none

Alors que les systèmes de reconnaissance gestuelle actuels privilégient souvent un usage intérieur, nous nous intéressons à la conception d'un système dont l'utilisation est possible en environnement extérieur et en mobilité. Notre objectif est le contrôle temps-réel d'un robot mobile dont l'usage est destiné aux fantassins débarqués. La contribution principale de cette thèse est le développement d'une chaîne de reconnaissance gestuelle temps réel, qui peut être entraînée en quelques minutes avec: un faible nombre d'exemples ("small data"); des gestes choisis par l'utilisateur; une résilience aux gestes mal réalisés; ainsi qu'une faible empreinte CPU. Ceci est possible grâce à deux innovations clés: d'une part, une technique pour calculer des distances entre séries temporelles en flux, basée sur DTW; d'autre part, une rétro-analyse efficace du flux d'apprentissage afin de déterminer les hyperparamètres du modèle sans intervention de l'utilisateur. D'autre part, nous avons construit notre propre gant de données et nous l'utilisons pour confirmer expérimentalement que la solution de reconnaissance gestuelle permet le contrôle temps réel d'un robot en mobilité. Enfin, nous montrons la flexibilité de notre technique en ce sens qu'elle permet de contrôler non seulement des robots, mais aussi des systèmes de natures différentes. / Although gesture recognition has been studied for several decades, much research stays in the realm of indoors laboratory experiments. In this thesis, we address the problem of designing a truly usable, real- world gesture recognition system, focusing mainly on the real-time control of an outdoors robot for use by military soldiers. The main contribution of this thesis is the development of a real-time gesture recognition pipeline, which can be taught in a few minutes with: very sparse input ("small data"); freely user-invented gestures; resilience to user mistakes during training; and low computation requirements. This is achieved thanks to two key innovations: first, a stream-enabled, DTW-inspired technique to compute distances between time series; and second, an efficient stream history analysis procedure to automatically determine model hyperparameters without user intervention. Additionally, a custom, hardened data glove was built and used to demonstrate successful gesture recognition and real-time robot control. We finally show this work's flexibility by furthermore using it beyond robot control to drive other kinds of controllable systems.

A study involving the movement of small-sized nursery plant containers was conducted using surface electromyography (EMG) to assess the effect that glove type has on forearm and shoulder muscle activation. A total of 24 participants were asked to move weighted nursery containers simultaneously with both left and right hands (one onegallon, two one-gallon and one three-gallon) from a floor location to a table located twenty feet away while wearing one of four glove treatments (1. No Glove; 2. Thick Leather; 3. Grip Assist Mechanics; 4. Thin Nitrile). Muscle activation was evaluated as a percentage of the participants’ maximum voluntary exertion (MVE). The results show no glove effect difference for the smaller pot configuration. With the larger container treatments, muscle activity was affected by glove treatment, specifically for the left and right flexor and extensor radialis muscles.

Thesis (Ph. D.)--University of Texas at Austin, 2004. / Supervisor: Steven Biegalski. Vita. Includes bibliographical references. Also available from UMI.

No description available.

This thesis aims to investigate how different feature extraction methods applied to textual data affect the results of multi-label classification. Two different Bag of Words extraction methods are used, specifically the Count Vector and the TF-IDF approaches. A word embedding method is also investigated, called the GloVe extraction method. Multi-label classification can be useful for categorizing items, such as pieces of music or news articles, that may belong to multiple classes or topics. The effect of using different pre-processing methods is also investigated, such as the use of N-grams, stop-word elimination, and stemming. Two different classifiers, an SVM and an ANN, are used for multi-label classification using a Binary Relevance approach. The results indicate that the choice of extraction method has a meaningful impact on the resulting classifications, but that no one method consistently outperforms the others. Instead the results show that the GloVe extraction method performs the best for the recall metrics, while the Bag of Words methods perform the best for the precision metrics. / Detta arbete ämnar att undersöka vilken effekt olika metoder för att extrahera särdrag ur textdata har när dessa används för att multi-tagga textdatan. Två metoder baserat på Bag of Words undersöks, närmare bestämt Count Vector-metoden samt TF-IDF-metoden. Även en metod som använder sig av word embessings undersöks, som kallas för GloVe-metoden. Multi-taggning av data kan vara användbart när datan, exempelvis musikaliska stycken eller nyhetsartiklar, kan tillhöra flera klasser eller områden. Även användandet av flera olika metoder för att förbehandla datan undersöks, såsom användandet utav N-gram, eliminering av icke-intressanta ord, samt transformering av ord med olika böjningsformer till gemensam stamform. Två olika klassificerare, en SVM samt en ANN, används för multi-taggningen genom använding utav en metod kallad Binary Relevance. Resultaten visar att valet av metod för extraktion av särdrag har en betydelsefull roll för den resulterande multi-taggningen, men att det inte finns en metod som ger bäst resultat genom alla tester. Istället indikerar resultaten att extraktionsmetoden baserad på GloVe presterar bäst när det gäller 'recall'-mätvärden, medan Bag of Words-metoderna presterar bäst gällade 'precision'-mätvärden.

