Grasping, the skill to hold objects and tools while doing in-hand manipulation, still is in many cases an unsolvable problem for robotics, but a natural act for humans. An efficient grasping requires not only human-like robotic hands with articulated fingers but also tactile, force, and kinesthetic sensors for the precise control of the forces and motions exerted during the manipulation. As a fully autonomous robotic dexterous manipulation is too difficult to develop for changing and unstructured environments, an alternative approach is to combine the low-level robot computer control with the higher-level perception and task planning abilities of a human operator equipped with an adequate human-computer interface (HCI). This thesis presents theoretical and experimental contributions to the development of an upgraded haptic-enabled anthropomorphic Ring Ada dexterous robotic hand and a biology-inspired synergistic real-time control system for teleoperated grasping of different objects using a CyberTouch HCI data glove. A fuzzy logic controller module was developed to efficiently control the underactuated Ring Ada’ robotic hand during grasping. A machine learning classification system was developed to recognize grasped objects. Experiments have convincingly demonstrated that our novel Ring Ada robotic hand equipped with kinematic position sensors and touch sensors is able to efficiently grasp different lightweight objects through teleoperation.

The study of fixed wing aeronautical engineering has matured to the point where years of research result in small performance improvements.  In the past decade, micro air vehicles, or MAVs, have gained attention of the aerospace and robotics communities.  Many researchers have begun investigating aircraft schemes such as ones which use rotary or flapping wings for propulsion.  While the engineering of rotary wing aircraft has seen significant advancement, the complex physics behind flapping wing aircraft remains to be fully understood.  Some studies suggest flapping wing aircraft can be more efficient when the aircraft operates in low Reynolds regimes or requires hovering.  Because of this inherent complexity, the derivation of flapping wing control methodologies remains an area with many open research problems.  This thesis investigates flapping wing vehicles whose design is inspired by avian flight.  The flapping wing system is examined in the cases where the core body is fixed or free in the ground frame.  When the core body is fixed, the Denavit Hartenberg representation is used for the kinematic variables.  An alternative approach is introduced for a free base body case.  The equations of motion are developed using Lagranges equations and a process is developed to derive the aerodynamic contributions using a virtual work principle.  The aerodynamics are modeled using a quasi-steady state formulation where the lift and drag coefficients are treated as unknowns.  A collection of nonlinear controllers are studied, specifically an ideal dynamic inversion controller and two switching dynamic inversion controllers.  A dynamic inversion controller is modified with an adaptive term that learns the aerodynamic effects on the equation of motion.  The dissipative controller with adaptation is developed to improve performance.  A Lyapunov analysis of the two adaptive controllers guarantees boundedness for all error terms.  Asymptotic stability is guaranteed for the derivative error in the dynamic inversion controller and for both the position and derivative error in the dissipative controller.  The controllers are simulated using two dynamic models based on flapping wing prototypes designed at Virginia Tech.  The numerical experiments validate the Lyapunov analysis and illustrate that unknown parameters can be learned if persistently excited. / Master of Science

Unmanned ground vehicles are increasingly employing a combination of active sensors such as LIDAR with passive sensors like cameras to perform at all levels of perception, which includes detection, recognition and classification. Typical cameras measure the intensity of light at a variety of different wavelengths to classify objects in different areas of an image. A polarimetric camera not only measures intensity of light, but can also determine its state of polarization. The polarization of light is the angle the electric field of the wave of light takes as it travels. A polarimetric camera can identify the state of polarization of the light, which can be used to segment highly polarizing areas in a natural environment, such the surface of water. The polarimetric camera designed and built for this thesis was created with low cost in mind, as commercial polarimetric cameras are very expensive. It uses multiple beam splitters to split incoming light into four machine vision cameras. In front of each machine vision camera is a linear polarizing filter that is set to a specific orientation. Using the data from each camera, the Stokes vector can be calculated on a pixel by pixel basis to determine what areas of the image are more polarized. Test images of various scenes that included running water, standing water, mud, and vehicles showed promise in using polarization data to highlight and identify areas of interest. This data could be used by a UGV to make more informed decisions in an autonomous navigation mode. / Master of Science

