Dissertations / Theses on the topic 'Inertial and magnetic sensors'

Positioning systems for trains are traditionally based on track-side infrastructure, implying costs for both installation and maintenance. A reliable on-board system would therefore be attractive. Sufficient reliability for on-board systems is likely going to require a multi-sensor solution. This thesis investigates how measurements from bogie-mounted inertial and magnetic sensors can contribute to such a system. The first part introduces and compares two different methods for estimating the speed. The first one estimates the fundamental frequency of the variations in the magnetic field, and the second one analyses the mechanical vibrations using the accelerometer and gyro, where one mode is due to the wheel irregularities. The second part introduces and evaluates a method for train localization using magnetic signatures. The method is evaluated both as a solution for localization along a given track and at switchways. Overall, the results in both parts show that bogie-mounted inertial and magnetic sensors provide accurate estimates of both speed (within 0.5 m/s typically) and location (3-5 m accuracy typically).

Magnetic interference in the motion capture environment is caused primarily by ferromagnetic objects and current-carrying devices disturbing the ambient, geomagnetic field. Inertial sensors gather magnetic data to determine and stabilize their global heading estimates, and such magnetic field disturbances alter heading estimates. This decreases orientation accuracy and therefore decreases motion capture accuracy. The often used Kalman Filter approach deals with magnetic interference by ignoring the magnetic data during periods interference is encountered, but this method is only effective when the disturbances are ephemeral, and cannot not retroactively repair data from disturbed time periods. The objective of this research is to develop a method of magnetic interference mitigation for environments where magnetic interference is the norm rather than the exception. To the knowledge of this author, the ability to use inertial and magnetic sensors to capture accurate, global, and drift-free orientation data in magnetically disturbed areas has yet to be developed. Furthermore there are no methods known to this author that are able to use data from undisturbed time periods to retroactively repair data from disturbed time periods. The investigation begins by exploring the use of magnetic shielding, with the reasoning that application of shielding so as to impede disturbed fields from affecting the inertial sensors would increase orientation accuracy. It was concluded that while shielding can mitigate the effect of magnetic interference, its application requires a tedious trial and error testing that was not guaranteed to improve results. Furthermore, shielding works by redirecting magnetic field lines, increasing field complexity, and thus has a high potential to exacerbate magnetic interference. Shielding was determined to be an impractical approach, and development of a magnetic inference mitigation algorithm began. The algorithm was constructed such that magnetic data would be filtered before inclusion in the orientation estimate, with the result that exposure in an undisturbed environment would improve estimation, but exposure to a disturbed environment would have no effect. The algorithm was designed for post-processing, rather than real-time use as Kalman Filters are, which enabled magnetic data gathered before and after a time point could affect estimation. The algorithm was evaluated by comparing it with the Kalman Filter approach of the company XSENS, using the gold standard of optical motion capture as the reference point. Under the tested conditions of stationary periods and smooth planar motion, the developed algorithm was resistant to magnetic interference for the duration of testing, while the Kalman Filter began to degrade after approximately 15 seconds. In a 190 second test, of which 180 were spent in a disturbed environment, the developed algorithm resulted in 0.4 degrees of absolute error, compared to the of the Kalman Filter’s 78.8 degrees. The developed algorithm shows the potential for inertial systems to be used effectively in situations of consistent magnetic interference. As the benefits of inertial motion capture make it a more attractive option than optical motion capture, immunity to magnetic interference significantly expands the usable range of motion capture environments. Such expansion would be beneficial for motion capture studies as a whole, allowing for the cheaper, more practical inertial approach to motion capture to supplant the more expensive and time consuming optimal option.

Wearable inertial and magnetic measurements units (IMMU) are an important tool for underwater motion analysis because they are swimmer-centric, they require only simple measurement set-up and they provide the performance results very quickly. In order to estimate 3D joint kinematics during motion, protocols were developed to transpose the IMMU orientation estimation to a biomechanical model. The aim of the thesis was to validate a protocol originally propositioned to estimate the joint angles of the upper limbs during one-degree-of-freedom movements in dry settings and herein modified to perform 3D kinematics analysis of shoulders, elbows and wrists during swimming. Eight high-level swimmers were assessed in the laboratory by means of an IMMU while simulating the front crawl and breaststroke movements. A stereo-photogrammetric system (SPS) was used as reference. The joint angles (in degrees) of the shoulders (flexion-extension, abduction-adduction and internal-external rotation), the elbows (flexion-extension and pronation-supination), and the wrists (flexion-extension and radial-ulnar deviation) were estimated with the two systems and compared by means of root mean square errors (RMSE), relative RMSE, Pearson’s product-moment coefficient correlation (R) and coefficient of multiple correlation (CMC). Subsequently, the athletes were assessed during pool swimming trials through the IMMU. Considering both swim styles and all joint degrees of freedom modeled, the comparison between the IMMU and the SPS showed median values of RMSE lower than 8°, representing 10% of overall joint range of motion, high median values of CMC (0.97) and R (0.96). These findings suggest that the protocol accurately estimated the 3D orientation of the shoulders, elbows and wrists joint during swimming with accuracy adequate for the purposes of research. In conclusion, the proposed method to evaluate the 3D joint kinematics through IMMU was revealed to be a useful tool for both sport and clinical contexts.

Questo progetto di tesi fa parte di un più ampio studio clinico condotto all’interno dell’azienda NCS Lab (Carpi,(MO)), in collaborazione con il Dr. Claudio Chillemi (ICOT, Latina (RM)) che mira ad eseguire un confronto tra diverse tecniche chirurgiche per la riparazione della cuffia dei rotatori. Lo studio clinico in questione durerà circa due anni: per questo motivo i dati analizzati in questo progetto di tesi provengono solo dal gruppo di pazienti acquisiti nella fase preoperatoria. Tutti i dati sono stati acquisiti utilizzando i sensori magneto-inerziali WISE (tecnologia proprietaria dell’azienda NCS Lab). Questo lavoro di tesi si propone, quindi, di valutare la ripetibilità del movimento in termini di coefficiente di correlazione multipla e di estrapolare alcuni parametri di interesse clinico come, ad esempio, i range di movimento (ROM) della scapola e dell’omero e il ritmo scapolo-omerale (SHR). Questi parametri sono stati poi caratterizzati da un punto di vista statistico al fine di valutare le differenze tra arto patologico e controlaterale. Sono state calcolate, inoltre, le prediction bands con lo scopo di descrivere le differenze tra arto patologico e controlaterale nella coordinazione scapolo-omerale dei pazienti. Per quanto riguarda la ripetibilità del movimento, i risultati ottenuti in questo lavoro di tesi mostrano che la rotazione medio-laterale è caratterizzata da un eccellente CMC sia per l'arto patologico che per il controlaterale. Inoltre, sono state riscontrate differenze significative dal punto di vista statistico tra le distribuzioni dei range di movimento dell'arto patologico e controlaterale. Tali differenze sono state trovate anche per quanto riguarda il ritmo scapolo-omerale.

Accurate risk assessment tools and methods are necessary to understand the relationship between occupational exposure to physical risk factors and musculoskeletal disorders. Ergonomists typically consider direct measurement methods to be the most objective and accurate of the available tools. However, direct measurement methods are often not used due to cost, practicality, and worker/workplace disruption. Inertial measurement units (IMUs), a relatively new direct measurement technology used to assess worker kinematics, are attractive to ergonomists due to their small size, low cost, and ability to reliably capture information across full working shifts. IMUs are often touted as a field-capable alternative to optical motion capture systems (OMCs). The error magnitudes of IMUs, however, can vary significantly (>15°) both within and across studies. The overall goals of this thesis were to (i) provide knowledge about the capabilities and limitations of IMUs in order to explain the inconsistencies observed in previous studies that assessed IMU accuracy, and (ii) provide guidance for the ergonomics community to leverage this technology. All three studies in this dissertation systematically evaluated IMUs using a repetitive material transfer task performed by thirteen participants with varying movement speeds (15, 30, 45 cycles/minute) and magnetic disturbance (absent, present). An OMC system was used as the reference device. This first study systematically evaluated the effects of motion speed and magnetic disturbance on the spatial orientation accuracy of an inertial measurement unit (IMU) worn on the hand. Root-mean-square differences (RMSD) exceeded 20° when inclination measurements (pitch and roll) were calculated using the IMU’s accelerometer. A linear Kalman filter and a proprietary, embedded Kalman filter reduced inclination RMSD to < 3° across all movement speeds. The RMSD in the heading direction (i.e., about gravity) increased (from < 5° to 17°) under magnetic disturbance. The linear Kalman filter and the embedded Kalman filter reduced heading RMSD to < 12° and < 7°, respectively. This study indicated that the use of IMUs and Kalman filters can improve inclinometer measurement accuracy. However, magnetic disturbances continue to limit the accuracy of three-dimensional IMU motion capture. The goal of the second study was to understand the capability of IMU inclinometers to improve estimates of angular displacements and velocities of the upper arm. RMSD and peak displacement error exceeded 11° and 28° at the fastest transfer rate (45 cycles/min) when upper arm elevation was calculated using the IMU accelerometer. The implementation of a Kalman filter reduced RMS and peak errors to < 1.5° and < 2.3°, respectively. Similarly, the RMS and peak error for accelerometer-derived velocities exceeded 81°/s and 221.3°/s, respectively, at the fastest transfer rate. The Kalman filter reduced RMS and peak errors to < 9.2°/s and < 25.1°/s, respectively. The third study was conducted to evaluate the relationship between magnetic field strength variation and magnetic heading deviation. In this study, the presence of the metal plate increased magnetic heading deviations from < 12° (90th-10th percentile) to approximately 30°. As expected, the magnetic field strength standard deviation increased from 1.0uT to 2.4uT. While this relationship may differ across other sources of magnetic disturbance, the results reinforce the notion that local magnetic field disturbances should be minimized when using IMUs for human motion capture. Overall, the findings from this thesis contribute to the ergonomics community’s understanding of the current capabilities and limitations of IMUs. These studies suggest that while the touted capabilities of the IMUs (full-body motion capture in workplace settings) may be unattainable based on current sensor technology, these sensors are still significantly more accurate than the accelerometer-based inclinometers commonly used by ergonomists to measure motions of the upper arms.