Hand impairments have a significant impact on quality of life and career performance. This effect is specially bold in the agricultural community, since farming activities involve continuously carrying and lifting heavy objects. Assistive robotic technologies hold considerable promise in alleviating such impairment issues. However, no portable assistive device is developed for farming applications, which requires additional considerations to ensure functionality of the device and its practicality in agricultural settings. In this work, a bi-layered structure for a robotic glove is presented, which consists of a passive extension and an active flexion layers. The former is responsible for extending the fingers, using a set of elastic bands. The flexion layer, which helps with flexing the fingers and grasping of objects, is a lightweight, self-contained, portable, cable-driven, and single-actuated robotic glove. The cable configuration is inspired from the human hand flexor tendons. Due to uncertainties associated with the fabric's flexibility and potential slippage between the cable and the glove, the designed mechanisms and sensory and control systems are initially implemented on a robotic hand. The rigid structure of the robotic hand provides a suitable proving ground for the design and control ideas. The novel power transmission system design enables the active layer to perform adaptive grasp of objects with unknown shapes, sizes, and material textures. The sensory system includes a bend sensor to detect the wearer's intention to perform grasp or release actions. Additionally, a PVDF-based sensor is developed for slip-detection, which is used as feedback to prevent further slipping of the grasped objects. Overall, the active flexion layer weighs 265 gr and can provide the maximum grasping force of 122 N, which is a noticeable improvement in comparison to the literature. / Doctor of Philosophy / Hand impairments have a significant impact on quality of life and career performance. This effect is specially bold in the agricultural community, since farming activities involve continuously carrying and lifting heavy objects. Assistive robotic technologies hold considerable promise in alleviating such impairment issues. However, no portable assistive device is developed for farming applications, which requires additional considerations to ensure functionality of the device and its practicality in agricultural settings. In this work, a bi-layered structure for a robotic glove is presented, which helps with grasping objects. The first layer is responsible for extending the fingers, using a set of elastic bands. The second layer, which helps with flexing the fingers, is a lightweight, self-contained, and portable robotic glove. A novel cable-driven power transmission system is designed to perform reliable grasps using only one actuator. The power transmission system design enables the robotic glove to grasp objects with unknown shapes, sizes, and material textures. The intention of the wearer for performing a grasp or releasing an object is detected using a bend sensor. Additionally, a vibration sensor is utilized for detecting the slip of the grasped object and preventing further slipping and dropping the object. The functionality of the developed robotic gloved is evaluated through experiments, where different geometry and weight of objects are grasped.

An experimental study was performed to investigate the integrity of gloveboxes when subjected to lathe drip pan fires. These fires are potentially dangerous since glovebox failure may allow hazardous gases to escape containment. A full scale mockup of a glovebox and corresponding air flow system was constructed. Careful consideration was given to the two components expected to cause glovebox failure: the gloves and windows. In addition to normal tests, tests which introduced added ventilation openings (i.e. missing gloves, missing window) were also performed. The glovebox ventilation system places these fires in the category of overhead forced ventilation compartment fires. Since little data has been obtained previously for this type of fire, further experiments were conducted to determine the effect of fuel surface area on fire behavior. In the past, these fires have been successfully modelled as well-stirred reactors. Results showed that overall containment was achieved under normal glovebox conditions. Added ventilation opening tests, however, showed that these scenarios would lead to a loss of containment. Nevertheless, under no conditions did a catastrophic glovebox failure occur. Furthermore, experiments with reduced fuel surface areas showed that the fires became less hazardous as the pan diameter decreased. Exhaust gas concentrations, temperature data, burn rates, smoke generation, and heat releases were the criteria used to form this conclusion. Neither a well-stirred or two-layer environment was observed. / Master of Science

