The aim of this diploma thesis is to create a comprehensive study of measurement accuracy of coordinate measuring machine (hereinafter CMM) Zeiss UPMC 850 Carat S-ACC (hereinafter Zeiss Carat) for the needs of the national metrological institute in Brno in accordance with ČSN EN ISO / IEC 17025 and the follow-up system standards of the ČSN EN ISO 10360 series. Additionally, it includes the creation of instructions for the calculation of measurement uncertainty, which will be put into effect in an accredited calibration laboratory. The first part of the work focuses on the description of the current state of knowledge and the definition of basic concepts in the field of metrology and accurate measurements on CMM. The second part describes the Zeiss Carat measuring machine, identifies the individual contributors to the resulting measurement uncertainty and defines the methodology for their quantification. The last part deals with the evaluation of calibration data and the calculation of the expanded measurement uncertainty of the Zeiss Carat instrument, which is used to quantify its accuracy.

Avec l’augmentation progressive de la population dans les grandes villes, comme Paris, nous prévoyons d’ici 2050 une augmentation de 50% du trafic routier. En considérant les embouteillages et la pollution que cette augmentation va générer, on voit clairement la nécessité de nouveaux système de mobilité plus durables, comme le covoiturage, ou plus généralement toute la mobilité partagée. En parlant de mobilité partagée, ce n’est pas seulement le partage de trajets de personnes qui ont le même itinéraire au même temps, elle inclut aussi les marchandises.Cette thèse aborde le défi de la synchronisation des flux de passagers et de marchandises dans les systèmes de mobilité urbaine et elle vis à développer des méthodes d’optimisation pour que cette synchronisation dans la mobilité partagée soit faisable. Plus précisément, elle aborde les questions de recherche suivantes:*Q1: Quelles sont les variantes des systèmes de mobilité partagée et comment les optimiser?*Q2: Comment synchroniser les déplacements de personnes et quels gains cette synchronisation peut-elle générer?*Q3: Comment combiner les flux de passagers et de fret et quels sont les avantages attendus?*Q4: Quels sont les effets de l'incertitude sur la planification et l'exploitation de systèmes de mobilité partagée?Dans un premier temps, nous étudions les différentes variantes des systèmes de mobilité partagée et nous les classifions en fonction de leurs modèles, caractéristiques, approches de résolution et contexte d'application. En se basant sur cette revue de littérature, nous identifions deux problèmes de mobilité partagés, que nous considérons en détails dans cette thèse et nous développons des méthodes d'optimisation pour les résoudre.Pour synchroniser les flux de passagers, nous étudions un modèle de covoiturage en utilisant les véhicules autonomes, personnels et partagés, et des points de rencontre où la synchronisation entre passagers peut avoir lieu. Pour cela, une méthode heuristique en deux phases est proposée et une étude de cas sur la ville de New York est présentée.Ensuite, nous développons un modèle d’optimisation qui combine les flux de passagers et de marchandises dans une région urbaine. Le but de ce modèle est d’utiliser les capacités disponibles sur une ligne de transport fixe pour transporter les passagers et des robots transportant des petits colis à leurs destinations finales en considérant que la demande de passagers est stochastique. Les résultats obtenus montrent que les solutions proposées par ces deux modèles peuvent conduire à une meilleure utilisation des systèmes de transport dans les régions urbaines. / The rise of research into shared mobility systems reflects emerging challenges, such as rising urbanization rates, traffic congestion, oil prices and environmental concerns. The operations research community has turned towards more sharable and sustainable systems of transportation. Although shared mobility comes with many benefits, it has some challenges that are restricting its widespread adoption. More research is thus needed towards developing new shared mobility systems so that a better use of the available transportation assets can be obtained.This thesis aims at developing efficient models and optimization approaches for synchronizing people and freight flows in an urban environment. As such, the following research questions are addressed throughout the thesis:*Q1: What are the variants of shared mobility systems and how to optimize them?*Q2: How can people trips be synchronized and what gains can this synchronization yields?*Q3: How can people and freight flows be combined and what are the intended benefits?*Q4: What impacts uncertainty can have on planning and operating shared mobility systems?First, we review different variants of the shared mobility problem where either (i) travelers share their rides, or (ii) the transportation of passengers and freight is combined. We then classify these variants according to their models, solution approaches and application context and We provide a comprehensive overview of the recently published papers and case studies. Based on this review, we identify two shared mobility problems, which we study further in this thesis.Second, we study a ridesharing problem where individually-owned and on-demand autonomous vehicles (AVs) are used for transporting passengers and a set of meeting points is used for synchronizing their trips. We develop a two-phase method (a pre-processing algorithm and a matching optimization problem) for assessing the sharing potential of different AV ownership models, and we evaluate them on a case study for New York City.Then, we present a model that integrates freight deliveries to a scheduled line for people transportation where passengers demand, and thus the available capacity for transporting freight, is assumed to be stochastic. We model this problem as a two-stage stochastic problem and we provide a MIP formulation and a sample average approximation (SAA) method along with an Adaptive Large Neighborhood Search (ALNS) algorithm to solve it. We then analyze the proposed approach as well as the impacts of stochastic passengers demand on such integrated system on a computational study.Finally, we summarize the key findings, highlight the main challenges facing shared mobility systems, and suggest potential directions for future research.