Dans cette thèse, nous étudions la propagation des atomes dans un guide magnétique toroïdal, dans le but de développer un capteur inertiel. Ici, nous présentons différentes stratégies pour créer un guide sur une puce atomique pour un interférometre Sagnac atomique guidé. Nous avons mis au point trois solutions qui peuvent être realisé avec la même configuration des fils. Ils utilise la technique de modulation de courant avec un nouveau point de vue qui traite simultanément la problème de rugosité des fils et les pertes de Majorana dépendant du spin. L'effect de la propagation multimode des atomes dan le guide est aussi quantifié dans cette thèse. En utilisant un modèle simple, nous avons couvert les cas de la propagation de gaz non interactif ultra froids et thermique. Nous avons identifié les conditions operationelles pour realiser un interferometre à atomes froids avec une grande gamme dynamique, essentielle pour les application en navigation inertielle. Expérimentalement, cette thèse decrit la réalisation et la characterisation de la source atomes froids proche d'un substrat avec un dépôt d'or, ainsi que l'implémentation et la caracterisation des systèmes de détection<br>In this thesis we study the propagation of atoms in a magnetic toroidal waveguide, with the aim of developing an inertial sensor. Here, we present different strategies to create the waveguide on an atom chip for a guided Sagnac atom interferometer. We devised three solutions which can be achieved using the same wire configuration. They use the current modulation technique, from a new point of view, which simultaneously tackles the problem of wire corrugation and spin dependent Majorana atom losses. The effect of the multimode propagation of the atoms in the guide is also quantified in this thesis. Using a simple model, we covered the propagation of noninteracting ultracold and thermal gases. We identified the operating conditions to realize a cold atom interferometer with a large dynamic range essential for applications in inertial navigation. Experimentally, the thesis describes the realisation and characterisation of the cold atom source close to a gold coated substrate, as well as the implementation and the characterisation of the atom detection systems

Thesis (M.S. in Electrical Engineering)--Naval Postgraduate School, June 2010.<br>Thesis Advisor(s): Yun, Xiaoping ; Second Reader: Romano, Marcello. "June 2010." Description based on title screen as viewed on July 14, 2010. Author(s) subject terms: micro-electro-mechanical systems, MEMS, magnetic, angular rate, gravity sensor, MARG sensors, inertial navigation system, INS, inertial test, MicroStrain, 3DM-GX1, 3DMGX3, CompactRIO, MATLAB GUI, dynamic accuracy test. Includes bibliographical references (p. 187-189). Also available in print.

Approved for public release; distribution is unlimited.<br>This dissertation describes the development of an integrated locomotion interface for building self-contained, portable, and immersive virtual environment (VE) systems. Such VE systems do not rely on any infrastructure support and can be used in indoor/outdoor open spaces. The natural walking motions of the user are utilized as a means of signal generation to drive the locomotion interface, which provides the user with a higher sense of presence. This work investigates the use of two types of measurement systems, the inertial/magnetic measurement units and the ranging measurement systems, to develop a locomotion interface for portable VE systems. Algorithms were developed for each of the two systems to provide the necessary functionalities of the desired locomotion interface. Fusing measurements from a head-mounted and a foot-mounted inertial/magnetic sensor, a locomotion interface was developed for allowing the use of natural walking motions to navigate through virtual environments. To prevent collisions with physical environment boundaries such as walls, a ranging measurement system was used to detect the presence of obstacles. An improved Iterative Closest Point (ICP) algorithm was developed for map-building of the physical environment and for estimating the user's orientation and position within the map. A redirected-walking mechanism was utilized for redirecting the user's walking direction away from boundaries in the physical environment. The two types of measurement systems were integrated to constitute a novel locomotion interface for portable VE systems, and its effectiveness was experimentally tested and demonstrated.

Le développement croissant d'objets intelligents offre de nouvelles opportunités de localisation du voyageur connecté. Cependant, le suivi de la trajectoire du piéton reste problématique et les applications de navigation ne proposent pas de suivre la trajectoire du voyageur à l’échelle multimodale de façon autonome. Ce travail s’intéresse à la mise en place d’une solution unique capable de localiser l’utilisateur selon différents mode de déplacement et quel que soit l’environnement, à partir de capteurs inertiels, magnétique et GNSS. Dans un premier temps, une nouvelle méthode de localisation du cycliste est mise en place. Les mesures de phases GNSS sont utilisées pour corriger le vecteur vitesse par différences temporelles et la direction de déplacement est contrainte à l'aide des signaux inertiels. Ces éléments ont été utilisés dans un second temps et adaptés pour mettre en place une nouvelle méthode de localisation du piéton avec un capteur en main. L’approche PDR qui est une technique de navigation inertielle à l’estime est paramétrée dans un filtre de Kalman étendu. Une mise à jour innovante fusionnant l’estimation de l’attitude du boîtier et une estimation statistique de la direction de marche permet de corriger l’estimation du cap de marche et obtenir une estimation cohérente et lissée. Les mesures GNSS sont utilisées pour corriger le vecteur vitesse, l’orientation, la longueur de pas et la position absolue. Enfin, une approche multimodale est proposée et la gestion des transitions entre les différents algorithmes, assistée par l’utilisation d’un capteur innovant, est étudiée. Des validations expérimentales multimodales en conditions réelles sont conduites pour analyser les performances d’estimation de la solution proposée<br>The growing development of smart objects offers new opportunities for locating the connected traveller. However, tracking the trajectory of the pedestrian remains problematic and navigation applications do not offer to track the traveller's trajectory on a multimodal level in an autonomous way. This work focuses on the implementation of a single solution able to locate the user according to different travel modes and whatever the environment, using inertial, magnetic and GNSS sensors. In a first step, a new method for locating the cyclist is implemented. GNSS phase measurements are used to correct the velocity vector by time differences and the motion direction is constrained using inertial signals. These elements were used in a second step and adapted to implement a new method of pedestrian localization with a handheld sensor. The PDR approach, which is an inertial dead reckoning navigation technique, is parameterized in an extended Kalman filter. An innovative update merging the device attitude estimation and a statistical estimation of the walking direction allows to correct the walking heading estimation and obtain a consistent and smoothed estimation. GNSS measurements are used to correct speed vector, orientation, step length and absolute position. Finally, a multimodal approach is proposed and the management of transitions between the different algorithms, assisted by the use of an innovative sensor, is studied. Multimodal experimental validations in real conditions are conducted to analyze the estimation performances of the proposed solution

Dans ce travail de thèse on s’intéresse à l’estimation de l’attitude d’un corps rigideen mouvement dans l’espace 3D en utilisant les quaternions comme représentation. Cetteproblématique a été largement étudiée dans la littérature sous divers domaines d’application.L’objectif de la thèse est de proposer de nouvelles méthodes de fusion de données en combinantdes mesures inertielles et magnétiques. Dans un premier temps, nous nous sommesintéressés à l’estimation de l’attitude en cas de mouvement accéléré où l’accélération linéairedu corps n’est plus négligeable devant la gravité. Deux approches ont été proposées dans cecadre. La première utilise un filtre de Kalman adaptatif pour la compensation des accélérationslinéaires. Précisément, des lois de détection ont été développées pour distinguer d’unefaçon automatique les différentes phases de mouvement (statiques et dynamiques). Ainsi, lamatrice de covariance associée à l’accélération linéaire est estimée afin d’ajuster le gain dufiltre. La deuxième approche consiste à intégrer un filtre singulier élaboré sur la base d’unnouveau modèle, dans lequel le modèle du processus est défini en se basant sur les mesuresissues de l’accéléromètre tandis que le modèle d’observation est défini par les mesures issuesdu gyromètres et du magnétomètres. Cette formulation permet de prendre en compte l’effetdes accélérations linéaires d’une manière efficace. Dans un deuxième temps, on s’est focalisésur l’estimation de l’attitude avec utilisation intermittente de gyromètres, considérés commecapteurs énergivores. Nous avons étudié dans ce cas la façon la plus adéquate afin de réduirel’acquisition des mesures de vitesse angulaire tout en gardant une qualité acceptable de l’estimationde l’attitude. Toutes les approches développées ont été validées par des simulationsnumériques ainsi que des expérimentations utilisant des données réelles<br>In this PhD. thesis we deal with attitude estimation of accelerated rigid body moving in the 3D space using quaternion parameterization. This problem has been widely studied in the literature in various application areas. The main objective of the thesis is to propose new methods for data fusion to combine inertial gyros) and magnetic measurements. The first challenge concerns the attitude estimation during dynamic cases, in which external acceleration of the body is not negligible compared to the Gravity. Two main approaches are proposed in this context. Firstly, a quatenion-based adaptive Kalman filter (q-AKF) was designed in order to compensate for such external acceleration. Precisely, a smart detector is designed to decide whether the body is in static or dynamic case. Then, the covariance matrix of the external acceleration is estimated to tune the filter gain. Second, we developed descriptor filter based on a new formulation of the dynamic model where the process model is fed by accelerometer measurements while observation model is fed by gyros and magnetometer measurements. Such modeling gives rise to a descriptor system. The resulting model allows taking the external acceleration of the body into account in a very efficient way. The second challenge is related to the energy consumption issue of gyroscope, considered as the most power consuming sensor. We study the way to reduce the gyro measurements acquisition by switching on/off the sensor while maintaining an acceptable attitude estimation. The effciency of the proposed methods is evaluated by means of numerical simulations and experimental tests