This thesis explores the field of hand exoskeleton robotic systems with slip detection and its applications. It presents the design and control of the intelligent sensing and force- feedback exoskeleton robotic (iSAFER) glove to create a system capable of intelligent object grasping initiated by detection of the user’s intentions through motion amplification. Using a combination of sensory feedback streams from the glove, the system has the ability to identify and prevent object slippage, as well as adapting grip geometry to the object properties. The slip detection algorithm provides updated inputs to the force controller to prevent an object from being dropped, while only requiring minimal input from a user who may have varying degrees of functionality in their injured hand. This thesis proposes the use of a high dynamic range, low cost conductive elastomer sensor coupled with a negative force derivative trigger that can be leveraged in order to create a controller that can intelligently respond to slip conditions through state machine architecture, and improve the grasping robustness of the exoskeleton. The mechanical and electrical improvements to the previous design, the sensing and force- feedback exoskeleton robotic (SAFER) glove, are described while details of the controller design and the proposed assistive and rehabilitative applications are explained. Experimental results confirming the validity of the proposed system are also presented. In closing, this thesis concludes with topics for future exploration. / Master of Science / Exoskeletons are robotic systems that have rigid external covering, such as links, joints, and/or soft artificial tendons or muscles, for the desired body part to provide support and/or protection. These are typically used to enhance power and strength, provide rehabilitation and assistance, and teleoperate other robots from a distance. While the US Army developed exoskeletons for strengthening purposes, another potential purpose of exoskeletons, which is serving medical needs, such as rehabilitation, attracted a lot of attention. Among numerous illnesses and injuries that may lead to impaired hand functionality, the U.S. Department of Health and Human Services estimated that approximately 470,000 people survive strokes every year in the United States and require continuous rehabilitation to recover their motor functions. Though medical professionals believe that the intensity and duration of rehabilitation is the key for maximizing the rate of recovery, it is often limited due to many reasons, such as cost or difficulty in attending rehabilitation sessions. To augment the availability and quality of rehabilitation, the study of exoskeletons has earned popularity. Beyond the capability of providing simple movements, such as passive rehabilitation, many scientists researched to provide active rehabilitation, which involves active participation from the patients. Furthermore, detecting the patient’s intention to activate the rehabilitation glove became a topic of interest, and many types of sensors were utilized in research. This thesis explores the design and control of the intelligent sensing and force- feedback exoskeleton robotic (iSAFER) glove, which detects the user’s intentions to activate the system through motion amplification. The iSAFER glove performs soft initial grasp until the fingers touch an object. After the object is gently grabbed and lifted, the grasp is autonomously adjusted through slip detection until there is no more slip. To facilitate this idea, a low cost force sensor was created and leveraged to improve the grasping control of the exoskeleton. The mechanical and electrical improvements to the previous design, the sensing and force-feedback exoskeleton robotic (SAFER) glove, are described while details of the controller design and the proposed assistive and rehabilitative applications are explained. Experimental results confirming the validity of the proposed system are also presented. In closing, this thesis concludes with topics for future exploration.

This thesis explores the field of robotic hand exoskeletons and their applications. These systems have emerged in popularity over the years, due to their potentials to advance the medical field as assistive and rehabilitation devices, and the field of virtual reality as haptic gloves. Although much progress has been made, hand exoskeletons are faced with several design challenges that are hard to overcome without having some tradeoffs. These challenges include: (1) the size and weight of the system, which can affect both the comfort of wearing it and its portability, (2) the ability to impose natural joint angle relationships among the user's fingers and thumb during grasping motions, (3) safety in terms of limiting the range of motions produce by the system to that of the natural human hand and ensuring the mechanical design does not cause harm or injury to the user during usage, (4) designing a device that is user friendly to use, and (5) the ability to effectively perform grasping motions and provide sensory feedback for the system to be applicable in various application fields. In order to address these common issues of today's state-of-the-art hand exoskeleton systems, this thesis proposes a mechanism design for a novel hand exoskeleton and presents the integration of several prototypes. The proposed hand exoskeleton is designed to assist the user with grasping motions while maintaining a natural coupling relationship among the finger and thumb joints to resemble that of a normal human hand. The mechanism offers the advantage of being small-size and lightweight, making it ideal for prolong usage. Several applications are discussed to highlight the proposed hand exoskeleton functionalities in processing sensory information, such as position and interactive forces. / MS / Hand exoskeletons are wearable devices that are designed to augment, reinforce, and/or restore hand performances and movements among the fingers and thumb. These hand exoskeleton systems have emerged in popularity within the medical field, where they serve as rehabilitation devices or assistive gloves, and within the field of virtual reality as haptic devices. Throughout the years, many hand exoskeleton designs have been proposed and even developed further into commercial products. Unfortunately, there still exist many design challenges for making an efficient and feasible hand exoskeleton without experiencing major tradeoffs. Some of the common challenges include designing a hand exoskeleton that is small in size, lightweight, and able to achieving natural grasping motions efficiently. As an attempt to overcome these design challenges, the work of this thesis presents a mechanism design for a novel hand exoskeleton that can serve as a general purpose glove across several applications. The design of the mechanism is described in detail with preliminary analysis. In addition, this thesis presents the design and development of several prototypes, which were made by extending the mechanism into fully integrated systems. The experimental validations of these prototypes are presented as well as their application potentials. To conclude the thesis, a discussion of the on-going future work is given.