Un des freins majeurs au développement de la manipulation d'objet avec une main robotisée est sans aucun doute leur fragilité. C'est l'une des raisons pour laquelle un système bras-main anthropomorphe, extrêmement robuste, est développé au centre de robotique et de mécatronique de DLR. Le système est unique à la fois par sa complexité, utilisant 52 moteurs et plus de200 capteurs, ainsi que par ses capacités dynamiques. En effet, ce nouveau système a la particularité d'être mécaniquement flexible ce qui offre la possibilité de stocker de l'énergie à court terme et remplit ainsi deux fonctions essentielles pour un robot humanoïde: les impacts sont filtrés et les performances dynamiques sont augmentées.Dans cette thèse, on se concentre plus particulièrement sur la main. Elle dispose de 19 degrés de liberté dont chacun est actionné par deux tendons flexibles antagonistes. La rigidité des tendons étant non linéaire il est possible, tout comme peut le faire l'être humain, de co-contracter les <<muscles>> et donc d'ajuster la rigidité des doigts afin de s'adapter au mieux aux tâches à effectuer. Cependant, cette flexibilité entraine de nouveau défis de modélisation et de contrôle. L'état de l'art se concentre majoritairement sur le problème de la répartition des forces internes ou du contrôle d'articulation flexible mais peu de travaux considèrent les deux problèmes simultanément.Le travail présenté dans la première partie de la thèse se concentre sur la modélisation de la main et du poignet. Les problématiques spécifiques aux systèmes actionnés par des tendons, tels que les matrices de couplage et l'estimation du déplacement des articulations à partir du déplacement des tendons, sont étudiées.La seconde partie se concentre sur le contrôle d'articulations actionnées par des tendons flexibles antagonistes. Les problèmes de distribution des forces internes et de correction de la rigidité perçue par l'utilisateur sont présentés.Des approches de contrôle linéaire et non linéaire sont utilisées et des expériences sont réalisées pour comparer ces approches. En particulier, il est montré que le <<backstepping>>, une méthode de contrôle non linéaire peut être utilisée et permet d'obtenir le comportement d'impédance souhaité tout en garantissant la stabilité en boucle fermée. / One of the major limitations of object manipulation with a robotic hand is the fragility of the hardware.This is one of the motivations for developing the new anthropomorphic and extremely robust Hand Arm System at the robotics and mecatronics center of DLR.The system is unique in terms of complexity, with 52 motors and more than 200 sensors, and also in terms of dynamics.Indeed, the system is mechanically compliant, thus offers the possibility to store and release energy, thereby providing two essential functions: The impacts are filtered and the dynamics are enhanced.This thesis focuses on the hand. It has 19 degrees of freedom, each being actuated by two flexible antagonistic tendons. Because the stiffnes of the tendons is not linear, it is possible to adjust the mechanical stiffness of the joints, similar to the co-contraction of human muscles, in order to adapt to a task. However, the stiffness adjustability rises new challenges in modeling and control. The state of the art usually focuses on the problems of tendon-driven systems or flexible joint robots but seldomly both simultaneously.In the first part, the modeling of the hand and the wrist is conducted. Several problems specific to tendon-driven systems are presented, such as the coupling matrices and the joint position estimation based on the tendon displacement. The second part focuses on the control of a single joint actuated by two flexible tendons. The distribution of the tendon forces and the correction of the effective stiffness are reported. Linear and nonlinear approaches are used and multiple experiments are realised to compare them. The major result is that the backstepping, a nonlinear control method, can be used and provides the desired impedance behavior while guaranting closed-loop stability.