Es ampliamente conocido que los modelos de error para sensores inerciales tienen dos componentes: El primero es un componente determinista que normalmente es calibrado por el fabricante en el firmware de la IMU. El segundo es un componente aleatorio que es caracterizado a través de un modelo estocástico que tiene su impacto en la solución de navegación calculada con las observaciones no procesadas obtenidas con dichos sensores. La caracterización de este comportamiento estocástico se apoya en el análisis de observaciones adquiridas en condiciones estáticas. El sensor inercial es aislado de cualquier perturbación externa y se llevan a cabo adquisiciones estáticas de larga duración. Estas observaciones se evalúan a través de herramientas de análisis como la varianza de Allan. Esta herramienta permite caracterizar el modelo estocástico que mejor se ajusta a las observaciones, permitiendo obtener parámetros como el ruido de observación o el bias del sensor, los factores de escala, etcétera. Existen referencias que indican que la caracterización de sensores en condiciones estáticas podría ser imprecisa cuando el sensor está funcionando en entornos no estáticos. Esta es la principal motivación de esta tesis doctoral. De cara a entender la variación del error del sensor con la dinámica aplicada, una serie de campañas experimentales (en laboratorio y en vehículos) fueron llevadas a cabo. Estos experimentos consistieron en un conjunto de IMUs de prueba acopladas a una plataforma rígida a la que a su vez se integraba una IMU de grado de navegación que se utilizaba como referencia. Primero, se debe resolver el alineamiento entre las IMUs de prueba y la IMU de referencia. Un método para calcular dicho alineamiento es presentado en esta tesis. Una vez que la IMU de referencia está correctamente alienada con las observaciones de la IMU en pruebas, y asumiendo que las observaciones de la IMU de referencia son ideales (no tienen errores), las observaciones de ambas IMUS son comparadas. Las diferencias entre ambos conjuntos de observaciones se pueden considerar que son los errores de la IMU en pruebas. El análisis de la varianza de Allan es entonces aplicada a estas diferencias. Se muestra que debido a las limitaciones de esta técnica de análisis y a la longitud del conjunto de datos, la única medida fiable que se puede obtener con la varianza de Allan es el ruido de observación (que normalmente sigue un modelo estocástico de ruido blanco). Por esta razón, esta tesis propone una aproximación diferente al problema. Se demuestra que existe una conexión entre el error del sensor y la dinámica aplicada a dicho sensor. Esta dinámica está caracterizada por las medidas directas del sensor y las derivadas enésimas de de dichas medidas. Por tanto, gracias a esta nueva aproximación, un nuevo modelo de error inercial es mostrado. Este nuevo modelo está compuesto por un sesgo, un factor de escala que afecta a la magnitud medida por el sensor y una serie de coeficientes que multiplican a las enésimas derivadas de las medidas del sensor. Esta tesis establece dos maneras que este modelo se puede usar para determinar estos coeficientes. Si existe un sensor de referencia, el modelo se puede implementar en un ajuste de mínimos cuadrados para estimar dichos coeficientes. El resultado de este ajuste es un conjunto de observaciones “mejoradas” del sensor inercial que pueden procesarse en un navegador INS/GNSS convencional. En caso de no existir el sensor de referencia, la otra opción para determinar los coeficientes es a través de un filtro de Kalman extendido que incorpora dichos coeficientes. En esta tesis, dichos coeficientes fueron implementados y testeados en un navegador INS/GNSS con arquitectura “loosely coupled”. Por lo que conoce el autor de esta tesis, la utilización de las derivadas de las observaciones nunca se han investigado y se demuestra que ayudan a mejorar la precisión de la trayectoria suministrada por el navegador INS/GNSS.<br>The inertial sensor error model is widely known to have two components. The first is a deterministic component that is usually calibrated by the manufacturer in the IMU integrated firmware and, secondly, there is a random component that is characterized through a stochastic model that has its impact on the navigation solution computed with the raw data obtained with these sensors. The characterization of this stochastic behavior relies on the analysis of datasets acquired under static conditions. The inertial sensor is isolated from any external perturbation and long periods of static acquisitions are conducted. These datasets are then evaluated through complex tools like the Allan variance analysis. This analysis tool lets the stochastic model be determined that best fits with the sensor observation noise, the sensor bias scale factors and so forth. There have been reports that these sensor characterization models start to become inaccurate when the sensor is submitted to non-static environments. This is the main motivation for this PhD thesis. In order to understand the sensor error variation with dynamics, a series of laboratory and vehicular experiments were performed, using a set of IMUs rigidly attached to a navigation grade reference IMU. First, the reference IMU misalignment must be resolved, with respect to the IMU under test (IUT) that was selected for this study. A method for resolving the issue is presented in this thesis. Once the reference IMU data is correctly aligned with the test IMU observations, and assuming that the reference IMU measurements are error-free, observations of the two sensors are compared. The difference between both observations is the error of the sensor from the IMU under test. A classic Allan variance analysis is then applied to this varying error. It is shown that due to the limitations of the Allan variance tool and the length of the dataset, the only reliable measurement that can be obtained by employing the Allan variance analysis is the observation of white noise. For this reason, a different approach is presented in this thesis. A connection is shown between the sensor error and the dynamic applied to the sensor. This dynamic is characterized by the sensor raw measurement and the nth order derivatives of these measurements. Therefore, thanks to this new approach, a new sensor error model is presented. This sensor model is composed of a bias and a scale factor that affects the actual sensed magnitude and a series of nth order derivatives of the sensed magnitude that are multiplied by a coefficient. This thesis sets out two ways that this model can be used to determine these coefficients. If there is a reference sensor, it can be used in a least squares adjustment to estimate these coefficients. The result of this adjustment is "improved" sensor observation that can be processed via an ordinary INS/GNSS navigator. Another way to determine these coefficients is by implementing this new sensor model in an extended Kalman filter. In the case of this thesis, it was implemented and tested in a loosely coupled INS/GNSS. To the knowledge of the author, the utilization of sensor observation derivatives has never been researched and it is demonstrated that it helps to improve the precision of the trajectory provided by the INS/GNSS navigator.<br>Es àmpliament conegut que els models d’error per a sensors inercials tenen dues components. La primera es una component determinista que normalment es calibrada pel fabricant en el firmware de la IMU. El segon es una component aleatòria que es caracteritzada a traves d’un model estocàstic que te el seu impacte en la solució de navegació calculada amb les observacions no processades obtingudes amb aquests sensors. La caracterització d’aquest comportament estocàstic es recolza en la anàlisi d’observacions adquirides en condicions estàtiques. El sensor inercial es aïllat de qualsevol pertorbació externa i es porten a terme adquisicions estàtiques de llarga durada. Aquestes observacions s’avaluen a través d’eines d’anàlisi com la variança d’Allan. Aquesta eina d’anàlisi permet caracteritzar el model estocàstic que millor s’ajusta a les observacions, permetent obtenir paràmetres com el soroll d’observació, el biaix del sensor, els factors d’escala, etcètera. N’hi han referències que indiquen que la caracterització de sensors en condicions estàtiques podria ser imprecisa quan el sensor està funcionant en entorns no estàtics. Aquesta es la principal motivació d’aquesta tesi doctoral. De cara a entendre la variació del error del sensor amb una dinàmica aplicada, una sèrie de campanyes de mesura experimentals (en laboratori i vehicles) van ser portades a terme. Aquests experiments van consistir en un conjunt d’IMUs de prova acoblades a una plataforma rígida que a la seva vegada integrava una IMU de grau de navegació que s’utilitzava com a referència. Primer de tot, s’ha de resoldre l’alineament entre les IMUs de prova i la IMU de referència. Un mètode per a calcular aquest alineament es presentat en aquesta tesi. Una vegada que la IMU de referència està correctament alineada amb les observacions de la IMU en proves, i assumint que les observacions de la IMU de referència son ideals (no tenen errors), les observacions de ambdues IMUs son comparades. Les diferencies entre els dos conjunts d’observacions es poden considerar que son els errors de la IMU de proves. L’anàlisi de la variança d’Allan es llavors aplicat a aquestes diferències. Es mostra que degut a les limitacions d’aquesta tècnica d’anàlisi i a la durada del conjunt de dades, l’única mesura fiable que es pot obtenir amb la variança d’Allan es el soroll d’observació (que normalment segueix un model estocàstic de soroll blanc). Per aquesta raó, aquesta tesi proposa una aproximació diferent al problema. Es demostra que existeix una connexió entre el error del sensor i la dinàmica aplicada a aquest sensor. Aquesta dinàmica està caracteritzada per les mesures directes del sensors i les derivades enèsimes d’aquestes mesures. Per tant, gràcies a aquesta nova aproximació, un nou model d’error inercial es mostrat. Aquest nou model està composat per un biaix, un factor d’escala que afecta a la magnitud mesurada pel sensor i una sèrie de coeficients que multipliquen a les enèsimes derivades de les mesures del sensor. Aquesta tesi estableix dues maneres que aquest model es pugui utilitzar per a determinar aquests coeficients. Si existeix un sensor de referència, el model es pot implementar en un ajust de mínims quadrats per a estimar aquests coeficients. El resultat d’aquest ajust es un conjunt d’observacions “millorades” del sensor inercial que es poden processar en un navegador INS/GNSS convencional. En cas de no haver-hi sensor de referència, l’altra opció per a determinar els coeficients es a través d’un filtre de Kalman estès que incorpora aquests coeficients. En el cas d’aquesta tesi, aquests coeficients van ser implementats i comprovats en un navegador INS/GNSS amb arquitectura “loosely coupled”. Pel que coneix l’autor d’aquesta tesi, la utilització de les derivades de les observacions mai s’han investigat i es demostra que ajuden a millorar la precisió de la trajectòria subministrada pel navegador INS/GNSS.

Understanding the motion of a camera from only the image(s) it captures is a di cult problem. At best we might hope to estimate the relative motion between camera and scene if we assume a static subject, but once we start considering scenes with dynamic content it becomes di cult to di↵erentiate between motion due to the observer or motion due to scene movement. In this thesis we show how the invaluable cues provided by inertial sensor data can be used to simplify motion analysis and relax requirements for several computer vision problems. This work was funded by the University of Bath.

The need to characterize normal and pathological human movement has consistently driven researchers to develop new tracking devices and to improve movement analysis systems. Movement has traditionally been captured by either optical, magnetic, mechanical, structured light, or acoustic systems. All of these systems have inherent limitations. Optical systems are costly, require fixed cameras in a controlled environment, and suffer from problems of occlusion. Similarly, acoustic and structured light systems suffer from the occlusion problem. Magnetic and radio frequency systems suffer from electromagnetic disturbances, noise and multipath problems. Mechanical systems have physical constraints that limit the natural body movement. Recently, the availability of low-cost wearable inertial sensors containing accelerometers, gyroscopes, and magnetometers has provided an alternative means to overcome the limitations of other motion capture systems. Inertial sensors can be used to track human movement in and outside of a laboratory, cannot be occluded, and are low cost. To calculate changes in orientation, researchers often integrate the angular velocity. However, a relatively small error or drift in the measured angular velocity leads to large integration errors. This restricts the time of accurate measurement and tracking to a few seconds. To compensate that drift, complementary data from accelerometers and magnetometers are normally integrated in tracking systems that utilize the Kalman filter (KF) or the extended Kalman filter (EKF) to fuse the nonlinear inertial data. Orientation estimates are only accurate for brief moments when the body is not moving and acceleration is only due to gravity. Moreover, success of using magnetometers to compensate drift about the vertical axis is limited by magnetic field disturbance. We combine kinematic models designed for control of robotic arms with state space methods to estimate angles of the human shoulder and elbow using two wireless wearable inertial measurement units. The same method can be used to track movement of other joints using a minimal sensor configuration with one sensor on each segment. Each limb is modeled as one kinematic chain. Velocity and acceleration are recursively tracked and propagated from one limb segment to another using Newton-Euler equations implemented in state space form. To mitigate the effect of sensor drift on the tracking accuracy, our system incorporates natural physical constraints on the range of motion for each joint, models gyroscope and accelerometer random drift, and uses zero-velocity updates. The combined effect of imposing physical constraints on state estimates and modeling the sensor random drift results in superior joint angles estimates. The tracker utilizes the unscented Kalman filter (UKF) which is an improvement to the EKF. This removes the need for linearization of the system equations which introduces tracking errors. We validate the performance of the inertial tracking system over long durations of slow, normal, and fast movements. Joint angles obtained from our inertial tracker are compared to those obtained from an optical tracking system and a high-precision industrial robot arm. Results show an excellent agreement between joint angles estimated by the inertial tracker and those obtained from the two reference systems.

