Fusion Motion Capture (FMC), a wearable motion capture system was developed, and applied to the optimisation of athlete performance in alpine ski racing. In what may be a world first, the three-dimensional movements of a skilled athlete (with less than 20 FIS1 points) skiing through a complete training giant slalom racecourse were analysed. FMC consists of multiple light weight sensors attached to the athlete including inertial measurement units (IMUs), pressure sensitive insoles and a global position system (GPS) receiver. The IMUs contain accelerometers, gyroscopes, and magnetometers. Limb orientation and location are obtained by mathematically combining the most reliable data from each sensor using fusion algorithms developed by the author. FMC fuses the signals from the IMUs and GPS without the need for the post filtering, usually applied to motion capture data, and therefore, maintains maximum bandwidth. The FMC results were stable and relatively independent of motion type and duration unlike other inertial systems available in 2005, when the research was initiated. Analysis of data collected from an athlete skiing giant slalom contradict the traditional „going straight turning short? race strategy. The shortest path may not always be the fastest. Instead each gate has a different optimum approach arc. Optimum turn radius increases with both increasing speed and increasing terrain slope. The results also contradict laboratory measurements of ski/snow sliding friction and suggest that snow resistance in giant slalom is of similar importance to wind drag. In addition to gravity, the athlete increased speed using the techniques of „lateral projection? and „pumping?. Race performance was determined from the analysis of the athlete skiing through the entire course. FMC proved, therefore, to be more suitable than traditional optical systems that are practically limited to capturing small sections of a race course. The athlete experienced high and rapidly fluctuating torques about all three axes of the lower joints. This information could be useful in designing training programmes racecourses and equipment to reduce knee injuries. Data driven animations and colour coded force vector diagrams were developed to enhance athlete feedback. Inline skating data was also analysed.

A percussion performance by a skilled musician will often extend beyond a written score in terms of expressiveness. This assertion is clearly evident when comparing a human performance with one that has been rendered by some form of automaton that expressly follows a transcription. Although music notation enforces a significant set of constraints, it is the responsibility of the performer to interpret the piece and “bring it to life” in the context of the composition, style, and perhaps with a historical perspective. In this sense, the sheet music serves as a general guideline upon which to build a credible performance that can carry with it a myriad of subtle nuances. Variations in such attributes as timing, dynamics, and timbre all contribute to the quality of the performance that will make it unique within a population of musicians. The ultimate goal of this research is to gain a greater understanding of these subtle nuances, while simultaneously developing a set of stochastic motion models that can similarly approximate minute variations in multiple dimensions on a purpose-built robot. Live or recorded motion data, and algorithmic models will drive an articulated robust multi-axis mechatronic system that can render a unique and audibly pleasing performance that is comparable to its human counterpart using the same percussion instruments. By utilizing a non-invasive and flexible design, the robot can use any type of drum along with different types of striking implements to achieve an acoustic richness that would be hard if not impossible to capture by sampling or sound synthesis. The flow of this thesis will follow the course of this research by introducing high-level topics and providing an overview of related work. Next, a systematic method for gesture acquisition of a set of well-defined percussion scores will be introduced, followed by an analysis that will be used to derive a set of requirements for motion control and its associated electromechanical subsystems. A detailed multidiscipline engineering effort will be described that culminates in a robotic platform design within which the stochastic motion models can be utilized. An analysis will be performed to evaluate the characteristics of the robotic renderings when compared to human reference performances. Finally, this thesis will conclude by highlighting a set of contributions as well as topics that can be pursued in the future to advance percussion robotics. / Graduate / 2019-12-10

Dans cette thèse, nous proposons de nouvelles representations pourles poses du mouvement humain, apprises sur des données réelles, envue d’une synthèse de nouveaux mouvements en temps-réel. Dans unepremière partie, nous exploitons une méthode statistique adaptée auxgroupes de Lie (Analyse en Géodésiques Principales, AGP) pour approximerla variété des poses d’un sujet en mouvement, à partir de donnéesde capture de mouvement. Nous proposons un algorithme de cinématiqueinverse exploitant cette paramétrisation réduite, permettantpar construction de synthétiser des poses proches des données initiales.Nous validons ce modèle cinématique par une application à la compressionde données de mouvements, dans laquelle seules quelques trajectoiresdes extrémités des membres du squelettes permettent de reconstruireune bonne approximation de l’ensemble des données initiales.Dans une deuxième partie, nous étendons cette approche à l’animationphysique de personnages virtuels. La paramétrisation réduitepar AGP fournit les coordonnées généralisées de la formulation Lagrangiennede la mécanique. Nous dérivons un intégrateur temporelexplicite basé sur les intégrateurs variationnels. Afin d’en améliorer lastabilité, nous proposons un modèle d’amortissement inspiré de l’algorithmede Levenberg-Marquardt. Nous présentons également une méthodegéométrique d’apprentissage des limites angulaires sur des donnéesde capture de mouvement, ainsi que leur application comme contraintescinématiques.Dans une troisième partie, nous abordons le problème du contrôledu mouvement. En formulant les étapes de la simulation physique d’unepart, et de la cinématique inverse d’autre part comme deux programmesquadratiques, nous proposons un algorithme de pseudo-contrôle parinterpolation des métriques, permettant un compromis intuitif entre simulationphysique non-contrôlée, et cinématique inverse. Cette approchefaisant intervenir des forces externes, nous proposons une formulationalternative, utilisant uniquement les forces associées à la paramétrisationréduite des poses. Cette formulation est obtenue par relaxationdu problème théorique de contrôle sous contraintes unilatérales, nonconvexe,en un programme quadratique convexe. Ces algorithmes sontévalués sur des contrôleurs d’équilibre et de suivi. / In this thesis, we propose novel, data-driven representations for humanposes, suitable for real-time synthesis of novel character motion. Inthe first part, we exploit Lie group statistical analysis techniques (PrincipalGeodesic Analysis, PGA) to approximate the pose manifold of amotion capture sequence by a reduced set of pose geodesics. We proposean inverse kinematics algorithm using this reduced parametrizationto automatically produce poses that are close to the learning set. Wedemonstrate the efficiency of the resulting pose model by an applicationto motion capture data compression, where only a few end-effector trajectoriesare used to recover a good approximation of the initial data.In the second part, we extend this approach to the physically-basedanimation of virtual characters. The PGA-reduced parametrization providesgeneralized coordinates in a Lagrangian formulation of mechanics.We derive an explicit time integrator by approximating existingvariational integrators, and propose a damping model based on theLevenberg-Marquardt algorithm. We also describe a geometric, datadriven,angular limit learning algorithm, and the associated kinematicconstraints.In the third part, we reach the problem of task-space motion control.By formulating both physical simulation and inverse kinematicstime stepping schemes as two quadratic programs, we propose a simplepseudo-control algorithm that interpolates between the two metrics.This allows for an intuitive trade-off between uncontrolled simulationand kinematic manipulation. Since this approach makes use of externalforces, we propose an alternate formulation using only the generalizedforces associated to the pose parametrization. A control algorithmis obtained by the relaxation of the exact, non-convex control problemunder unilateral constraints, into a convex quadratic program. Thesealgorithms are evaluated on simple balance and tracking controllers.

À quel point les mouvements humains sont-ils optimaux ? Cette thèse aborde cette question en se concentrant particulièrement sur les mouvements en escalade, étudiés ici sous trois aspects complémentaires que sont la collecte expérimentale de séquences de grimpe, l'analyse biomécanique de ces données, et la synthèse de gestes par optimisation temporelle. La marche fut l'objet de nombreux travaux, avec de bons résultats notamment en animation [Mordatch 2013]. Nous nous intéressons ici spécialement au problème original des mouvements d'escalade, dont la diversité et leur caractère multicontact présentent une complexité intéressante pour l'évaluation des caractéristiques du mouvement humain. L'hétérogénéité du répertoire gestuel rencontrée en escalade s'explique par plusieurs facteurs que sont l'évolution sur des parois de formes variées, la multiplicité des niveaux d'expertise des pratiquants, et des disciplines différentes au sein même de l'activité, à savoir le bloc, la difficulté, ou encore l'escalade de vitesse. Notre démarche d'exploration de ce sport se décompose en trois étapes : la collecte de données par une capture de mouvements multicaméra avec marqueurs, couplée à un ensemble de capteurs de force montés sur un mur de bloc en laboratoire ; une analyse du geste par dynamique inverse, prenant exclusivement des données cinématiques pour entrées, basée sur une minimisation des couples internes pour résoudre l'ambiguïté du multicontact, intrinsèque à l'activité d'escalade, validée par comparaison avec les mesures capteurs ; et enfin, l'utilisation d'un critère d'efficacité énergétique pour synthétiser la meilleure temporisation associée à une séquence de déplacements donnés. Les enregistrements expérimentaux se sont fait à l'université McGill qui dispose d'un mur instrumenté de 6 capteurs de forces, et d'un dispositif de capture de mouvements 24 caméras, nous ayant permis de collecter des données sur une population de 9 sujets. L'analyse de ces données constitue la deuxième partie de cette thèse. Le défi abordé est de retrouver les forces externes et les efforts internes à partir uniquement des déplacements du grimpeur. Nous supposons pour cela une répartition optimale des efforts internes. Après analyse, cette répartition s'avère être plutôt uniforme que proportionnelle aux capacités musculaires des différentes articulations du corps. Finalement, dans une troisième et dernière partie, nous nous intéressons à la temporisation des gestes en escalade, en prenant en entrée la trajectoire du grimpeur, éventuellement issue de cinématique inverse pour s'affranchir de la nécessité d'une capture par marqueurs et caméras infra-rouges. En sortie, une temporisation idéale pour cette trajectoire est trouvée. Cette temporisation s'avère réaliste, mais manque d'une modélisation des instants d'hésitation et de prise de décision, ainsi que d'un modèle d'établissements de contact, phénomène présentant un délai temporel non pris en compte pour l'instant. / How optimal are human movements ? This thesis tackles this issue by focusing especially on climbing movements, studied here under three complementary aspects which are the experimental gathering of climbing sequences, the biomechanical analysis of these data, and the synthesis of gestures by timing optimization. Walking has been largely studied, with good results in animation [Mordatch 2013]. We are interested here especially in the original question of climbing motions, whose diversity and multicontact aspect present an interesting complexity for the evaluation of the human motion characteristics. The heterogeneity of climbing gestures can be linked to several factors which are the variety of wall shapes, the multiplicity of climber skill levels, and different climbing categories, namely bouldering, route climbing or speed climbing. Our exploratory approach of this sport consists in three steps: the data collection by multicamera marker-based motion capture, combined with a set of force sensors mounted on an in-laboratory bouldering wall; a gesture analysis by inverse dynamics, taking only kinematic data as inputs, based on the minimization of internal torques to resolve the multicontact ambiguity, intrinsic to the climbing activity, validated by comparison with sensor measurements; and finally, the use of the energy efficiency criterion for synthesizing the best timing associated with a given sequence of movements. Experimental recordings were made at McGill University which has a climbing wall instrumented of 6 force sensors, and a motion capture device of 24 cameras, which allowed us to collect data on a population of nine subjects. The analysis of these data is the second part of this thesis. The addressed challenge is to find the external forces and internal torques from the climber's movements only. To this end we assume an optimal distribution of internal torques. After analysis, the distribution turns out to be rather uniform than proportional to the muscle capacity associated to each body joint. Finally, in a third and last part, we focus on the timing of climbing gestures, taking as input the path of the climber, possibly after inverse kinematics in order to overcome the need for a capture with markers and infrared cameras. As output, an optimal timing for this path is found. This timing is realistic, but lacks of a modelization for hesitation and decision making instants, as well as a model for the contact establishment, with the associated temporal delay currently not taken into account.

Submitted by Izabel Franco (izabel-franco@ufscar.br) on 2016-10-10T13:03:01Z No. of bitstreams: 1 DissIJRF.pdf: 11147756 bytes, checksum: c2f35d4805dd8c9a93b43189bacf2899 (MD5) / Approved for entry into archive by Marina Freitas (marinapf@ufscar.br) on 2016-10-20T19:38:57Z (GMT) No. of bitstreams: 1 DissIJRF.pdf: 11147756 bytes, checksum: c2f35d4805dd8c9a93b43189bacf2899 (MD5) / Approved for entry into archive by Marina Freitas (marinapf@ufscar.br) on 2016-10-20T19:39:03Z (GMT) No. of bitstreams: 1 DissIJRF.pdf: 11147756 bytes, checksum: c2f35d4805dd8c9a93b43189bacf2899 (MD5) / Made available in DSpace on 2016-10-20T19:39:09Z (GMT). No. of bitstreams: 1 DissIJRF.pdf: 11147756 bytes, checksum: c2f35d4805dd8c9a93b43189bacf2899 (MD5) Previous issue date: 2016-05-20 / Não recebi financiamento / The computational analysis of human mobility can aid in treatment and rehabilitation of people who have some kind of physical disability. Physiotherapeutic studies, called kinematics, aims to correct body’s dysfunction, caused by genetic malformations, degenerative diseases or posture problems. A computer system can be used to precisely describe motion, anatomical position and angular terms during gait. Therefore, the movements need to be captured and represented in three-dimensional space to be evaluated. This work is a study, development and evaluation of a system for capturing human movement based on recorded videos, using low cost computing resources and adapted to the reality of gait analysis. The automation of computing movements capture, as well as aid in neurological, motor or orthopedic rehabilitation will expand its application to community. In this context, a study was conducted about computing methods of human movements capture by passive markers, and defined the four main steps: calibration and synchronization of the cameras, detection and three-dimensional reconstruction. Computer experiments were made to facilitate the theoretical understanding of this work. Kinematic analyzes data were used as a study case for the implementati / A análise computacional de movimentos humanos pode auxiliar no tratamento e reabilitação de pessoas que possuem algum tipo de deficiência motora. Estudos na fisioterapia, chamado de avaliação cinemática, tem por objetivo direcionar tratamentos para corrigir as disfunções do corpo, causadas pela má formação genética, doenças degenerativas ou problemas de postura. Um sistema computacional pode ser utilizado para descrever precisamente os movimentos, posição anatômica e termos angulares durante o período da marcha. Para tanto, os movimentos precisam ser capturados e reconstruídos no espaço tridimensional para posteriormente serem avaliados. Este trabalho consiste no estudo, desenvolvimento e avaliação de um sistema para captura de movimentos humanos baseado em registros de vídeos, utilizando recursos computacionais de baixo custo e adaptado a realidade da análise de marcha. A automatização da captura computacional destes movimentos, além de auxiliar na reabilitação motora neurológica e ou ortopédica de pessoas que possuem alguma deficiência do gênero, ampliará sua aplicação na comunidade. Nesse contexto, foi realizado o levantamento bibliográfico das principais técnicas de captura computacional de movimentos, e no caso dos sistemas ópticos com marcadores passivos, e definido quatro etapas principais: calibração, sincronização das câmeras, detecção e reconstrução tridimensional. Ainda, experimentos computacionais foram realizados para o entendimento prático e teórico do trabalho. Registros de marcha de crianças foram utilizados como estudo de caso para a implementação das etapas de calibração, detecção dos marcadores e reconstrução tridimensional.

Made available in DSpace on 2016-06-02T19:52:04Z (GMT). No. of bitstreams: 1 5892.pdf: 5880561 bytes, checksum: db40111a324a7986148a35761e5c50a4 (MD5) Previous issue date: 2014-02-17 / Financiadora de Estudos e Projetos / The Motion Capture technology is considered a potential to overcome the limitations of traditional human simulation, reducing the time spent on creating and assigning greater reality simulations the simulated movements. This research presents an analysis of a technical system, formed by integrating a Motion Capture System with a human model and simulation software, aiming to establish ways to use the system in project process, contextualized by ergonomics. Different configurations of the main steps of the integration of the Motion Capture System Moven (Xsens), integrated with CAD DELMIA (Dassault Systèmes) software, which has a module of human simulation (Human Builder), through four experimental tests. Four trials were defined with factorial design to test the combination of different possible system configurations: i) spatial positioning of virtual environments, ii) mitigation of drift, iii) digital human kinematics and iv) anthropometric compatibility. From the results obtained in each test operating procedures were constructed in order to solve the problems and limitations of using these integrated technologies. The following procedures show how to configure the integrated system for positioning a virtual environment in a CAD system known coordinates, which enables the correction of the position due to drift of the capture system. Procedures that relate to the possible interactions between the body and capture the workplace, with the characteristics of the activity and the objectives of the simulation were also created. In addition, it was elaborated a systematic iterative to match the anthropometry between real human being, avatar of the capture system and digital human. The development of ways to use these technologies provides improvement of techniques used to construct the simulation, contributing to integration the perspective of future activity in the design in the design of productive situations. / A tecnologia de Captura de Movimentos é considerada potencial para superar as limitações da simulação humana tradicional, reduzindo o tempo despendido na criação de simulações e atribuindo maior realidade aos movimentos simulados. Esta pesquisa apresenta uma análise de um sistema técnico, formado pela integração de um Sistema de Captura de Movimentos com um software de modelagem e simulação humana, visando estabelecer formas de utilização do sistema no processo de projetos, contextualizados pela ergonomia. São analisadas diferentes configurações das principais etapas do processo de integração do Sistema de Captura de Movimentos Moven (Xsens), integrado com software CAD DELMIA (Dassault Systèmes), que possui um módulo de simulação humana (Human Builder), através de ensaios experimentais. Foram definidos quatro ensaios com delineamento fatorial para testar a combinação das diferentes configurações possíveis do sistema: i) posicionamento espacial de ambientes virtuais, ii) mitigação do drift, iii) cinemática do manequim digital e iv) compatibilidade antropométrica. A partir dos resultados obtidos em cada ensaio foram construídos procedimentos operacionais de configuração do sistema integrado, para equacionar as limitações e problemas de utilização destas tecnologias integradas. Os procedimentos mostram como configurar o sistema integrado, para posicionar um ambiente virtual em um sistema de coordenadas CAD conhecido, que possibilita a correção do posicionamento, devido ao drift do sistema de captura. Também foram criados procedimentos que relacionam as interações possíveis, entre o corpo de captura e o ambiente de trabalho, com as características da atividade e os objetivos da simulação. Além disso, foi elaborada uma sistemática iterativa para compatibilizar a antropometria entre o ser humano real, avatar do sistema de captura e o manequim digital. O desenvolvimento de formas de uso destas tecnologias proporciona melhoria das técnicas utilizadas para construção das simulações com manequins digitais, contribuindo para inserção da perspectiva da atividade futura no projeto das novas instalações.

The purpose of this study was investigate the measurement techniques of human movement and to develop a system capable of capturing and reconstructing the movement trajectory on real world. The devised system uses video cameras to record the movement and a computer program to process video footage assessing the movement trajectory and calculating its kinematical variables. Those data are showed by the program as graphics or animation which simulates the reconstructed movement. To reach these objectives, the system used techniques from image processing, computer vision and anthropometric models. In the end of this work, it is exhibited an experimental evaluation of the precision and accuracy of the system and some case studies to show its potential. / O propósito deste estudo foi pesquisar as técnicas de medição do movimento humano e desenvolver um sistema capaz de capturar e reconstruir a trajetória do movimento no espaço. O sistema desenvolvido utiliza câmeras de vídeo para registrar o movimento e um programa de computador capaz de processar as imagens extraídas dos vídeos de forma a estimar a trajetória do movimento e calcular as suas variáveis cinemáticas. Estes dados são exibidos pelo programa na forma de gráficos ou animações que reconstroem o movimento capturado. Para alcançar estes objetivos, o sistema utilizou técnicas de processamento de imagens, visão computacional, cálculos numéricos e modelos antropométricos do corpo humano. Ao final desta dissertação é feita uma avaliação experimental da precisão e exatidão do sistema proposto e alguns estudos de casos que tiveram como objetivo demonstrar o potencial deste sistema. / Mestre em Ciências

La complexité de l’anatomie fait de son apprentissage une tâche difficile. Au fil des années, différents supports de connaissances ont vu le jour dans le but de représenter et structurer l’anatomie : des dessins au tableau, aux livres d’anatomie, en passant par l’étape incontournable de la dissection, et des travaux pratiques sur maquettes 3d. Il est néanmoins difficile d’appréhenderla dimension dynamique de l’anatomie avec les outils d’apprentissage conventionnels ; connaissance qui est pourtant essentielle à la formation des médecins. A travers ces travaux de thèse nous proposons un système original et innovant pour l’apprentissage de l’anatomie intitulé « Living Book of Anatomy » (LBA). L’idée étant, pour un utilisateur donné, de superposer à sa propre image une maquette anatomique 3d (peau, squelette, muscles et viscères) et del’animer en mimant les mouvements de celui-ci. Nous parlons ici d’une application temps-réel de type « miroir augmenté ». Nous utilisons la Kinect comme capteur de mouvement.Le premier défi à relever est l’identification de caractéristiques morphologiques qui nous permettront de recaler notre maquette anatomique 3d sur l’utilisateur. Nous proposons ici deux technologies interchangeables en fonction des besoins. La première méthode, temps-réel, est basée sur l’utilisation de transformations affines attachées entre les repères positionnés à chaque articulation du squelette donné par la Kinect pour déformer la maquette 3d à l’aide de poids de skinning prédéfinis. La seconde méthode, plus couteuse en temps (de l’ordre de quelques minutes), se découpe en trois parties : dans un premier temps nous déformons la peau à l’aide de la position des articulations du squelette d’animation Kinect et du nuage de pointpartiel de l’utilisateur ; à partir de cela et de règles anatomiques strictes nous déformons le squelette ; pour finir nous déformons les tissus mous pour qu’ils comblent l’espace entre le squelette et la peau. Le second défi concerne la capture réaliste et temps-réel des mouvements utilisateurs. Reproduire le comportement des structures anatomiques est une tâche complexe due aux informations Kinect souvent partielles et très bruitées. Nous proposons ici l’utilisation de règles anatomiques concernant les articulations du corps (axes de rotation et butées articulaires) pour contraindre les mouvements donnés par la Kinect et obtenir des mouvements réalistes. Pour obtenir des mouvements fluides nous nous proposons d’utiliser des filtrages, notamment le filtre de Kalman. Le dernier défi concerne la dominante de retour visuel et d’interaction.Lors de ces travaux nous nous sommes tout particulièrement intéressés à un renducorps complet pour montrer le fonctionnement général du corps humain et de ces différentes articulations. Nous avons également choisi le membre inférieur comme structure anatomique d’intérêt avec pour but la mise en avant de phénomènes anatomiques spécifiques, comme l’activité musculaire.Les différents éléments ont été intégrés dans un système opérationnel présenté en détails dans ce manuscrit de thèse. Grâce à des expérimentations - avec des étudiants et des professionnels de différents domaines - et la présentation de ces travaux sous forme de démonstrations lors de différents congrès, nous avons validé cet outil / To ease the complex task of anatomy learning, there exist many ways to represent and structure anatomy : illustrations, books, cadaver dissections and 3d models. However, it is difficult to understand and analyse anatomy motion, which is essential for medicine students. We present the "Living Book of Anatomy" (LBA), an original and innovative tool to learn anatomy. For a specific user, we superimpose a 3d anatomical model (skin, skeleton, muscles and visceras) onto the user’s color map and we animate it following the user’s movements. We present a real-time mirror-like augmented reality (AR) system. A Kinect is used to capturebody motions.The first innovation of our work is the identification of the user’s body measurements to register our 3d anatomical model. We propose two different methods to register anatomy.The first one is real-time and use affine transformations attached to rigid positioned on each joint given by the Kinect body tracking skeleton in order to deform the 3d anatomical model using skinning to fit the user’s measurements.The second method needs a few minutes to register the anatomy and is divided in 3 parts : skin deformation (using Kinect body tracking skeleton and the Kinect partial point cloud), with it and strict anatomical rules we register the skeleton. Lastly we deformm the soft tissues to completly fill the space inbetween the registered skeleton and skin.Secondly, we want to capture realistically and in real-time the user’s motion. To do that we need to reproduce anatomical structure motion but it is a complex task due to the noisy and often partial Kinect data. We propose here the use of anatomical rules concerning body articulations (angular limits and degrees of freedom) to constraint Kinect captured motion in order to obtain/gain plausible motions. a kalman filter is used to smooth the obtaiined motion capture.Lastly, to embed visual style and interaction, we use a full body reproduction to show general knowledge on human anatomy and its differents joints. We also use a lower-limb as structure of interest to higlight specific anatomical phenomenon, as muscular activity.All these tools have been integrated in a working system detailed in this thesis.We validated our tool/system by presenting it as a live demo during different conferences and through user studies done with students and professionnals from different backgrounds

Qu'il tente de prévenir la chute d'une personne âgée, de traduire la langue des signes ou de contrôler un humain virtuel, l'analyse de gestes est un vaste domaine de recherche qui s'attelle à reconnaître, classifier, segmenter, indexer ou encore évaluer différents types de mouvements. Cependant, peu de travaux se concentrent sur cette dernière approche d'évaluation. Ce travail de thèse propose de mettre en place un outil d’évaluation automatique et générique d’un geste sportif, reposant sur l’utilisation d’une base de données de gestes experts acquis via un système de capture de mouvements. Afin d’extraire un mouvement de référence, l’algorithme de déformation temporelle dynamique (DTW) est considéré pour aligner puis moyenner les gestes. Les méthodes d’alignements et de moyennage de séries temporelles se confrontant aux conséquences néfastes de chemins de déformation du DTW pathologiques, des contraintes locales sont ajoutées et donnent lieu à un nouvel algorithme appelé CDBA. La qualité d’un geste est estimée spatialement et temporellement à chaque instant et pour chaque membre par comparaison avec le geste de référence et pondérée par la dispersion des données expertes autour de ce geste moyen. Le processus ainsi mis en place est validé à partir de gestes de karaté et de tennis annotés par des entraîneurs. Un premier prototype d’outil d’entraînement en ligne est finalement proposé et laisse entrevoir les potentialités d’usage qui pourraient être menées à la suite de ce travail. / Either to reduce falling risks in elderly people, to translate the sign language or to control a virtual human being, gesture analysis is thriving research field that aims at recognizing, classifying, segmenting, indexing and evaluating different types of motions. As few studies tackle the evaluation process, this PhD focuses on the design of an autonomous system for the generic evaluation of sport-related gestures. The tool is trained on the basis of experts’ motions recorded with a motion capture system. Dynamic Time Warping (DTW) is deployed to obtain a reference gesture thanks to data alignment and averaging. Nevertheless, this standard method suffers from pathological paths issues that reduce its effectiveness. For this reason, local constraints are added to the new DTW-based algorithm, called CDBA (Constrained DTW Barycenter Averaging). At each time step and for each limb, the quality of a gesture is spatially and temporally assessed. Each new motion is compared to the reference gesture and weighted in terms of data dispersion around the reference.The process is validated on annotated karate and tennis databases. A first online training prototype is given in order to prompt further research on this subject.

Detta kandidatarbete fokuserar på att undersöka aspekter som kan förstärka en användares upplevelse inom digitala spel. De aspekter som vi valt att fokusera på är användandet av Oculus Rift och Motion Capture då de tar upp viktiga punkter som Virtual Reality, närvaro och inlevelse i spel. Vi valde även att fokusera vår produktion mot skräckgenren då detta är ett område vi båda har stor passion för i diverse media. I vårt arbete har vi fått fram en hypotes om att spel som fokuserar på att lämna karaktären i ett tomt skal (ingen personlighet), lämnar en större chans för inlevelse hos spelaren då användarens egna känslor speglas på karaktären. Användandet av Oculus Rift och realistiska röresler via Motion Capture har en tendens att även stärka detta. / Detta är en reflektionsdel till en digital medieproduktion.