Les robots sériels de grandes dimensions apportent dextérité, répétabilité et flexibilité dans les chaînes de production. Sur ces chaînes, des opérations telles que la mesure de pièces peuvent exploiter ces avantages très attractifs. Il est cependant impératif de mieux maîtriser le positionnement de l’effecteur de ces robots, pour répondre aux exigences de la mesure 3D. Dans ce contexte, une cellule de numérisation robotisée a été développée, exploitant un robot sériel 6 axes comme support d’un capteur laser plan (KZ25 Kreon), et utilisant un système de stéréovision externe pour le suivi de l’opération et la calibration de la cellule (C-Track Creaform). La calibration que nous proposons pour cette cellule permet de maîtriser la qualité et d’optimiser la vitesse de numérisation, et se veut à la fois rapide et pratique (peu de contraintes et de matériel) pour répondre à un contexte industriel.Cette calibration passe par l’identification des paramètres d’un modèle géométrique pour le robot, à l’aide d’une nouvelle méthode que nous proposons, généralisant le concept d’étude d’arc de cercle proposé dans la méthode CPA (Circle Point Analysis). Une étude comparative démontre les avantages de cette nouvelle méthode par rapport aux méthodes classiquement utilisées. Une méthode de sélection que nous avons développée permet ensuite de compléter le modèle du robot avec des paramètres non-géométriques pertinents (flexibilités, jeux). Au cours de la calibration, nous étudions également les capacités du robot en vitesse et en qualité de positionnement au travers d’indices de performance originaux. Enfin nous avons élaboré et validé une méthode de calibration en position et orientation du KZ25 sur son support (le robot) ce qui permet une numérisation à 6 DDL.En perspective, un générateur de trajectoires, à donner en consigne au robot, devra utiliser cette calibration de la cellule entière pour maîtriser la qualité de la numérisation et optimiser sa durée. / Serial robots are designed for repetitive tasks in wide workspaces, and provide good dexterity and flexibility to production lines. Restrictive applications such as parts digitizing can make use of those attractive benefits. However, digitizing needs on accuracy require to work carefully on the quality of the robot end-effector positioning. In this context, a digitizing cell using a 6 axis robot as displacement system for a laser plane sensor (KZ25) was developed. In this cell all calibrations are ensured by an external measurement system (C-Track Creaform) following the end-effector. The goal is to master the digitizing quality and to optimize the operating speed thanks to quick and convenient calibrations (few equipment, installation and constraints).First, the parameters of the robot geometric model are identified with a quick and convenient method we developed, based on the existing circle point analysis method (CPA). A comparative study shows the advantages of our identification method over classic methods. Then a selective method we proposed allows us to complete this geometric model with relevant non-geometric parameters such as flexibilities or backlashes. Robot performances in terms of speed and posing quality are also studied through new performance indexes. Finally, we had to create and validate a calibration method for the position and orientation of the KZ25 sensor in order to exploit the unrestricted orientation provided by the robot end- effector.In our work prospects, a path generation strategy will use those calibrations to create paths for the robot, with mastered digitizing quality and optimized speed.

Perception is a key feature in how any creature or autonomous system relates to its environment. While there are many types of perception, this thesis focuses on the improvement of the visual robotics perception systems. By implementing a broadband passive sensing system in conjunction with current perception algorithms, this thesis explores scene reconstruction and world modeling. The process involves two main steps. The first is stereo correspondence using block matching algorithms with filtering to improve the quality of this matching process. The disparity maps are then transformed into 3D point clouds. These point clouds are filtered again before the registration process is done. The registration uses a SAC-IA matching technique to align the point clouds with minimum error.  The registered final cloud is then filtered again to smooth and down sample the large amount of data. This process was implemented through software architecture that utilizes Qt, OpenCV, and Point Cloud Library. It was tested using a variety of experiments on each of the components of the process.  It shows promise for being able to replace or augment existing UGV perception systems in the future. / Master of Science