The young moving groups are collections of nearby (<200 pc), young (5-150 Myr) pre-main sequence stars; these stars offer us one of the best opportunities to characterise stellar multiplicity, sub-stellar phenomena, disc evolution and planet formation. Here we present results from a series of multiplicity studies aimed at producing comprehensive multiplicity statistics of the young moving groups. The aim was to compare the derived statistics of the young moving groups to other populations in order to investigate whether the abundance and properties of multiple systems are environment independent. We have combined high-resolution spectroscopy, AO-imaging and direct imaging to identify and characterise multiple systems across a huge range of orbital periods (1- 10e10 day). The observational techniques also allow us to constrain the abundance of multiple systems in these populations by calculating detection limits. We found many similarities (frequency of spectroscopic binaries; frequency, mass-ratio and physical separation of visual binaries) between the young moving groups and both younger and older regions, for multiple systems with physical separations smaller than 1000 au. We did, however, identify a significant number of new wide (>1000 au) companions. We reconciled the apparent excess of wide binary systems, when compared to the field population, by arguing that the wide systems are weakly bound and most likely decaying. By comparing the multiplicity statistics in one particular moving group we showed that the dynamical evolution of non-hierarchical protostars could lead to the population of wide binaries we can observe today. Our results indicate that the majority of low-mass stars form in small groups with 3 or 4 components that undergo significant dynamical evolution. The multiplicity properties of the young nearby moving groups are statistically similar to many other populations, supporting the environment-independent formation of multiple systems.

The mass of a protoplanetary disk limits the formation and future growth of any planet. Masses of protoplanetary disks are usually calculated from measurements of the dust continuum emission by assuming an interstellar gas-to-dust ratio. To investigate the utility of CO as an alternate probe of disk mass, we use ALMA to survey (CO)-C-13 and (CO)-O-18 J = 3-2 line emission from a sample of 93 protoplanetary disks around stars and brown dwarfs with masses from 0.03 to 2 M-circle dot in the nearby Chamaeleon I star-forming region. We detect (CO)-C-13 emission from 17 sources and (CO)-O-18 from only one source. Gas masses for disks are then estimated by comparing the CO line luminosities to results from published disk models that include CO freeze-out and isotope-selective photodissociation. Under the assumption of a typical interstellar medium CO-to-H-2 ratio of 10(-4), the resulting gas masses are implausibly low, with an average gas mass of similar to 0.05M(Jup) as inferred from the average flux of stacked (CO)-C-13 lines. The low gas masses and gas-to-dust ratios for Cha I disks are both consistent with similar results from disks in the Lupus star-forming region. The faint CO line emission may instead be explained if disks have much higher gas masses, but freeze-out of CO or complex C-bearing molecules is underestimated in disk models. The conversion of CO flux to CO gas mass also suffers from uncertainties in disk structures, which could affect gas temperatures. CO emission lines will only be a good tracer of the disk mass when models for C and CO depletion are confirmed to be accurate.

Visual Question Answering (VQA) models should have both high robustness and accuracy. Unfortunately, most of the current VQA research only focuses on accuracy because there is a lack of proper methods to measure the robustness of VQA models. There are two main modules in our algorithm. Given a natural language question about an image, the first module takes the question as input and then outputs the ranked basic questions, with similarity scores, of the main given question. The second module takes the main question, image and these basic questions as input and then outputs the text-based answer of the main question about the given image. We claim that a robust VQA model is one, whose performance is not changed much when related basic questions as also made available to it as input. We formulate the basic questions generation problem as a LASSO optimization, and also propose a large scale Basic Question Dataset (BQD) and Rscore (novel robustness measure), for analyzing the robustness of VQA models. We hope our BQD will be used as a benchmark for to evaluate the robustness of VQA models, so as to help the community build more robust and accurate VQA models.

M.Tech. (Chemical Engineering) / Please refer to full text to view abstract

This thesis focuses on the design and development of novel versatile diamido ligands for transition metal and main group element chemistry. The central concept of this work deal relied on the design of N, N'-disubstituted-1,8-diaminonaphthalene (H2RR’-DAN) as proligands to dianionic diamido ligand scaffolds. These ligands would then be employed for stabilization of main group element (e.g. Li, B, Al) and transition metal (e.g. Ti, Zn) compounds.