In order to use high power headlights on heavy duty vehicles, an automatic mechanism for adjusting the level of the headlights must be used. This is in order to avoid glare while still maintaining good visibility. For headlights exceeding 2000 lumen, this control must be done automatically. The main reason for a change in the headlight level is when the truck is being loaded, and the suspension is compressed causing the headlights to point slightly up or down. Due to inherent limitations of the Scania trucks, the most commonly used approach of estimating the vehicle pitch angle is not possible to implement. Thus, a set of pitch estimation methods largely using sensors detecting acceleration are investigated and presented. Further, three methods are chosen for further study and evaluation. One method using previously recorded data about road slope as well as an on board acceleration sensor is shown to produce a high quality vehicle pitch estimate.

Thesis: S.M., Massachusetts Institute of Technology, Department of Mechanical Engineering, 2017.<br>Cataloged from PDF version of thesis.<br>Includes bibliographical references (pages 80-82).<br>Agility is defined as the ability to quickly change speed or direction. Planned agility refers to the physical act of changing direction and reactive agility addresses the additional cognitive responses needed to react quickly to an external cue. This work specifically considers reactive agility. Agility performance is often evaluated using time-based metrics, which provide little information about which factors aid or limit success. Two studies were completed to identify key factors contributing to agility performance. The objective of the first study was to determine how novices and experts working in athletic, clinical, and military environments qualitatively and quantitatively evaluate agility performance. Thirty-three participants completed a survey which involved scoring 16 athletes on a 7 point Likert scale of not agile to agile. The spread of the scores indicated that even within groups, participants had different opinions about which aspects of technique contributed to high performance. Participant responses were used to link several terms to agility technique. The objective of the second study was to apply these terms to the development of objective biomechanical metrics. An array of body-worn inertial sensors was used to calculate metrics that were sensitive to performance speed. Five metrics were defined (normalized number of foot contacts, stride length variance, arm swing variance, mean normalized stride frequency, and number of body rotations). Eighteen participants donned 13 sensors to complete a reactive agility task, which involved navigating a set of cones in response to a vocal cue. Participants were grouped into fast, medium, and slow performance based on their completion time. Participants in the fast group had the smallest number of foot contacts after normalizing by height, highest stride length variance, highest forearm angular velocity variance, and highest stride frequency after normalizing by height.These metric values translate to an efficient strategy for making turns by minimizing path length between cues and cones, effectively adjusting stride in reaction to turn points, and using tight pumping arm motions to aid in accelerating out of endpoint cones.The results of this study have the potential to inform the development of a composite agility score constructed from the list of significant metrics. Study 1 informed the quantification of qualitative agility terminology and Study 2 mapped these terms to speed of performance. The outcomes from these studies can assist in strategy development for training and rehabilitation across athletic, clinical, and military domains.<br>by Chika U. Eke.<br>S.M.

Inertial sensors, specifically MEMS gyroscopes, suffer in performance with down scaling. Non linear amplification techniques, such as parametric resonance, can be employed in many resonant structures to alleviate this degradation in performance, improve sensitivity and Signal to Noise Ratio (SNR). In this thesis the application of parametric resonance amplification and damping to both modes of a vibratory gyroscope is carried out using specialized combs. Gap-varying combs, which are usually used for the sensing mode are known for producing electrostatic spring modulations. They are used in this thesis to achieve parametric modulation in sense mode, for increasing spectral selectivity and to reduce the equivalent input noise angular rate (from 0.0046 deg/s/√Hz to 0.0026 deg/s/√Hz , for a parametric gain of 5). Additionally, novel shaped combs were used for performing parametric modulation of the driven mode of a resonant gyroscope as well. Analytical modes for both types of parametric amplification are derived and experimentally verified. In order to study the effect of parametric modulation for large signal operation, the dynamic pull-in process is analyzed and modeled in inertial MEMS sensors. The dynamic analytical model is derived and experimentally verified for parametric amplification. The dependence of dynamic pull-in voltage amplitudes on the values of externally-induced accelerations (e.g. Coriolis accelerations in the case of vibratory gyroscopes) is experimentally. The measurements indicate that the dynamic pull-in voltages reduce from 100 V to 56 V for a designed and fabricated MEMS gyroscope (device A) and from 21.77 V to 17.3 V for a MEMS accelerometer (device B), for an equivalent input acceleration signal of 0.319 ms-2, when the structures are actuated at their resonance frequency. In order to further analyze the fundamental limitations of sensing at microscale, a separate noise analysis of MEMS resonant sensors is performed. The frequency-dependent damping theory is used to suggest new optimization methods for the design of MEMS vibratory gyroscopes.

The gravitational wave observatory and many other large ground-based instruments need to be decoupled from the Earth’s ever-present motion to improve their performance. In such scenarios, inertial sensors which measure the ground motion are necessary, especially those with a high resolution and a large dynamic range. This thesis aims to develop high performance inertial sensors which outperform the commercially available ones in terms of resolution and dynamic range in low frequency down to 0.01 Hz.Inertial sensors essentially consist of two parts: a single-degree-of-freedom mechanism and a transducer which converts mechanical quantities into electrical quantities. In this work, a novel interferometric readout based on homodyne quadrature interferometer is proposed and examined. Experimental results show that its resolution is 1e-11, 1e-12 and 2e-13 m/rtHz at 0.01, 0.1 and 1 Hz respectively. For the mechanical parts, the leaf spring pendulum and Lehman pendulum are used respectively as the restoring springs for the vertical and horizontal inertial sensors. With these, the resonance frequencies are made to 0.26 and 0.11 Hz, respectively. Combined with the interferometric readout, a Vertical Interferometric Inertial Sensor (VINS) and a Horizontal Interferometric Inertial Sensor (HINS) are developed. They are placed together in a vacuum chamber as an inertial unit to measure vertical and horizontal motion.A critical investigation of the developed HINS and VINS is performed. The passive VINS and HINS are compared, firstly, with a commercial seismometer (Guralp 6T) the results showed that they provide equivalent seismograms in frequencies from tides to 10 Hz. Secondly, both simulations and measurements have been conducted in this study, a noise budget of the interferometric readout itself was constructed, which corresponds to the case when the proof-mass of the inertial sensors is blocked. At present, the resolution of the interferometric readout is found to be limited by the photodetector noise from 0.01 to 1 Hz. Moreover, huddle tests were conducted for the inertial units to examine their overall performance. However, extra experiments and simulations are performed and it is found that the resolution identified from the experimental means is worse than that from the simulation. Nevertheless, the mismatch can be reduced by reducing the magnitude of input ground vibration, by reducing undesired inputs and improving the stability of the interferometric readout output signal.<br>Doctorat en Sciences de l'ingénieur et technologie<br>info:eu-repo/semantics/nonPublished

ABSTRACT DETERMINATION OF STOCHASTIC MODEL PARAMETERS OF INERTIAL SENSORS &Uuml<br>nver, Alper PhD, Department of Electric Electronic Engineering Supervisor: Prof. Dr. M&uuml<br>beccel Demirekler January 2013, 82 pages Gyro and accelerometer systematic errors due to biases, scale factors, and misalignments can be compensated via an on-board Kalman filtering approach in a Navigation System. On the other hand, sensor random noise sources such as Quantization Noise (QN), Angular Random Walk (ARW), Flicker Noise (FN), and Rate Random Walk (RRW) are not easily estimated by an on-board filter, due to their random characteristics. In this thesis a new method based on the variance of difference sequences is proposed to compute the powers of the above mentioned noise sources. The method is capable of online or offline estimation of stochastic model parameters of the inertial sensors. Our aim in this study is the estimation of ARW, FN and RRW parameters besides the quantization and the Gauss-Markov noise parameters of the inertial sensors. The proposed method is tested both on the simulated and the real sensor data and the results are compared with the Allan variance method. Comparison shows very satisfactory results for the performance of the method. Computational load of the new method is less than the computational load of the Allan variance on the order of tens. One of the usages of this method is the individual noise characterization. A noise, whose power spectral density has a constant slope, can be identified accurately by the proposed method. In addition to this, the parameters of the GM noise can also be determined. Another idea developed here is to approximate the overall error source as a combination of ARW and some number of GM sources only. The reasons of selecting such a structure is the feasibility of using these models in a Kalman filter framework for error propagation as well as their generality of modeling other noise sources.

<p> In the past several years, IMU&rsquo;s have been widely used to measure the orientation of a moving body over a continuous period of time. Although, inertial navigation is a common approach for estimating the orientation, it greatly suffers from the accumulation of error in the orientation estimation. Most of the current common practices apply zero velocity update as a calibration method to address this problem and improve the estimation accuracy. However, this approach requires the sensors to be stationary frequently. </p><p> This thesis introduces a novel method of calibration for estimating the elevation and bank angles of the orientation over a persistent human movement utilizing accelerometers and gyroscopes. The proposed technique incorporates the prior knowledge about the human motion to the estimation of the orientation to prevent the estimated position from growing unboundedly. The measurement model is designed to estimate the position for <i>T</i> seconds in the future. The knowledge of the estimated position for few seconds further in the future provides a feedback for orientation estimation during the periods of time when the accelerometer&rsquo;s readings are significantly deviated from gravity. </p><p> This work evaluates the performance of the proposed method in two different ways: 1. a model of human movement is designed to generate synthetic data which resembles human motion. 2. an experimental design is implemented using a robot arm and an actual IMU to capture real data. The performance of the new technique is compared with the results from the inertial navigation approach. It is demonstrated that the new method significantly improves the accuracy of the orientation estimation.</p><p>

Falls are the most frequent cause of unintentional injuries among older adults; afflicting 30 percent of persons aged 65 and older and more than 50 percent of persons aged 85 and older. There is a serious need for strategies to prevent falls in elderly individuals, but an important challenge in fall prevention is the paucity of objective evidence regarding the mechanisms that lead directly to falls. There exists no mechanisms about how to predict and manage elderly falls, which has multifactorial risk factors associated with its occurrence in the elderly. As the U.S. population continues to age, both the number of falls as well as the cost of treatment of fall injuries will continue to grow. Decades of research in fall prevention has not led to a decrease in the fall incidence; thus new strategies need to be introduced to understand and prevent falls. Aging reduces the adaptability of various physical and environmental stressors that hinder stability and balance maintenance and may therefore result in a fall. Movement variability in an individual's task performance can be used to assess the limitations of the movement control system. Maintaining variation in movement engenders flexible and adaptable modalities for elderly individuals to prevent falls in an unpredictable and ever changing external environment. Conversely, excessive variability of movement may drive the control system closer to its stability limits during balance and walking tasks. Accordingly, inertial sensors are an emerging wearable technology that can facilitate noninvasive monitoring of fall prone individuals in clinical settings. This research examined the potential of inertial sensors for use in clinical settings, and evaluated their effectiveness in comparison to mature laboratory systems (i.e., force platform and camera system). Study findings showed a relationship between movement variability and fall risk among healthy young and older adults. Further, the outcomes of this work translates to the clinical environment to better understand the health status (leading to frailty) of cardiac patients; reflected by the underlying adaptability of the control system, but requires further improvements if to be used as robust clinical tool. This research provides the groundwork for rapid clinical assessments in which its validity and robustness should be investigated in future efforts.<br>Ph. D.

In this thesis, we consider the problem of estimating position and orientation (6D pose) using inertial sensors (accelerometers and gyroscopes). Inertial sensors provide information about the change in position and orientation at high sampling rates. However, they suffer from integration drift and hence need to be supplemented with additional sensors. To combine information from the inertial sensors with information from other sensors we use probabilistic models, both for sensor fusion and for sensor calibration. Inertial sensors can be supplemented with magnetometers, which are typically used to provide heading information. This relies on the assumption that the measured magnetic field is equal to a constant local magnetic field and that the magnetometer is properly calibrated. However, the presence of metallic objects in the vicinity of the sensor will make the first assumption invalid. If the metallic object is rigidly attached to the sensor, the magnetometer can be calibrated for the presence of this magnetic disturbance. Afterwards, the measurements can be used for heading estimation as if the disturbance was not present. We present a practical magnetometer calibration algorithm that is experimentally shown to lead to improved heading estimates. An alternative approach is to exploit the presence of magnetic disturbances in indoor environments by using them as a source of position information. We show that in the vicinity of a magnetic coil it is possible to obtain accurate position estimates using inertial sensors, magnetometers and knowledge of the magnetic field induced by the coil. We also consider the problem of estimating a human body’s 6D pose. For this, multiple inertial sensors are placed on the body. Information from the inertial sensors is combined using a biomechanical model which represents the human body as consisting of connected body segments. We solve this problem using an optimization-based approach and show that accurate 6D pose estimates are obtained. These estimates accurately represent the relative position and orientation of the human body, i.e. the shape of the body is accurately represented but the absolute position can not be determined. To estimate absolute position of the body, we consider the problem of indoor positioning using time of arrival measurements from an ultra-wideband (uwb) system in combination with inertial measurements. Our algorithm uses a tightlycoupled sensor fusion approach and is shown to lead to accurate position and orientation estimates. To be able to obtain position information from the uwb measurements, it is imperative that accurate estimates of the receivers’ positions and clock offsets are known. Hence, we also present an easy-to-use algorithm to calibrate the uwb system. It is based on a maximum likelihood formulation and represents the uwb measurements assuming a heavy-tailed asymmetric noise distribution to account for measurement outliers.

Micro-Electro-Mechanical Systems (MEMS) inertial sensors are used to measure acceleration and rate of angular rotation based on the deformation of a small sensitive structure, and are widely used from low-cost commercial to high-performance space and defence applications. Packaging processes can induce thermo-mechanical stresses and curvature within the MEMS die, leading to stressing and deformation of the sensitive MEMS element, which change over time and temperature. These effects cause performance changes over time and temperature, including zero-g offsets and bias and scale factor changes in MEMS accelerometers and quadrature bias changes in MEMS gyroscopes. The die bond, a small adhesive bond attaching the die to the package, can induce significant stresses and deformation in the die that have potential to cause these effects. This thesis focuses on the die curvature induced by the die bonding process, the mechanical properties of bonds, and investigates the changes in properties over temperature and time. The effects of bond design parameters are investigated to evaluate the performance of 4-dot and 9-dot bond shapes to reduce stress and reduce changes in curvature and mechanical properties over temperature and time. The behaviour of these bond shapes and the traditional square full coverage bond with thickness reductions is also investigated. Methods to characterise the mechanical properties of bonds are developed based on vibration testing of plain bonded dies, in bounce and shear directions, and using a single degree of freedom Kelvin-Voigt model with hysteretic damping. These methods are used to characterise different bond designs over the typical operating temperature range of a MEMS inertial sensor, in terms of apparent modulus and loss factor, as well as over time caused by thermal cycling and high temperature storage. White light interferometry is used to assess the die deformation induced by bonding and changes over time. The influence of imperfection effects on the dynamic response of the die is investigated, including coupling between translational and rotational coordinates and the resulting errors induced in the characterised properties. Finite Element models are developed to investigate the bond design parameters and their effect on bond induced curvature and translational die mode frequencies, along with their change over temperature. The results indicate the bond design parameters that are expected to reduce die curvature and property changes over temperature, as well as beneficial bond material properties. It was found that a bond material with a reduced thermal expansion coefficient provided the most significant reduction in die curvature from varying the bond material properties, and greatly reduced the percentage change in bond properties over temperature. It is thought that minimisation of performance effects would be achieved through matching of the bond and adherends thermal expansion rates. The 9-Dot bond was observed to induce the lowest die curvature of the bond shapes investigated and it was indicated that further discretisation would lead to further improvements. To reduce the temperature dependency of the bond properties, the bond should be designed such that it does not experience significant slenderness stiffening effects. To achieve the small thicknesses used in MEMS devices a very low shape factor is required, this can be achieved through a high perimeter to cross-sectional area ratio of the bond shape. Improvements and future works are proposed to allow full assessment of bond design effects experimentally and using finite element modelling. It is recommended for future investigations into die bond designs, to use larger-scale bonded samples, but maintaining slenderness ratios, to reduce the effects of imperfections and noise observed in the experiments, through reduced coupling and increased vibration response and die curvature.

The aim of this thesis was to describe the development of motion analysis protocols for applications on upper and lower limb extremities, by using inertial sensors-based systems. Inertial sensors-based systems are relatively recent. Knowledge and development of methods and algorithms for the use of such systems for clinical purposes is therefore limited if compared with stereophotogrammetry. However, their advantages in terms of low cost, portability, small size, are a valid reason to follow this direction. When developing motion analysis protocols based on inertial sensors, attention must be given to several aspects, like the accuracy of inertial sensors-based systems and their reliability. The need to develop specific algorithms/methods and software for using these systems for specific applications, is as much important as the development of motion analysis protocols based on them. For this reason, the goal of the 3-years research project described in this thesis was achieved first of all trying to correctly design the protocols based on inertial sensors, in terms of exploring and developing which features were suitable for the specific application of the protocols. The use of optoelectronic systems was necessary because they provided a gold standard and accurate measurement, which was used as a reference for the validation of the protocols based on inertial sensors. The protocols described in this thesis can be particularly helpful for rehabilitation centers in which the high cost of instrumentation or the limited working areas do not allow the use of stereophotogrammetry. Moreover, many applications requiring upper and lower limb motion analysis to be performed outside the laboratories will benefit from these protocols, for example performing gait analysis along the corridors. Out of the buildings, the condition of steady-state walking or the behavior of the prosthetic devices when encountering slopes or obstacles during walking can also be assessed. The application of inertial sensors on lower limb amputees presents conditions which are challenging for magnetometer-based systems, due to ferromagnetic material commonly adopted for the construction of idraulic components or motors. INAIL Prostheses Centre stimulated and, together with Xsens Technologies B.V. supported the development of additional methods for improving the accuracy of MTx in measuring the 3D kinematics for lower limb prostheses, with the results provided in this thesis. In the author’s opinion, this thesis and the motion analysis protocols based on inertial sensors here described, are a demonstration of how a strict collaboration between the industry, the clinical centers, the research laboratories, can improve the knowledge, exchange know-how, with the common goal to develop new application-oriented systems.

<p>This thesis deals with estimating position and orientation in real-time, using measurements from vision and inertial sensors. A system has been developed to solve this problem in unprepared environments, assuming that a map or scene model is available. Compared to ‘camera-only’ systems, the combination of the complementary sensors yields an accurate and robust system which can handle periods with uninformative or no vision data and reduces the need for high frequency vision updates.</p><p>The system achieves real-time pose estimation by fusing vision and inertial sensors using the framework of nonlinear state estimation for which state space models have been developed. The performance of the system has been evaluated using an augmented reality application where the output from the system is used to superimpose virtual graphics on the live video stream. Furthermore, experiments have been performed where an industrial robot providing ground truth data is used to move the sensor unit. In both cases the system performed well.</p><p>Calibration of the relative position and orientation of the camera and the inertial sensor turn out to be essential for proper operation of the system. A new and easy-to-use algorithm for estimating these has been developed using a gray-box system identification approach. Experimental results show that the algorithm works well in practice.</p>

Integration of GPS with Inertial Navigation Systems (INS) can provide reliable and complete positioning and geo-referencing parameters including position, velocity and attitude for dynamic platforms for a variety of applications. This research has been focusing on four modelling and implementation issues for a GPS/INS integrated platform in order to improve the overall integration performance, in particular: a) Time synchronization Having recognised that having a precise time synchronisation of measurements is fundamental in constructing a multi-sensor integration platform and is critical for achieving high data fusion performance, various time synchronisation scenarios and solutions have been investigated. A criterion for evaluating synchronisation accuracy and error impacts has been derived; an innovative time synchronisation solution has been proposed; an applicable data logging system has been implemented with off-the-shelf components and tested. b) Noise suppression of INS raw measurements Low cost INS sensors, especially MEMS INS, would normally exhibit much larger measurement noise than conventional INS sensors. A novel method of using vehicle dynamic information for de-noising raw INS sensor measurements has been proposed in this research. Since the vehicle dynamic model has the characteristic of a low pass filter, passing the raw INS sensor measurements through it effectively reduces the high frequency noise component. c) Adaptive Kalman filtering The present data fusion algorithms, which are mostly based on the Kalman filter, have the stringent requirement on precise a priori knowledge of the system model and noise properties. This research has investigated the utilization issues of online stochastic modelling algorithm, and then proposed a new adaptive process noise scaling algorithm which has shown remarkable capability in autonomously tuning the process noise covariance estimates to the optimal magnitude. d) Integration of a low cost INS sensor with a standalone GPS receiver To improve the performance where a standalone GPS receiver integrated with a MEMS INS, additional velocity aiding and a new integration structure has been adopted in this research. Field test shows that velocity determination accuracy could reach the centimetre level, and the errors of MEMS INS have been limited to such a level that it can generate stable attitude and heading references under low dynamic conditions.

Back pain is a common and costly disorder affecting 80% of the population, with 80-90% of the symptoms reported to have no pathological cause and it is suggested that this non-specific low back pain can be improved by the adoption of proper posture and body mechanics during normal daily life.

Kernel methods are able to exploit high-dimensional spaces for representational advantage, while only operating implicitly in such spaces, thus incurring none of the computational cost of doing so. They appear to have the potential to advance the state of the art in control and signal processing applications and are increasingly seeing adoption across these domains. Applications of kernel methods to fault detection and isolation (FDI) have been reported, but few in aerospace research, though they offer a promising way to perform or enhance fault detection. It is mostly in process monitoring, in the chemical processing industry for example, that these techniques have found broader application. This research work explores the use of kernel-based solutions in model-based fault diagnosis for aerospace systems. Specifically, it investigates the application of these techniques to the detection and isolation of IMU/INS sensor faults – a canonical open problem in the aerospace field. Kernel PCA, a kernelised non-linear extension of the well-known principal component analysis (PCA) algorithm, is implemented to tackle IMU fault monitoring. An isolation scheme is extrapolated based on the strong duality known to exist between probably the most widely practiced method of FDI in the aerospace domain – the parity space technique – and linear principal component analysis. The algorithm, termed partial kernel PCA, benefits from the isolation properties of the parity space method as well as the non-linear approximation ability of kernel PCA. Further, a number of unscented non-linear filters for FDI are implemented, equipped with data-driven transition models based on Gaussian processes - a non-parametric Bayesian kernel method. A distributed estimation architecture is proposed, which besides fault diagnosis can contemporaneously perform sensor fusion. It also allows for decoupling faulty sensors from the navigation solution.

The robotics field has grown in recent years to a point where unmanned systems are no longer limited by their capabilities. As such, the mission profiles for unmanned systems are becoming more and more complicated, and a demand has risen for the deployment of unmanned systems into the most complex of environments. Additionally, the objectives for unmanned systems are once more complicated by the necessity for beyond line of sight teleoperation, and in some cases complete vehicle autonomy. Such systems require adequate sensory devices for appropriate situational awareness. Additionally, a large majority of what is currently being done with unmanned systems requires visual data acquisition. A stereo vision system is ideal for such missions as it doubles as both an image acquisition device, and a range finding device. The 2D images captured with a stereo vision system can be mapped to three dimensional point clouds with reference to the optic center of one of the stereo cameras. While stand alone commercial stereo vision systems are capable of doing just that, the GPS/INS aided stereo vision system also has integrated 3-axis accelerometers, 3-axis gyros, 3-axis magnetometer, and GPS receiver allowing for the measurement of the systemâ s position and orientation in global coordinates. This capability provides the potential to georeference the 3D data captured with the stereo camera. The GPS/INS aided stereo vision system integrates a combination of commercial and inhouse developed devices. The total system includes a Point Grey Research Bumblebee stereovision camera, a Versalogic PC104 computer, a PCB designed for sensor acquisition and power considerations, and a self contained battery. The entire system is all contained within a 9.5â x 5â x 6.5â aluminum enclosure and weighs approximately 6 lbs. The system is also accompanied with a graphical user interface which displays the geo-referenced data within a 3D virtual environment providing adequate sensor feedback for a teleoperated unmanned vehicle. This thesis details the design and implementation of the hardware and software included within this system as well as the results of operation.<br>Master of Science

The main topics of this master's project are non-linear ltering techniques,system identication, and supervised classication. The main goal is to develop,implement, and test methods using inertial sensors to monitor doorandwindow movement. Secondly, it aims to investigate if it is possible tomonitor door movements by using event classication. The thesis beginswith an overview of the available sensors and a theoretical part, in which theltering algorithms and the modelling of door- and window movements aresummarized. Later a supervised learning algorithm is developed. Lastly, thehardware used to test the algorithm and the results from the tests are shown.The goal of the project is to develop ltering techniques using accelerometers,gyroscopes, and magnetometers. They measure various inertial units of amoving body. By combining measurements from dierent sensors and fusingthem, the movement of a body can be tracked. Door- and window movementare modelled and the Kalman lter is used to estimate the true parameters ofthe model. Due to the imperfections of the sensors, properties such as osetand drift are included in the models. The models are to be used in doorsand windows to track the angle of which they are open. Due to dicultiesof tracking the movement when they hit the door- or window frame, theK-Nearest Neighbours classication algorithm is used to predict whether thedoor- or window closes fully.A general method of implementing the ltering techniques is constructed.Due to limitations in computational power, limited sampling rate of thesensors, and the nonlinear nature of the movement, some generalizations arerequired. Hardware which can be placed on doors and windows are used toverify the models. As a rst verication, the models and ltering techniqueswere tested in Matlab. Secondly, the algorithms were tested live on doorsand windows. The result shows it is possible to track the movement of doorsand windows with an accuracy of over 10. The classication algorithmshows some success with tracking a closing door but requires more work tobecome more accurate. The result also shows that the sampling rate has tobe taken into consideration when designing the algorithms, a higher samplingrate results in more accurate tracking but increases power consumption andcomputing power.<br>Huvudamnet i detta examensarbete ar ickelinjara ltertekniker, systemidentikation och handledd klassiering. Huvudmalet ar att utveckla, implementeraoch testa metoder dar accelerometrar, gyroskop och magnetometraranvands for att monitorera dorr- och fonsterrorelser. Examensarbetet amnaraven att undersoka mojligheten att upptacka plotsliga dorrorelser genom attanvanda klassiering av handelser. Uppsatsen borjar med en overblick avtillgangliga sensor och en teoretisk del dar Kalmanltret och modelleringenav dorr- och fonsterrorelser sammanfattas. I den andra delen tas en klassi-eringsalgoritm fram. Slutligen visas resultatet fran den praktiska delen darhardvara och sensorer har anvants for att utvardera algoritmerna.Malet med projektet ar att utveckla ltertekniker dar accelerometer, gyroskopoch magnetometer anvands. De mater olika fysiska egenskaper hosen kropp i rorelse. Genom att kombinera sensorerna och fusera data kandorr- och fonsterrorelser sparas. Pagrund av fel i sensorerna uppstar egenskapersom forskjutning och drift vilket over tid ger felaktiga resultat. Dorrochfonsterrorelser ar ickelinjart modellerade och det utvidgade Kalmanlteranvands for att estimera de riktiga parametrarna i modellerna. Modellernaanvands mer specikt for att estimera vinkeln som dorrar och fonster aroppna. Pa grund av svarigheten med att spara vinkeln nar de slar i karmenutvecklades en K-Nearest Neighbours klassieringsalgoritm som anvands foratt spara om dorren eller fonstret stangdes helt.En generell metod for att implementera ltren anvands. Pa grund av begransningar nar det galler berakningskraft, samplingsfrekvens och modellernasickelinjara beteende gjorde vissa generaliseringar. Existerande hardvarasom kunde placeras pa dorrar och fonster anvandes for att veriera modellerna.Som en forsta verikation anvandes Matlab. For veriering anvandesen mikrokontroller forsedd med sensorer monterad pa dorrar och fonster, daralgoritmerna kunde utvarderas. Resultatet ar att det ar mojligt att spara vinkelnmed en noggrannhet pa over 10. Klassieringen visar framgang med attestimera om en dorr eller fonster stangdes helt men behover mer jobb for attbli stabilare. Noggrannheten pa resultatet ar beroende pa berakningskomplexitetenoch samplingsfrekvensen for sensorerna. Hogre frekvens ger battre resultatmen ar kostsammare och kraver storre berakningskraft.

Most mobile video-recording devices of today, e.g. cell phones and music players, make use of a rolling shutter camera. A rolling shutter camera captures video by recording every frame line-by-line from top to bottom of the image, leading to image distortions in situations where either the device or the target is moving. Recording video by hand also leads to visible frame-to-frame jitter. In this thesis, methods to decrease distortion caused by the motion of a video-recording device with a rolling shutter camera are presented. The methods are based on estimating the orientation of the camera from gyroscope and accelerometer measurements. The algorithms are implemented on the iPod Touch 4, and the resulting videos are compared to those of competing stabilization software, both commercial and free, in a series of blind experiments. The results from this user study shows that the methods presented in the thesis perform equal to or better than the others.<br>Mobiltelefoner, mp3-spelare och andra bärbara enheter som kan spela in film har ofta en kamera med rullande slutare. En sådan kamera fångar varje bild rad för rad, från topp till botten. Detta resulterar i distortion i bilden om antingen enheten eller objekt i bilden rör sig. Att filma för hand introducerar också skakighet i filmen. I det här examensarbetet presenteras metoder för att minska den distortion som uppstår till följd av att en enhet med rullande slutare förflyttas under inspelning av en film. Metoderna bygger på estimering av kamerans orientering, utgående från mätdata från gyroskop och accelerometer. Algoritmerna har implementerats på en iPod Touch 4 och de resulterande filmerna har jämförts med de från konkurrerande program i en serie blindtester. Resultaten från denna undersökning visar att metoderna som presenteras i examensarbetet är lika bra eller bättre än de övriga.

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, 2010.<br>Page 104 blank. Cataloged from PDF version of thesis.<br>Includes bibliographical references.<br>The monitoring of physiological biomarkers is fundamental to the diagnosis and treatment of disease. We describe here the development of molecular sensors which can be read by magnetic resonance (MR) relaxometry. MR is an advantageous bio-sensor readout because it can be determined from opaque samples and through intervening layers of matter. Wash steps can therefore be avoided in in vitro MR assays and non-invasive interrogation achieved for in vivo MR sensing. Functionalized magnetic nanoparticles originally developed as in vivo contrast agents have recently been adapted for use in magnetic relaxometry assays. The first half of this thesis describes a simple particle functionalization strategy and its application to the detection of myocardial infarction ("heart attack") associated biomarkers. The particles were subcutaneously implanted in the form of small discrete sensors and shown to be efficacious in measuring the physiological release of three protein biomarkers. Alternative contrast mechanisms may also be employed by MR readable sensors. The second half of this thesis introduces the novel use of 'smart' polymers which produce analyte-responsive changes in MR relaxivity. We show that MR responsive calcium-crosslinked and pH-swelling hydrogels can be incorporated within discrete sensors.<br>by Yibo Ling.<br>Ph.D.

The purpose of this thesis is to propose a novel application of amorphous magnetic ribbons for use as knee force measurement sensors, without the need for secondary windings. This thesis demonstrates that the magnetic properties of amorphous ribbons are retained when embedded in Ultra-high molecular weight polyethylene (UHMWPE) tibial inserts, and these properties can be interrogated non-invasively. This is of importance, as it offers a viable solution for instrumented prosthesis which can be used for in-vivo monitoring. The research conducted also demonstrates that the tibiofemoral contact force on the instrumented tibial insert can be measured by observing the impedance changes in adjacent coils. Other conventional methods, though effective, require additional circuitry for non-invasive retrieval of measured data. The work contained herein eliminates this need, thereby reducing the structural modification of the implant required to accommodate the additional components. This research also shows that the variation in the coil impedance can be related to the permeability changes in the amorphous ribbons, and these can be quantified by tracking the resonant frequency of the coils. Amorphous ribbons have not been used in monitoring orthopaedic prosthesis before, and this work shows how the simplified measurement system can offer an alternative technique to knee implant monitoring.

Indoor positioning is a rapidly growing research area, enabling new innovative location-aware applications and user-oriented services. Location Fingerprinting (LF) is the positioning technique of coupling a physical location with observed radio signal measurements. In the terms of indoor LF using Wireless Local Area Network (WLAN) it refers to the use of network measurements from the WLAN Access Points (APs) to tag known locations. A data set is created containing reference fingerprints for the area of interest and is known as a radio map. A radio map can later be used to find a user's location in the area of interest. WLAN infrastructures are vulnerable to many kinds of faults and malicious attacks, including, an attacker jamming the signal from an AP, or an AP becoming unavailable during positioning due to power outage. These faults can be collectively characterized as an AP-failure. In LF positioning systems, AP-failure faults can significantly degrade the performance of a LF system due to the difference between the current fingerprints and radio map created with all APs being available. It is desirable to detect such faulty APs, in order to take actions towards fault-mitigation and restoration, in case of a malicious attack. In this work, we have developed a fault detection algorithm that uses inertial sensors (i.e., accelerometer, magnetometer) available in smartphones to detect AP-failure faults in LF systems. Inertial Measurement Unit (IMU) has become an integral part of all high-end smartphones. IMU can be used to infer location information on the smartphone. The main idea is to have two parallel position streams, the LF positioning and the IMU positioning, and to compare the mean positioning error between the two. Since IMU positioning is fairly accurate once provided with starting coordinates, we use it to detect abnormal behaviour in LF positioning system, such as highly erroneous estimates signifying an AP-failure fault present in the system. The performance of the proposed detection algorithm is evaluated with several real-life AP-related faults. The proposed algorithm exhibits low probability of false alarms in the detection of faulty APs. The conclusion is that using IMU based positioning is an effective and robust solution in terms of fault detection in LF systems.

PONTIFÍCIA UNIVERSIDADE CATÓLICA DO RIO DE JANEIRO<br>CONSELHO NACIONAL DE DESENVOLVIMENTO CIENTÍFICO E TECNOLÓGICO<br>Simuladores de movimentos são sistemas mecatrônicos que reproduzem as principais atitudes e movimentos de um veículo. Neste estudo serão analisados simuladores baseados em mecanismos com 3 e 6 graus de liberdade. No segundo caso, o mecanismo é capaz de reproduzir todos os ângulos de atitude (rolagem, arfagem e guinada) e todos os deslocamentos lineares (lateral, vertical e longitudinal) com limitações, porém com amplitude suficiente de modo a possibilitar os principais movimentos associados ao veículo. O uso de transdutores de deslocamento linear nestes mecanismos articulados introduzem elevados efeitos de inércia, além de aumentar a massa dos mesmos, diminuindo sua relação carga/peso e sua eficiência. Atualmente, o grande desenvolvimento de sensores do tipo unidade de medição inercial (IMU) aumentou a disponibilidade destes no mercado e reduziu muito seu custo. Como se trata de acelerômetros triaxiais em conjunto com girômetros também triaxiais, sensores como este podem ser usados para determinar a posição e a orientação no espaço de mecanismos com seis graus de liberdade, como a Plataforma Stewart. Neste trabalho será desenvolvida uma metodologia para modelagem da cinemática de mecanismos paralelos baseada nos derivativos de suas matrizes jacobianas. Esta metodologia é avaliada em um mecanismo paralelo plano de três graus de liberdade e em uma Plataforma Stewart. Com a metodologia de modelagem validada, é implementada uma estratégia de controle baseada no uso de um sensor tipo central inercial para o controle de posição, velocidade e aceleração destes mecanismos. Os resultados das simulações indicam a possibilidade do uso destes sensores nestes tipos de equipamentos e apontam para a necessidade de avaliar esta metodologia em testes experimentais.<br>Movement simulators are mechatronic systems that reproduce the main attitudes and movements of a vehicle. In this study are examined simulators based on 3 and 6 degrees of freedom mechanisms. In the second case, the mechanism is able to reproduce all the attitude angles (roll, pitch and yaw) and all the linear displacements (sway, heave and surge) with limitations, but with sufficient amplitude to enable the main movements associated with the vehicle. The use of linear displacement transducers in these articulated mechanisms introduce high inertia effects and increase the mass, decreasing the load/weight ratio and efficiency. Currently, the great development of the inertial central type sensors (IMU – Inertial measurement unit) increased the availability of these transducers on market and greatly reduced cost. Since this is a conjunct of triaxial accelerometers with triaxial gyrometers, sensors such as these ones can be used to determine the position and orientation in space of mechanisms with six degrees of freedom, such as the Stewart Platform. In this work it will be developed a methodology for modeling the kinematics of parallel mechanisms based on derivatives of their jacobian matrices. This methodology is evaluated in a planar parallel mechanism of three degrees of freedom and on a Stewart Platform. With the modeling methodology validated, a control strategy based on the use of an inertial unit type sensor for controlling the position, velocity and acceleration of these mechanisms is implemented. The simulations results indicate the possibility of using these sensors in these types of equipment and point to the need to evaluate this methodology in experimental tests.

in the everyday clinical practice. Having this in mind, the choice of a simple setup would not be enough because, even if the setup is quick and simple, the instrumental assessment would still be in addition to the daily routine. The will to overcome this limit has led to the idea of instrumenting already existing and widely used functional tests. In this way the sensor based assessment becomes an integral part of the clinical assessment. Reliable and validated signal processing methods have been successfully implemented in Personal Health Systems based on smartphone technology. At the end of this research project there is evidence that such solution can really and easily used in clinical practice in both supervised and unsupervised settings. Smartphone based solution, together or in place of dedicated wearable sensing units, can truly become a pervasive and low-cost means for providing suitable testing solutions for quantitative movement analysis with a clear clinical value, ultimately providing enhanced balance and mobility support to an aging population.<br>Una analisi obiettiva richiede un approccio di misura strumentale. La valutazione clinica impiega strumenti veloci e di facile impiego ma i risultati sono spesso dipendenti dall’operatore e ci sono effetti soffitto e pavimento che rendono queste metodiche poco sensibile per poter rilevare e monitorare piccoli cambiamenti di stato. L’analisi strumentale applicata alla valutazione funzionale invece fa uso di algoritmi autimatici per un analisi obiettiva della performance motoria che va al di la delle semplici misure che si ottengono con gli strumenti tradizionalmente in uso come i cronometri. Non solo i clinici hanno bisogno di informazioni chiare ed affidabili ma è altrettanto importante fornire queste informazione con il minimo sforzo da parte dell’utente altrimenti queste metodiche non avranno mai una reale penetrazione nella pratica clinica. Per superare i limiti dell’approccio di valutazione tradizionale si è scelto di andare a strumentare i test di valutazione più utilizzati nella pratica ordinaria. Sono stati quindi sviluppati e validati nuovi metodi di analisi per dati e segnali che diventano parte inegrante della routine clinica. Facendo uso di soluzioni basate su smartphone, queste metodiche sono entrate a far parte dei cosidetti Personal Health Systems che forniscono soluzioni pervasive e a basso costo per l’analisi del movimento in ambiente supervisionato e non supervisionato.

Grâce à l’émergence dans la vie quotidienne des appareils de plus en plus populaires que sont les smartphones et les tablettes, la tâche de postionner l'utilisateur par le biais de son téléphone est une problématique fortement étudiée dans les domaines non seulement de la recherche mais également des communautés industrielles. Parmi ces technologies, les approches GPS sont devenues une norme et ont beaucoup de succès pour une localisation en environnement extérieur. Par contre, le Wi-Fi, les capteurs inertiels et le Bluetooth sont plutôt préférés pour les tâches de positionnement dans un environnement intérieur.Pour ce qui concerne le positionnement des smartphones, les approches basées sur les « empreintes digitales » (fingerprint) Wi-Fi sont bien établies. D'une manière générale, ces approches tentent d'apprendre la fonction de correspondance (cartographie) des caractéristiques du signal Wi-Fi par rapport à la position de l’appareil dans le monde réel. Elles nécessitent généralement une grande quantité de données pour obtenir une bonne cartographie. Lorsque ces données d'entraînement disponibles sont limitées, l'approche basée sur les empreintes digitales montre alors des taux d’erreurs élevés et devient moins stable. Dans nos travaux, nous explorons d’autres approches, différentes, pour faire face à cette problématique du manque de données d'entraînement. Toutes ces méthodes sont testées sur un ensemble de données public qui est utilisé lors d’une compétition internationale à la Conférence IPIN 2016.En plus du système de positionnement basé sur la technologie Wi-Fi, les capteurs inertiels du smartphone sont également utiles pour la tâche de suivi. Les trois types de capteurs, qui sont les accéléromètres, le gyroscope et la boussole magnétique, peuvent être utilisés pour suivre l'étape et la direction de l'utilisateur (méthode SHS). Le nombre d'étapes et la distance de déplacement de l'utilisateur sont calculés en utilisant les données de l'accéléromètre. La position de l'utilisateur est calculée par trois types de données avec trois méthodes comprenant la matrice de rotation, le filtre complémentaire et le filtre de Madgwick. Il est raisonnable de combiner les sorties SHS avec les sorties de Wi-Fi, car les deux technologies sont présentes dans les smartphones et se complètent. Deux approches combinées sont testées. La première approche consiste à utiliser directement les sorties Wi-Fi comme points de pivot pour la fixation de la partie de suivi SHS. Dans la deuxième approche, nous comptons sur le signal Wi-Fi pour construire un modèle d'observation, qui est ensuite intégré à l'étape d'approximation du filtre à particules. Ces combinaisons montrent une amélioration significative par rapport au suivi SHS ou au suivi Wi-Fi uniquement.Dans un contexte multiutilisateur, la technologie Bluetooth du smartphone pourrait fournir une distance approximative entre les utilisateurs. La distance relative est calculée à partir du processus de numérisation du périphérique Bluetooth. Elle est ensuite utilisée pour améliorer la sortie des modèles de positionnement Wi-Fi. Nous étudions deux méthodes. La première vise à créer une fonction d'erreur qui permet de modéliser le bruit dans la sortie Wi-Fi et la distance approximative produite par le Bluetooth pour chaque intervalle de temps spécifié. La seconde méthode considère par contre cette relation temporelle et la contrainte de mouvement lorsque l'utilisateur se déplace. Le modèle d'observation du filtre à particules est une combinaison entre les données Wi-Fi et les données Bluetooth. Les deux approches sont testées en fonction de données réelles, qui incluent jusqu'à quatre utilisateurs différents qui se déplacent dans un bureau. Alors que la première approche n'est applicable que dans certains scénarios spécifiques, la deuxième approche montre une amélioration significative par rapport aux résultats de position basés uniquement sur le modèle d'empreintes digitales Wi-Fi<br>With the popularity of smartphones and tablets in daily life, the task of finding user’s position through their phone gains much attention from both the research and industry communities. Technologies integrated in smartphones such as GPS, Wi-Fi, Bluetooth and camera are all capable for building a positioning system. Among those technologies, GPS has approaches have become a standard and achieved much success for the outdoor environment. Meanwhile, Wi-Fi, inertial sensors and Bluetooth are more preferred for positioning task in indoor environment.For smartphone positioning, Wi-Fi fingerprinting based approaches are well established within the field. Generally speaking, the approaches attempt to learn the mapping function from Wi-Fi signal characteristics to the real world position. They usually require a good amount of data for finding a good mapping. When the available training data is limited, the fingerprinting-based approach has high errors and becomes less stable. In our works, we want to explore different approaches of Wi-Fi fingerprinting methods for dealing with a lacking in training data. Based on the performance of the individual approaches, several ensemble strategies are proposed to improve the overall positioning performance. All the proposed methods are tested against a published dataset, which is used as the competition data of the IPIN 2016 Conference with offsite track (track 3).Besides the positioning system based on Wi-Fi technology, the smartphone’s inertial sensors are also useful for the tracking task. The three types of sensors, which are accelerate, gyroscope and magnetic, can be employed to create a Step-And-Heading (SHS) system. Several methods are tested in our approaches. The number of steps and user’s moving distance are calculated from the accelerometer data. The user’s heading is calculated from the three types of data with three methods, including rotation matrix, Complimentary Filter and Madgwick Filter. It is reasonable to combine SHS outputs with the outputs from Wi-Fi due to both technologies are present in the smartphone. Two combination approaches are tested. The first approach is to use directly the Wi-Fi outputs as pivot points for fixing the SHS tracking part. In the second approach, we rely on the Wi-Fi signal to build an observation model, which is then integrated into the particle filter approximation step. The combining paths have a significant improvement from the SHS tracking only and the Wi-Fi only. Although, SHS tracking with Wi-Fi fingerprinting improvement achieves promising results, it has a number of limitations such as requiring additional sensors calibration efforts and restriction on smartphone handling positions.In the context of multiple users, Bluetooth technology on the smartphone could provide the approximated distance between users. The relative distance is calculated from the Bluetooth inquiry process. It is then used to improve the output from Wi-Fi positioning models. We study two different combination methods. The first method aims to build an error function which is possible to model the noise in the Wi-Fi output and Bluetooth approximated distance for each specific time interval. It ignores the temporal relationship between successive Wi-Fi outputs. Position adjustments are then computed by minimizing the error function. The second method considers the temporal relationship and the movement constraint when the user moves around the area. The tracking step are carried out by using particle filter. The observation model of the particle filter are a combination between the Wi-Fi data and Bluetooth data. Both approaches are tested against real data, which include up to four different users moving in an office environment. While the first approach is only applicable in some specific scenarios, the second approach has a significant improvement from the position output based on Wi-Fi fingerprinting model only

In the diagnosis and analysis of shoulder instability a precise determination of the location and orientation of the Glenohumeral joint is important. A better understanding of shoulder kinematics and kinetics will help clinicians and therapists in the diagnosis and treatment of shoulder pathologies. To-date, non-invasive skin-based methods are often either restricted to quasi-static measurements or are inaccurate during dynamic assessments at high humeral elevations as a result of soft skin artefact. Tracking the orientation of the scapula is difficult because it is surrounded by soft tissues, is held mainly by muscles and has only one direct point of attachment to the thorax. Instability of the glenohumeral joint generates poor functionality of the shoulder labrum and capsule as well as in the muscle and connective tissue structures that surround the shoulder. As the clinical phenomenon of shoulder instability is extremely complex, one of the priorities for the specialist in avoiding a faulty diagnosis is to recognise, identify and classify shoulder pathologies such as muscle patterning instability in the early stages of the investigation. A two stage methodology for non-invasive tracking of the scapula under dynamic conditions is presented in this work. The methodology provides scapula location by combining data from two surface mounted sensors using a regression-type equation formulated from quasi-static trials undertaken using a scapula locator and three IMUs (first stage). In the second stage, the least square fit is used to improve the scapular orientation by utilising data from only two IMUs (humerus and scapula) under dynamic conditions. Accuracy was assessed in an animal study by comparing results with those from a bone based method during quasi static and dynamic tests. Tests were also undertaken to investigate the errors induced by the soft tissue artefact in surface based scapula location measurement. In dynamic trials the methodology proved more accurate in determining scapula location than a standard skin-based approach, and showed that the greatest contribution to soft tissue artefact was from the epidermal, dermal and subcutaneous tissue layers as opposed to the muscle layer. We confirmed that, in cases where subjects have relatively small amounts of soft tissue surrounding the scapula, surface based methods could provide reasonable accuracy. Our methodology utilised subject-specific data to formulate a regression equation, and can be used to provide accurate, non-invasive tracking of the scapula under dynamic conditions in subjects regardless of individual body morphology. After the methodology validation, study tests were undertaken in a case study in order to estimate the scapula orientation under dynamic conditions in a human without symptoms of any shoulder pathologies and in one participant diagnosed with shoulder instability due to muscle patterning. The two stage methodology is proven to work in a healthy human participant in dynamic tests, in a person with no suspicion of shoulder instability. This methodology allows the error reduction generated by the soft tissues surrounded the scapula. The work presented here can be used as a framework for developing diagnosis protocols by using modern technology.

This dissertation discusses the design, simulation and characterization of process-compatible accelerometers and gyroscopes for the implementation of multi-degree-of-freedom (multi-DOF) systems. All components presented herein were designed to operate under the same vacuum-sealed environment to facilitate batch fabrication and wafer-level packaging (WLP), enabling the development of small form-factor single-die inertial measurement units (IMUs). The high-aspect-ratio poly and single-crystal silicon (HARPSS) process flow was used to co-fabricate the devices that compose the system, enabling the implementation ultra-narrow capacitive gaps (< 300 nm) in thick device-layer substrates (40 um). The presented gyroscopes were implemented as high-frequency BAW disk resonators operating in a mode-matched condition. A new technique to reduced dependencies on environmental stimuli such as temperature, vibration and shock was introduced. Novel decoupling springs were utilized to effectively isolate the gyros from their substrate, minimizing the effect that external sources of error have on offset and scale-factor. The substrate-decoupled (SD) BAW gyros were interfaced with a customized IC to achieve supreme random-vibration immunity (0.012 (deg/s)/g) and excellent rejection to shock (0.075 (deg/s)/g). With a scale factor of 800 uV/(deg/s), the complete SD-BAW gyro system attains a large full-scale range (2500 deg/s) with excellent linearity. The measured angle-random walk (ARW) of 0.36 deg/rthr and bias-instability of 10.5 deg/hr are dominated by the thermal and flicker noise of the IC, respectively. Additional measurements using external electronics show bias-instability values as low as 3.5 deg/hr. To implement the final monolithic multi-DOF IMU, accelerometers were carefully designed to operate in the same vacuum environment required for the gyroscopes. Narrow capacitive gaps were used to adjust the accelerometer squeeze-film damping (SFD) levels, preventing an under-damped response. Robust simulation techniques were developed using finite-element analysis (FEA) tools to extract accurate values of SFD, which were then match with measured results. Ultra-small single proof-mass tri-axial accelerometers with Brownian-noise as low as 30 ug/rtHz were interfaced with front-end electronics exhibiting scale-factor values in the order of 5 to 10 mV/g and cross-axis sensitivities of less than 3% before any electronic compensation.

The usage of inertial sensors has traditionally been confined primarily to the aviation and marine industry due to their associated cost and bulkiness. During the last decade, however, inertial sensors have undergone a rather dramatic reduction in both size and cost with the introduction of MEMS technology. As a result of this trend, inertial sensors have become commonplace for many applications and can even be found in many consumer products, for instance smart phones, cameras and game consoles. Due to the drift inherent in inertial technology, inertial sensors are typically used in combination with aiding sensors to stabilize andimprove the estimates. The need for aiding sensors becomes even more apparent due to the reduced accuracy of MEMS inertial sensors. This thesis discusses two problems related to using inertial sensors in combination with aiding sensors. The first is the problem of sensor fusion: how to combine the information obtained from the different sensors and obtain a good estimate of position and orientation. The second problem, a prerequisite for sensor fusion, is that of calibration: the sensors themselves have to be calibrated and provide measurement in known units. Furthermore, whenever multiple sensors are combined additional calibration issues arise, since the measurements are seldom acquired in the same physical location and expressed in a common coordinate frame. Sensor fusion and calibration are discussed for the combination of inertial sensors with cameras, UWB or GPS. Two setups for estimating position and orientation in real-time are presented in this thesis. The first uses inertial sensors in combination with a camera; the second combines inertial sensors with UWB. Tightly coupled sensor fusion algorithms and experiments with performance evaluation are provided. Furthermore, this thesis contains ideas on using an optimization based sensor fusion method for a multi-segment inertial tracking system used for human motion capture as well as a sensor fusion method for combining inertial sensors with a dual GPS receiver. The above sensor fusion applications give rise to a number of calibration problems. Novel and easy-to-use calibration algorithms have been developed and tested to determine the following parameters: the magnetic field distortion when an IMU containing magnetometers is mounted close to a ferro-magnetic object, the relative position and orientation of a rigidly connected camera and IMU, as well as the clock parameters and receiver positions of an indoor UWB positioning system.<br>MATRIS (Markerless real-time Tracking for Augmented Reality Image), a sixth framework programme funded by the European Union<br>CADICS (Control, Autonomy, and Decision-making in Complex Systems), a Linneaus Center funded by the Swedish Research Council (VR)<br>Strategic Research Center MOVIII, funded by the Swedish Foundation for Strategic Research (SSF)