited by the skills they possess. Currently deployed autonomous robots lack the manipulation skills possessed by humans. To achieve general autonomy and applicability in the real world, robots must possess such skills. Autonomous manipulation requires algorithms that rapidly and reliably compute collision-free motion for robotic limbs with many degrees of freedom. Unfortunately, adequate algorithms for this task do not currently exist. Though there are many dimensions of the real-world planning task that require further research. A central problem of reliable real-world planning is that planners must rely on incomplete and inaccurate information about the world in which they are planning. The motion planning problem has exponential complexity in the robot's degrees of freedom. Consequently, the most successful planning algorithms use incomplete information obtained via sampling a subset of all possible movements. Additionally, real-world robots generally obtain information about the state of their environment through lasers, cameras and other sensors. The information obtained from these sensors contains noise and error. Thus the planner's incomplete information about the world is possibly inaccurate as well. Despite such limited information, a planner must be capable of quickly generating collision free motions to facilitate general purpose autonomous robots. This thesis proposes a new utility-guided framework for motion planning that can reliably compute collision-free motions with the efficiency required for real-world planning. The utility-guided approach begins with the observation there is regularity in space of possible motions available to a robot. Further, certain motions are more crucial than others for computing collision free paths. Together these observations form structure in the robot's space of possible movements. This structure provides a guide for the planner's exploration of possible motions. Because a complete understanding of this structure is computationally intractable, the utility-guided framework incrementally develops an approximate model discovered by past exploration. This model of the structure is used to select explorations that maximally benefit the planner. Information provided by each exploration improves the planner's approximation. The process of incremental improvement and further guided exploration iterates until an adequate model of configuration space is constructed. Discovering and exploiting structure in a robot's configuration space enables a utility-guided planner to achieve the performance and reliability required by real-world motion planning. This thesis describes applications of the utility-guided motion-planning framework to multi-query sampling-based roadmap and random-tree motion planning. Additionally, the utility-guided framework is extended to develop a planner that can successfully plan despite inaccuracies in its perception of the environment and to guide further sensing to reduce uncertainty and maximally improve the utility of the path.

Ce mémoire présente des manœuvres de réorientation appliquées à un robot articulé et à un robot mobile en chute libre. Ces manœuvres, initialement inspirées du phénomène du chat qui atterrit toujours sur ses pattes, sont aussi attribuées à d'autres animaux, tels que certains reptiles et même les humains dans le contexte de certains sports. Les manœuvres de réorientation ont aussi des applications dans le domaine de la robotique. En effet, de tels manœuvres s'avèrent utiles pour le contrôle de pose d'atterrissage pour des robots susceptibles aux chutes, comme les robots sauteurs ou les robots de secours qui doivent être déployés dans des environnements dangereux et difficiles à parcourir. Dans cette optique, le travail présenté dans ce mémoire vise à développer et démontrer des manœuvres de réorientation permettant une réorientation rapide (redressement de 180 degrés -- le pire cas possible -- dans le temps d'une chute d'un mètre) et multiaxe. Tout d'abord, une architecture articulée ainsi que deux manœuvres de réorientation sont conçues afin d'atteindre les capacités de réorientation visées et les performances de cette architecture sont testées en simulation. Les résultats obtenus démontrent que l'architecture proposée est capable de se réorienter selon plusieurs axes, mais n'atteint pas les performances visées en termes de vitesse de réorientation. Par la suite, une architecture mobile omnidirectionnelle et compacte est conçue afin d'adresser les limitations de la première architecture. Un prototype de cette architecture est développé et permet d'effectuer une réorientation de 179 degrés selon son axe de tangage en 0.44 secondes tout en conservant sa capacité de se redresser selon plusieurs axes. Les performances de réorientation visées sont alors atteintes avec ce deuxième prototype. Enfin, une méthode de fusion de données par filtre de Kalman étendu servant à estimer l'orientation d'une plateforme en apesanteur est explicitée et est validée dans des conditions contrôlées. Ces résultats démontrent l'utilité de telles méthodes de fusion de données pour implémenter la planification automatique des manœuvres de réorientation dans les itérations futures du prototype développé. / This thesis explores the application of reorientation manoeuvres to an articulated and a mobile robot architecture. These manoeuvres are often attributed to cats that are said to always land on their feet, but have also been observed in other animals and used by humans in certain sports. However, these manoeuvres are more than just a curiosity and have seen some use in the field of robotics. Indeed, reorientation manoeuvres are used for orientation control in falling robots, such as rescue robots deployed in dangerous environments, and in jumping robots. With such applications in mind, this thesis aims to develop and demonstrate fast (180-degree reorientation about one axis -- the worst-case scenario -- within the time of a one-metre fall), multi-axis reorientation manoeuvres. Firstly, an articulated architecture, along with two different manoeuvres, are designed in order to attain the desired reorientation capabilities and are tested in simulated conditions. The results obtained show that, although multi-axis reorientation is achieved, the required motor torques to reach the desired reorientation speeds are not feasible for the proposed architecture. Secondly, an omnidirectional mobile robot architecture is designed to address the limitations of the first architecture. A prototype of this mobile architecture is developed and is used to demonstrate a reorientation of 179 degrees about the pitch axis in 0.44 seconds as well as a reorientation about multiple axes. Therefore, with this prototype, the desired reorientation capabilities are achieved. Finally, the use of sensor fusion methods based on extended Kalman filtering in the context of estimation of the orientation of a free-floating platform is studied. The results obtained from this study support the viability of using such methods for on-board trajectory planning in future iterations of the developed prototype.