Incorporating uncertainties in the aeroelastic analysis increases the confidence levels of computational predictions and reduces the need for validation with experimental or flight test data. Helicopter rotor blades, which play a dominant role in the overall vehicle performance, are routinely made of composites. The material properties of composites are uncertain because of the variations in manufacturing process and other effects while in service, maintenance and storage. Though nominal values are listed, they are seldom accurate. In this thesis, the effect of uncertainty in composite material properties on the computational predictions of cross-sectional properties, natural frequencies, blade tip deflections, vibratory loads and aeroelastic stability of a four-bladed composite helicopter rotor is studied. The effect of material uncertainty is studied with the composite rotor blades modeled as components of soft-inplane as well as stiff-inplane hingeless helicopter rotors. Aeroelastic analysis based on finite elements in space and time is used to evaluate the helicopter rotor blade response in hover and forward flight. Uncertainty analysis is performed with direct Monte Carlo simulations based on a sufficient number of random samples of material properties. It is found that the cross-sectional stiffness parameters and natural frequencies of rotor blades show considerable scatter from their baseline predictions. The uncertainty impact on the rotating natural frequencies depends on the level of centrifugal stiffening of each mode. The propagation of material uncertainty into aeroelastic response causes large deviations from the baseline predictions. The magnitudes of 4/rev vibratory loads show deviations of 10 to 600 percent from their baseline predictions. The aeroelastic stability in hover and forward flight conditions also show considerable uncertainty in the predictions. In addition to the effects of material uncertainty, various factors influencing the propagation of material uncertainty are studied with the first-order based reliability methods. The numerical results have shown the need to consider the uncertainties in the helicopter aeroelastic analysis for reliable computational predictions. Uncertainty quantification using direct Monte Carlo simulation is accurate but computationally expensive. The application of response surface methodologies to reduce the computational cost of uncertainty analysis is studied. Response surface approximations of aeroelastic outputs are developed in terms of the composite material properties. Monte Carlo simulations are then performed using these computationally less expensive response surface models. The results of this study show that the metamodeling techniques can effectively reduce the computational cost of uncertainty analysis of composite rotor blades. In the last part of the thesis, an aeroelastic optimization method to minimize the vibration level is developed with due consideration to material uncertainty. Second-order polynomial response surfaces are used to approximate the objective function which smooths out the local minima or numerical noise in the design space. The aeroelastic optimization is carried out with the nominal values of composite material properties and the performance of final design is found to be optimum even for the perturbed values of material properties.

The central theme of the thesis is to study linear and non linear SVM formulations in the presence of uncertain observations. The main contribution of this thesis is to derive robust classfiers from partial knowledge of the underlying uncertainty. In the case of linear classification, a new bounding scheme based on Bernstein inequality has been proposed, which models interval-valued uncertainty in a less conservative fashion and hence is expected to generalize better than the existing methods. Next, potential of partial information such as bounds on second order moments along with support information has been explored. Bounds on second order moments make the resulting classifiers robust to moment estimation errors. Uncertainty in the dataset will lead to uncertainty in the kernel matrices. A novel distribution free large deviation inequality has been proposed which handles uncertainty in kernels through co-positive programming in a chance constraint setting. Although such formulations are NP hard, under several cases of interest the problem reduces to a convex program. However, the independence assumption mentioned above, is restrictive and may not always define a valid uncertain kernel. To alleviate this problem an affine set based alternative is proposed and using a robust optimization framework the resultant problem is posed as a minimax problem. In both the cases of Chance Constraint Program or Robust Optimization (for non-linear SVM), mirror descent algorithm (MDA) like procedures have been applied.