No description available.

No description available.

No description available.

No description available.

La technologie d’aujourd’hui donne la possibilité à chacun de se localiser grâce à son smartphone. Cependant les milieux intra-muros restent encore relativement dépourvus en service de géolocalisation. Pour des raisons d’ubiquité, les centrales inertielles et magnétiques de technologie MEMS présentes dans les smartphones offrent une solution pour la navigation pédestre. Dans ce contexte la stratégie « Pedestrian Dead-Reckoning » s’avère intéressante car elle limite la dérive temporelle de l’estimation de la distance parcourue. Cependant, l’estimation de la direction de marche s’avère critique dans la stratégie PDR, les mouvements de la main présentant un certain désordre/désalignement qui rend difficile une telle estimation. Dans un premier temps, l’estimation de l’orientation de la centrale inertielle est affinée afin de projeter avec un minimum d’erreur les mesures inertielles dans le plan horizontal. Un filtre d’estimation de l’orientation paramétré en quaternions et basé sur une exploitation opportune des champs magnétiques et de gravité a été développé. Dans un second temps, il s’agit d’estimer la direction de marche. Les méthodes de l’état de l’art proposent une estimation de la direction de marche à partir de la maximisation de l’énergie du signal. Cette approche est sensible aux mouvements de la main. Nous proposons une nouvelle approche basée sur les théories des probabilités et de l’information qui s’inspire de la description biomécanique de la marche. Des validations expérimentales sont conduites pour analyser les performances d’estimation de la direction de marche qui est directement reliée à la qualité de l’estimation du positionnement. / Thanks to new technological developments, it is now possible to get our localization with our own smartphone. However, indoor environments are still relatively lacking in localization based service. MEMS sensors, composed of inertial and magnetic sensors, offer a ubiquitous solution. These sensors can be merged with other technologies to give a reliable solution for the Pedestrian Navigation. In this context the “Pedestrian Dead Reckoning” strategy is attractive. Indeed, this strategy enables to estimate the walking distance with a limited drift. But the walking direction estimation remains critical in the PDR strategy. Hand movements are relatively erratic and cause a dynamic angular misalignment, which is difficult to estimate. Firstly, a new orientation estimation algorithm of the handheld unit is developed to reduce the errors in the horizontal inertial measurements. The filter is parametrized with quaternions and based on opportune invariant phases of the magnetic and gravity fields. Secondly, a novel walking direction estimator is proposed. State of the art methods to estimate the walking direction are based on the signal energy maximization and are sensitive to erratic hand movements. The new approach is based on the theories of probability and information that is built on the biomechanical description of walking. Experimental validations are conducted to analyze the performance of the new direction estimation filter whose quality directly depends on the quality of the position estimates

No description available.

No description available.

Thesis (M.S.)--Auburn University, 2007. / Abstract. Includes bibliographic references (ℓ. 114-116)

Thesis (Ph. D.)--University of Wisconsin, 1965. / Typescript. Vita. eContent provider-neutral record in process. Description based on print version record. Includes bibliography.

abstract: The advancements in the technology of MEMS fabrication has been phenomenal in recent years. In no mean measure this has been the result of continued demand from the consumer electronics market to make devices smaller and better. MEMS inertial measuring units (IMUs) have found revolutionary applications in a wide array of fields like medical instrumentation, navigation, attitude stabilization and virtual reality. It has to be noted though that for advanced applications of motion tracking, navigation and guidance the cost of the IMUs is still pretty high. This is mainly because the process of calibration and signal processing used to get highly stable results from MEMS IMU is an expensive and time-consuming process. Also to be noted is the inevitability of using external sensors like GPS or camera for aiding the IMU data due to the error propagation in IMU measurements adds to the complexity of the system. First an efficient technique is proposed to acquire clean and stable data from unaided IMU measurements and then proceed to use that system for tracking human motion. First part of this report details the design and development of the low-cost inertial measuring system ‘yIMU’. This thesis intends to bring together seemingly independent techniques that were highly application specific into one monolithic algorithm that is computationally efficient for generating reliable orientation estimates. Second part, systematically deals with development of a tracking routine for human limb movements. The validity of the system has then been verified. The central idea is that in most cases the use of expensive MEMS IMUs is not warranted if robust smart algorithms can be deployed to gather data at a fraction of the cost. A low-cost prototype has been developed comparable to tactical grade performance for under $15 hardware. In order to further the practicability of this device we have applied it to human motion tracking with excellent results. The commerciality of device has hence been thoroughly established. / Dissertation/Thesis / Masters Thesis Mechanical Engineering 2016