La connaissance du profil et des attributs de la route est un élément critique dans le développement des assistances à la conduite. Ceci est notamment le cas pour les applications d’alerte et d’aide au maintien de voie. Les résultats des travaux de perception, vidéo notamment, permettent uniquement une localisation relative du véhicule par rapport à la route. Ce type de localisation ne permet pas l’accès au positionnement absolu de la route. A ce titre, il est impossible d’avoir accès au devers de la route via une seule caméra vidéo montée en vision frontale. Pour surmonter cette difficulté, nous proposons dans cette thèse une approche de numérisation et de reconstruction 3D de la géométrie de la route (pente, devers et courbure) dans un repère absolu supposé galiléen. L’idée générale de la numérisation 3D de la route, consiste à localiser le véhicule en 3D, faire le recalage pour obtenir celui du centre de la voie et d’associer à chaque positionnement les valeurs numériques des différents attributs de la route. L’approche développée consiste en la fusion d’informations provenant d’une part de capteurs proprioceptifs et extéroceptifs et d’autre part d’observateurs. On y distingue trois phases principales : La première a pour objet l’estimation des attributs de la route à l’aide d’observateurs. La deuxième est dédiée au positionnement relatif de la route par rapport au véhicule à partir de capteurs de vision embarqués sur le véhicule. La dernière phase aboutit à la reconstruction 3D de la route par fusion des informations provenant des deux premières phases. Techniquement les états des véhicules et les attributs de la route sont reconstruit à l’aide de filtres de Kalman étendus et d’observateurs à entrées inconnues. Par la suite, la géométrie 3D de la route, dans le repère absolu, est obtenue en calculant la matrice de passage du repère relatif, lié aux capteurs embarqués, au repère absolu. On utilise pour cela un schéma de fusion avec les données issues du capteur vidéo et les données GPS en utilisant les techniques IMM (Interacting Multiple Models). / The knowledge of the profile and the road attributes is a critical item in the development of driving assistances. This is specially the case for alert applications and lateral control. The results of the perception studies, specially the video, only allow a relative localization of the vehicle on the road. This kind of localization does not allow the access to absolute positioning of the road. For this reason, it is impossible to have access to the road bank via only one video camera fixed in frontal view. To overcome this difficulty, we propose in this thesis an approach of digitalization and 3Dreconstruction of the geometry of the road (slope, road bank and curvature) in an absolute frame, supposed Galilean. The general idea of 3D-digitalization of the road consist in localizing the vehicle in 3D, doing the correction to obtain the one of the road center and to associate, to each positioning, the numerical values of the different attributes of the road. The developed approach consists in the fusion of information coming from proprioceptive and exteroceptive sensors on one hand and observers on the other. We find in it three principal cases: The first one has for object the estimation of the road attributes via observers. The second one is dedicated to the relative positioning of the road beside the vehicle via embedded video sensors in the vehicle. The last stage leads to the 3D-reconstruction of the road via fusion of information coming from the two first stages. Technically, the states of the vehicles and the road attributes are reconstructed by application of the extended Kalman filters and proportional integral observers with unknown inputs. Afterwards, the 3D-geometry of the road, in the absolute landmark, is obtained by calculating the transfer matrix from the relative frame, linked to embedded sensors, to the absolute frame. For that, we use a fusion scheme with the video sensors and GPS data by using the IMM techniques.

Thesis (M.S.)--West Virginia University, 2005. / Title from document title page. Document formatted into pages; contains vii, 96 p. : ill. (some col.). Includes abstract. Includes bibliographical references (p. 77-79).

Conventional wisdom dictates that imaging hardware should be optimized by use of an ideal observer (TO) that exploits full statistical knowledge of the class of objects to be imaged, without consideration of the reconstruction method to be employed. However, accurate and tractable models of the complete object statistics are often difficult to determine in practice. Moreover, in imaging systems that employ compressive sensing concepts, imaging hardware and (sparse) image reconstruction are innately coupled technologies. We have previously proposed a sparsity-driven ideal observer (SDIO) that can be employed to optimize hardware by use of a stochastic object model that describes object sparsity. The SDIO and sparse reconstruction method can therefore be "matched" in the sense that they both utilize the same statistical information regarding the class of objects to be imaged. To efficiently compute SDIO performance, the posterior distribution is estimated by use of computational tools developed recently for variational Bayesian inference. Subsequently, the SDIO test statistic can be computed semi-analytically. The advantages of employing the SDIO instead of a Hotelling observer are systematically demonstrated in case studies in which magnetic resonance imaging (MRI) data acquisition schemes are optimized for signal detection tasks.

No description available.

Monitoring the limiting reactant concentration is a key question to maximize the productivity and to guarantee the safety of exothermic fed-batch processes. However in most applications, the concentration cannot be measured in real-time since suitable devices do not exist or are too expensive; the concentrations are then measured by off-line analyses. In this context monitoring the concentrations via software sensors, or state observer based estimators, is an attractive option. The presence of model uncertainties is a major limitation when applying state observers to real processes. More precisely, in fed-batch exothermic reactors the bad knowledge of both the reaction kinetics and the heat transfer may prevent the use of classical observers. In this study, we propose two different approaches to estimate the concentration of the limiting reactant in a class of single phase exothermic fed-batch reactors with uncertain kinetics and heat transfer. The first approach is based on a finite time converging observer that provides an estimate for the reaction rate via the reactor energy balance equation. The concentration is then computed from the reaction rate estimate via a material balance equation. The main contribution of this approach is the use of a finite time observer to limit the reconstruction error by guaranteeing a small convergence time interval for the reaction rate estimate. The second approach is based on an interval observer that provides two bounds for the concentration by considering uncertainties related to both the heat transfer and the reaction kinetics. The final estimate is then computed as the mean of the bounds. A systematic tuning procedure has been developed for each of both estimation techniques. Both estimators have then been tested and validated with real data coming from the production of different kinds of resins carried out in 10 tons reactors.

A fault in the system operation is deemed to occur when the system practically experiences an abnormal condition, such as a malfunction in the actuators/sensors. Hence, detection and isolation of the faulty components is crucial in control applications. Effective control and monitoring of a system requires accurate information of internal behaviour of the system. This internal behaviour can be analyzed by system's states. Practically, in many real systems, state space variables are not fully available for measurements. The two critical problems stated have motivated significant research work in the area of robust state and fault estimation. Fault reconstruction and estimation is regarded as a stronger extension to fault detection and isolation (FDI) since accurate fault estimation automatically implies fault detection. It is well known that two promising control strategies to cope with uncertain control processes are H_infinity Control and Sliding Mode Control. Therefore, in this PhD thesis, we employ these tools and we propose observer based robust fault reconstruction (RFR) by integrating H_infinity filtering and Sliding Mode Control. We also employ adaptive control on the sliding motion to deal with faults with unknown bounds. Another open problem in the context of FDI and RFR is due to systems with multiple faults at different system's components since it is often the case where actuators and also sensors suffer from faults during the course of the system's operation. Both actuators and sensors can suffer from faults either alone, at separate times or simultaneously. The co-existence of unknown fault at both sensor(s) and actuator(s) has not been addressed in any earlier design of fault reconstruction schemes. In this Thesis, inspired by the theory of singular systems, we aim at solving this problem. A New structure for reduced-order unknown input observers (UIOs) with application to chaotic communication and sensor fault reconstruction is also proposed. / Controls

Successful Fisheries Management should be constructed upon accurate data to evaluate the fisheries stock correctly. Meanwhile, efficient monitoring system is required to oversee the execution of policies. Observer Program which is the most effective method to understand the data of fisheries was promoted since 2001 in Taiwan. However, due to the different values, roles, living habits between observers and captains, they still have conflicts. Amount 50 scientific observers, 25 randomly selected for the survey during October to December in 2008. The interview focus on five major subjects: (1) report of fishing activities; (2) Bycatch; (3) Biological sample collection; (4) Living area; (5) Daily schedule. The conflict factors and the conflict points between observers and captains were discussed (with qualitative and quantitative methods) in the research. It reveals that the most significant element leads to conflicts is relating Bycatch¡F88% (22 persons) of the reviewed observers reported conflict in this issue, and the main reason is the profits of shark fin. Besides, there are 20 observers, which is 80% of the sample, mentioned that Logbook is another factor leads to arguments of commercial benefits. Furthermore, 16 persons (64%) also agree that Biological sample collections is another factor; the main reason is the otolieh sampling will destroy the price catch. Finally, it shows that few observers chose Activity area and Living habits, which only have 12% (3 person) and 24% (6 person) respectively. In addition, in the factors of Logbook and Bycatch, the conflict level is effected by educational degree of observers. This research also investigates the opinions of observers, which could improve the relationship within the industry. To sum up, the study has reviewed and integrated available researches to deal with conflicts in Fisheries management. The findings of this research could be applied in setting up fisheries policies.

Thesis (Ph. D.)--University of Missouri-Columbia, 2003. / Typescript. Vita. Includes bibliographical references (leaves 209-215). Also available on the Internet.

Thesis (Ph. D.)--University of Missouri-Columbia, 2003. / Typescript. Vita. Includes bibliographical references (leaves 209-215). Also available on the Internet.

Thesis (Ph. D.)--West Virginia University, 2003. / Title from document title page. Document formatted into pages; contains xi, 172 p. : ill. (some col.). Includes abstract. Includes bibliographical references.