<p>Insects and hummingbirds not only can perform long-term stationary hovering but also are capable of acrobatic maneuvers. At their body scale, such extraordinary flight performance remains unmatched by state-of-the-art conventional man-made aerial vehicles with fixed or rotary wings. Insects' and hummingbirds' near maximal performance come from their highly intricate and powerful wing-thorax actuation systems, sophisticated sensory system, and precise neuromotor control. Flapping Wing Micro Air Vehicles (FWMAVs) with bio-inspired flapping flight mechanisms hold great promise in matching the performance gap of their natural counterparts. Developing such autonomous flapping-wing vehicles to achieve animal-like flight, however, is challenging. The difficulties are mainly from the high power density requirements under the stringent constraints of scale, weight, and power, severe system oscillations induced by high-frequency wing motion, high nonlinearity of the system, and lack of miniature navigation sensors, which impede actuation system design, onboard sensing, flight control, and autonomous navigation. </p><p><br></p><p>To address these open issues, in this thesis, we first introduce systematic modeling of a dual-motor hummingbird-scale flapping wing robot. Based upon it, we then present studies of the onboard sensor fusion, flight control, and navigation method. </p><p><br></p><p>By taking the key inspiration from its natural counterparts, the proposed hummingbird robot has a pair of independently controlled wings. Each wing is directly actuated by a dc motor. Motors undergo reciprocating motion. Such a design is a severely underactuated system, namely, it relies on only two actuators (one per wing) to control full six degrees of freedom body motion. As a bio-inspired design, it also requires the vehicle close to its natural counterparts’ size and weight meanwhile provide sufficient lift and control effort for autonomy. Due to stringent payload limitation from severe underactuation and power efficiency challenges caused by motor reciprocating motion, the design and integration of such a system is a challenging task. In this thesis, we present the detailed modeling, optimization, and system integration of onboard power, actuation, sensing, and flight control to address these unique challenges. As a result, we successfully prototyped such dual-motor powered hummingbird robot, either with power tethers or fully untethered. The tethered platform is used for designing onboard sensing, control, and navigation algorithms. Untethered design tackles system optimization and integration challenges. Both tethered/untethered versions demonstrate sustained stable flight. </p><p><br></p><p>For onboard attitude sensing, a real-time sensor fusion algorithm is proposed with model-based adaptive compensation for both sensor reading drift and wing motion induced severe system vibration. With accurate and robust sensing results, a nonlinear robust control law is designed to stabilize the system during flight. Stable hovering and waypoint tracking flight were experimentally conducted to demonstrate the control performance. In order to achieve natural flyers' acrobatic maneuverability, we propose a hybrid control scheme by combining a model-based robust controller with a model-free reinforcement learning maneuver policy to perform aggressive maneuvers. The model-based control is responsible for stabilizing the robot in nominal flight scenarios. The reinforcement learning policy pushes the flight envelope to pilot fierce maneuvers. To demonstrate the effectiveness of the proposed control method, we experimentally show animal-like tight flip maneuver on the proposed hummingbird robot, which is actuated by only two DC motors. These successful results show the promise of such a hybrid control design on severely underactuated systems to achieve high-performance flight.</p><p><br></p><p>In order to navigate confined space while matching the design constraints of such a small robot, we propose to use its wings in dual functions - combining sensing and actuation in one element, which is inspired by animals' multifunctional flapping wings. Based on the interpretation of the motor current feedback which directly indicates wing load changes, the onboard somatosensory-like feedback has been achieved on our hummingbird robot. For navigation purposes, such a method can sense the presence of environmental changes, including grounds, walls, stairs, and obstacles, without the need for any other sensory cues. As long as the robot can fly, it can sense surroundings. To demonstrate this capability, three challenging tasks have been conducted onto the proposed hummingbird robot: terrain following, wall detection and bypass, and navigating a confined corridor. </p><p><br></p><p>Finally, we integrate the proposed methods into the untethered platform, which enables stable untethered flight of such a design in both indoor and outdoor tests. To the best of our knowledge, this result presents the first bio-inspired FWMAV powered by only two actuators and capable of performing sustained stable flight in both indoor and outdoor environment. It is also the first untethered flight of an at-scale tailless hummingbird robot with independently controlled wings, a key inspiration from its natural counterparts.</p><div><div><div><div><div> </div> </div> </div></div></div>

No description available.

No description available.

No description available.

京都大学 / 新制・課程博士 / 博士(医学) / 甲第24480号 / 医博第4922号 / 新制||医||1063(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 波多野 悦朗, 教授 山本 洋介, 教授 小林 恭 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM