This dissertation describes the development of an assist-device aimed to deliver 3D graphic information to the visually impaired people. A human-in-loop approach was used to analyze whether a virtual 3D shape can be transferred correctly to the human users.The proposed device in this dissertation consists of two major parts: (a) A system of position sensors for real time localization based on magnetization, and (b) A single vibratory actuator working at varied frequencies based on its real time location. The error bound of the position measurement was tested to be 2 mm, which defined the machine resolution of the shape display. In order to realize the refresh rate of the localization that can follow user's scanning speed, the parallel data processing sequences for computer and microcontroller were designed. Additionally, vibratory electromagnetic (EM) actuators were discussed based on eddy current and permanent magnet methods. The simulation study showed that eddy current method was not applicable for millimeter size coil. Accordingly, the permanent magnet method was developed and the force detection threshold of human tactile perceptions was studied.Virtual shape perception experiments were made with participation of 3 volunteers who were not aware of the 3D shape information prior to the tests. Based on the four sets of shape tests, we conclude that the majority of the shape information is able to be delivered to users by using the proposed device. Difficulties for perceiving the local sharp profile e.g. thin plates and large curvature in small shapes may be better addressed by multiple actuators simultaneously providing shape information in the local boundary detection.The major contribution of this dissertation is the 3D shape display implemented by a miniature and low cost device. The developed device utilizes both passive stimulation and active search so that a commonly used large scale actuators matrix based on mere active touch method is avoided. The studies on the required force/energy input from the actuator showed that EM actuators can be miniaturized to millimeter scale without sacrificing the ability to induce tactile stimulation. Additional uniqueness of the proposed system is the ability to present hollow features, which is impossible to display by the existing devices.

Automated positioning systems designed to measure three-dimensional locations of objects are of paramount importance to flexible manufacturing applications. These systems should perform in an industrial environment, withstanding obstacles of solid objects and must be immune from external influences including changes in atmospheric conditions and surrounding noise. Automated positioning systems should also be free of mechanical contact and able to perform without having to establish a line-of-sight with the measured object. In this thesis, a novel design is proposed for the spatial measurement of the six degrees of freedom industrial robots and autonomous vehicles. Not only does the proposed system comply with the above characteristics, but it is also capable of achieving better resolutions than CCD cameras, easier to implement, safer than laser devices and more accurate than ultrasound systems.[...] / Doctor of Philosophy in Engineering

Cette thèse s’inscrit dans le cadre applicatif général de l'analyse de cellules uniques. Afin d’améliorer la sélectivité du tri de cellules uniques, les équipes de FEMTO-ST proposent de contrôler en boucle fermée les trajectoires des cellules entemps réel pendant leur trajet dans les puces dédiées au tri. Dans ce cadre, mes travaux de thèse portent sur une méthode novatrice de détection en temps-réel de la position des cellules, directement intégrée aux puces et basée sur le principephysique de la mesure d'impédance.Lors du passage d’une cellule dans un microcanal, celle-ci vient modifier l’impédance mesurable entre des électrodes placées sur les bords du canal.Une méthode générique permettant de formuler les variations d'impédance en fonction de la position de la particule été proposée (modèledirect).Une méthode d'estimation de la position d’une particule reposant sur les mesures d’impédance a également été proposée (modèle inverse). Celle-ci exploite un filtre de Kalman étendu, permettant la fusion de données en provenance de plusieurspaires d'électrodes, et exploitant les informations disponibles telles que la distribution du bruit de mesure et le modèle dynamique de la particule.La validation de la méthode a été effectuée sur un banc expérimental qui a été développé lors de cette thèse et sur des simulations numériques.Ces travaux montrent la pertinence d’exploiter l’impédancemétrie pour construire un capteur de position de particules immergées dans un microcanal. Cette méthode est une alternative à l’utilisation de microscopes optiques et présente l’avantage d’une grande compacité. / This thesis is part of the general application framework of single cell analysis. In order to improve the selectivity of single cell sorting, FEMTO-ST teams propose to control in closed loop the trajectories of the cells in real time during their journey in the chips dedicated to sorting. In this context, my thesis work focuses on an innovative method for real-time detection of cell position, directly integrated into chips and based on the physical principle of impedance measurement.When a cell passes through a microchannel, it changes the measurable impedance between electrodes placed on the edges of the channel.A generic method for formulating impedance variations as a function of particle position has been proposed (direct model). A method for estimating the position of a particle based on impedance measurements has also been proposed (inverse model). It uses an extended Kalman filter, allowing data from several pairs ofelectrodes to be merged, and exploiting available information such as measurement noise distribution and particle dynamic model. The validation of the method was carried out on an experimental bench that was developed during this thesis and on numerical simulations.This work shows the relevance of using impedance measurement to build a position sensor for particles immersed in a microchannel. This method is an alternative to the use of optical microscopes and has the advantage of being very compact.

Real-time, virtual and mixed reality systems have diverse uses for real-world data visualization, representation, and remote collaboration in distant learning settings, especially in universities. Design of such systems involves challenges in mapping the real world data and physical world structure accurately to digital form of physical space, also called as virtual models. Researchers have created similar systems using multiple cameras, stereo cameras, accelerometers, and motion detectors. This report presents a platform to detect and track real-time locations of people present in buildings and map their location information into virtual models as avatars using omni-directional cameras installed in the physical space. These models were created as part of the Mirror Worlds project. The project infrastructure is funded by National Science Foundation. This infrastructure enables users to connect virtual and physical aspects of the environment through a coordinate-based data networking system to enable interaction with the rest of the system including environment objects and other users. This is an interdisciplinary project where students from various departments have worked on the development of virtual model of the Moss Art Center and Torgersen Hall in Unity / X3D. Some students from the Department of Computer Science have developed a coordinate-based data networking system. The prototype of a detection and tracking algorithm to extract the location information was developed using background subtraction in MATLAB. The proposed approach was developed using the combination of background subtraction and neural networks along with heuristics based on spatial information about the physical space. The system was scaled to work across multiple buildings, extract the location information of people present in the physical space, and map location information into shared virtual space as an avatar. The concept of remote presence was extended to create a collaborative object manipulation application using Leap Motion controller. Effects of fidelity were evaluated to perform the collaborative object manipulation task in shared virtual space based on user study conducted for this application. Since no annotated people video dataset is publicly available with overhead view from omni-directional cameras, three videos were annotated manually to test the performance of the approach. The current approach almost works at near real-time rates. All three video sequences were evaluated to compute frame based detection accuracy. Precision and recall obtained for the first video sequence of people detection is 93.85% and 95.06% respectively. / Master of Science

Real-time lane detection or localization is a crucial problem in modern Advanced Driver Assistance Systems (ADAS), especially in Automated Driving System. This thesis estimates the possibility to implement a lane detection system in the available low-power embedded hardware. Various state-of-the-art Lane Detection algorithms are assessed based on a number of proposed criteria. From the result of the evaluation, three different algorithms are constructed and implemented in the hardware using OpenCV library. The lane detection stage is done with different methods: using Hough Transform for line detection or randomly sampling hypotheses which are straight lines or cubic splines over the pre-processed binary image. Weights of the hypotheses are then calculated based on their positions in the image. The hypothesis which has highest weight or best position will be chosen to represent lane marking. To increase the performance of the system, tracking stage is introduced with the help of Particle Filter or Kalman Filter. The systems are then tested with several different datasets to evaluate the speed, performance and ability to work in real-time. In addition, the system interfaces with CAN bus using a CAN interface, so that the output data can be sent as messages via the CAN bus to other systems.

Αντικείμενο της παρούσας εργασίας είναι η υψομετρική ανίχνευση θέσης κινούμενου ρομπότ. Για τον υπολογισμό της θέσης του ρομπότ χρησιμοποιήθηκε ένα στερεοσκοπικό σύστημα δύο ίδιων παράλληλων καμερών. Οι εσωτερικές παράμετροι των δύο καμερών και το μήκος της βασικής γραμμής είναι γνωστά. Θεωρείται για ευκολία ότι όλος ο όγκος του ρομπότ αναπαριστάται απο ένα σημείο P με συντεταγμένες (X,Y,Z) ώς προς σύστημα συντεταγμένων με αρχή το κέντρο προβολής της αριστερής κάμερας. Για τον εντοπισμό του ρομπότ αποκτήθηκαν δύο ακολουθίες βίντεο, μια για κάθε κάμερα, στις οποίες έγινε κατάτμηση κίνησης με την μέθοδο διαφοράς διαδοχικών καρέ. Στη συνέχεια ακολούθησε επεξεργασία εικόνας στα καρέ και των δύο ακολουθιών για την εξαγωγή των περιοχών που απεικονίζουν το αντικείμενο στο επίπεδο της εικόνας. Κατόπιν, υπολογίστηκαν οι θέσεις των κέντρων βάρους των περιοχών για τις δύο ακολουθίες , Pl(xl,yl) και Pr(xr,yr) αντίστοιχα. Τα σημεία αυτά αναπαριστούν το σημείο P στην αριστερή και στη δεξιά ακολουθία εικόνων. Ακολούθησε αντιστοίχιση σημείων μεταξύ των δύο ακολουθιών με βάση την επιπολική γεωμετρία. Τα σημεία στα οποία βρέθηκε αντιστοιχία χρησιμοποιήθηκαν για τον υπολογισμό της θέσης του σημείου P με την μέθοδο της τριγωνοποίησης. / The object of this thesis is the position detection of a moving robot. To calculate the position of the robot a stereoscopic system of two identical parallel cameras was used. The intrinsic parameters of the cameras and the length of the baseline are given. To simplify the problem it is assumed that the entire robot can be represented by a point P with (X,Y,Z) coordinates with respect to the coordinate system of the left camera center of projection. To detect the robot two video image sequences were acquired, one for each camera, to which motion segmentation was applied with the frame differencing method. After that, image proseccing was implemented to the frame sequences in order to find the region that represents the robot in the image plane. The centroids of the regions were calculated in the two sequences,Pl(xl,yl) and Pr(xr,yr) respectively. Those points represent the point P in the image planes of the left and right camera. Finally the restriction of epipole geometry was used to find matches between those points. Using those matches the coordinates of P were calculated with the triangulation method.

L’expérience Arc-En-Ciel permet d’étudier la dynamique de photofragmentation UV de biomolécules produites par une source « électrospray ». La spécificité du dispositif expérimental utilisé repose sur la détection en coïncidence des photo-fragments ioniques et neutres issus d’un même évènement physique de fragmentation. L’étude de molécules simplement chargées permet d’identifier chaque canal de fragmentation par la masse du fragment ionique émis. En corrélant les informations temporelles et spatiales des photo-fragments détectés, on définit :- le nombre et la masse des fragments neutres associés à chaque fragment ionique- le nombre d’étapes de fragmentation de chaque canal et leurs temps caractéristiques(20 ns ≤ τ < 1 μs).L’ensemble de ces informations permet une description complète de la dynamique de photofragmentation du système étudié.La dynamique de photofragmentation du tryptophane protoné est régie par des transferts concertés d’électron et de proton à l’état excité. Lorsque le tryptophane protoné est complexé à un éther-couronne, les transferts de protons sont inhibés. Nous observons alors une modification de la dynamique de fragmentation.Pour de petits peptides protonés contenant le tryptophane, la dynamique à l’état excité est gouvernée par la position du tryptophane dans la chaîne peptidique. Les voies de fragmentation spécifiques UV, mises en évidence pour ces peptides, sont expliquées par les mêmes mécanismes de transfert concerté d’électron et de proton. Nous montrons cependant que ces mécanismes diffèrent suivant la composition du peptide. / The Arc-En-Ciel experiment allows the investigation of UV photo-fragmentation dynamics of protonated biomolecules produced by an electrospray ion source. The specificity of the set-up is based on the detection in coincidence of ionic and neutral photo-fragments coming from the same fragmentation event. The study of simple charged molecules allows the identification of each fragmentation channel by the mass of the emitted ionic fragment. With the time and spatial correlation of the information of detected photo-fragments we identify:- the number of neutral fragments as well as their masses associated with each ionic fragment- the number of fragmentation steps of each channel as well as their fragmentation times (20 ns ≤ τ < 1 μs)This information provides a comprehensive understanding of the photo-fragmentation dynamics.The photo-fragmentation dynamics of protonated Tryptophan is driven by concerted electron and proton transfers in the excited state. When protonated Tryptophan is complexed witha crown-ether, proton transfers are inhibited and dynamics is modified.The excited state dynamics of small protonated peptides containing Tryptophan is governed by the position of Tryptophan in the peptide chain. The specific fragmentation channels involved are explained by concerted electron and proton transfers. We show how these mechanisms change with the composition of peptides.

Real-time lane detection or localization is a crucial problem in modern Advanced Driver Assistance Systems (ADAS), especially in Automated Driving System. This thesis estimates the possibility to implement a lane detection system in the available low-power embedded hardware. Various state-of-the-art Lane Detection algorithms are assessed based on a number of proposed criteria. From the result of the evaluation, three different algorithms are constructed and implemented in the hardware using OpenCV library. The lane detection stage is done with different methods: using Hough Transform for line detection or randomly sampling hypotheses which are straight lines or cubic splines over the pre-processed binary image. Weights of the hypotheses are then calculated based on their positions in the image. The hypothesis which has highest weight or best position will be chosen to represent lane marking. To increase the performance of the system, tracking stage is introduced with the help of Particle Filter or Kalman Filter. The systems are then tested with several different datasets to evaluate the speed, performance and ability to work in real-time. In addition, the system interfaces with CAN bus using a CAN interface, so that the output data can be sent as messages via the CAN bus to other systems.

Optical microcavities (OMs) are receiving increasing attention owing to their potential applications ranging from cavity quantum electrodynamics, optical detection to photonic devices. Recently, rolled-up structures have been demonstrated as OMs which have gained considerable attention owing to their excellent customizability. To fully exploit this customizability, asymmetric and topological rolled-up OMs are proposed and investigated in addition to conventional rolled-up OMs in this thesis. By doing so, novel phenomena and applications are demonstrated in OMs. The fabrication of conventional rolled-up OMs is presented in details. Then, dynamic mode tuning by a near-field probe is performed on a conventional rolled-up OM. Next, mode splitting in rolled-up OMs is investigated. The effect of single nanoparticles on mode splitting in a rolled-up OM is studied. Because of a non-synchronized oscillating shift for different azimuthal split modes induced by a single nanoparticle at different positions, the position of the nanoparticle can be determined on the rolled-up OM. Moreover, asymmetric rolled-up OMs are fabricated for the purpose of introducing coupling between spin and orbital angular momenta (SOC) of light into OMs. Elliptically polarized modes are observed due to the SOC of light. Modes with an elliptical polarization can also be modeled as coupling between the linearly polarized TE and TM mode in asymmetric rolled-up OMs. Furthermore, by adding a helical geometry to rolled-up structures, Berry phase of light is introduced into OMs. A -π Berry phase is generated for light in topological rolled-up OMs so that modes have a half-integer number of wavelengths. In order to obtain a deeper understanding for existing rolled-up OMs and to develop the new type of rolled-up OMs, complete theoretical models are also presented in this thesis.

The aim of this diploma thesis is to implement a robotic air hockey table in cooperation with the work of Dominik Jašek [37] . Specifically, the work deals with the implementation of simulation software that models behavior of robotic air hockey table, designing a game strategy for the robot and detecting position of a puck situated on the playing field. In addition, the work was extended by integrating said software modules under one control application and creating a user interface for controlling and managing all the functionalities of the robotic table through a capacitive touch screen.