>Magister Scientiae - MSc / The Karoo in South Africa is an area in which petroleum studies are being reviewed with particular interest, especially within the Tanqua and Laingsburg subbasins of the Karoo. The particular formations of interest comprises of the Laingsburg, Fort Brown, Kookfontein and Vischkuil of the Upper Ecca group, located within the Laingsburg subbasin as well as the Kookfontein formation found within the Tanqua subbasin. The present work dealt with interpreting and distinguishing the different types of source rock lithologies, to identify the source rock zones and organic matter within the Laingsburg, Fort Brown, Kookfontein and Vischkuil formations. This was achieved by utilising different geochemical analyses (such as rock eval pyrolysis and TOC) and organic chemistry (XRD) to assess the maturity, mineralogy and the quality of the source rocks.

Les origines des Primates font l’objet d’importantes controverses. La radiation initiale des premiers Primates ainsi que leurs liens phylogénétiques précis au sein des Euarchonta (le clade incluant les Primates, Scandentia, Dermoptères et Plesiadapiformes) sont débattus. De plus, l’interprétation fonctionnelle et évolutive de certains caractères morphologiques qui définissent les Primates est incertaine. Parmi eux, se trouvent l’acquisition de capacités de préhension manuelle et pédale, avec un pied spécialisé dans la saisie et un gros orteil opposable, ainsi que des ongles remplaçant les griffes sur les phalanges distales. De ce fait, le morphotype ancestral des Primate est très étudié, bien que l’arborealité et la petite taille de nos premiers ancêtres soient consensuelles. Le but de cette thèse était de revisiter certains aspects encore flous des origines des Primates, en se concentrant sur les mécanismes de préhension de la main et du pied, à travers une approche interdisciplinaire mêlant éthologie, biomécanique, anatomie comparée et analyse phylogénétique. Un réexamen du genre Plesiadapis (Plésiadapiforme) conduit au questionnement de l’hypothèse récente concernant les relations phylogénétiques des premiers primates. De plus, une étude quantitative des postures manuelles et pédales en relation au type de support utilisé lors de la locomotion, suivie d’une analyse morphologique des métapodes et phalanges de mains et pieds, ont été conduites sur différentes espèces de Primates et non-Primates. Les résultats furent ensuite couplés de façon intégrative afin de relier les caractères morphologiques à leur fonction, tout en évaluant leur importance phylogénétique. Les résultats de ces travaux permettent de proposer des hypothèses alternatives concernant deux caractères clés chez les Primates, comme la fonction initiale des ongles : liés plutôt à une capacité sensorielle que mécanique ; ainsi que concernant le scenario environnemental qui a pu conduire à l’évolution de leurs capacités de préhension pédale : supports fins verticaux et non la niche de fines branches. Également, un nouveau type de capteur de force spatialement résolu a été créé dans l’optique de mieux caractériser les contraintes biomécaniques en jeu lors de la locomotion arboricole. Ce dernier a des applications dans différents domaines, comme la robotique. / Primate origins are subject to important controversies. The initial radiation of first Primates and their precise relationships within Euarchontans (the clade including Primates, Scandentians, Dermopterans, and Plesiadapiformes) are still debated. Moreover, the functional and evolutionary interpretation of some of the morphological characters that define Primates is still uncertain. Among them are the acquisition of manual and pedal prehensile abilities, with a specialized grasping foot bearing an opposable hallux, and nails instead of claws on the distal phalanges. Thus, the ancestral morphotype of Primates is under active investigation, despite the consensus on the arboreality and small size of our early ancestors. This PhD dissertation aimed at revisiting some blurry aspects of primate origins focusing on hand and foot grasping mechanisms, through an interdisciplinary approach blending ethology, biomechanics, comparative morphology and phylogenetics. A reappraisal of the genus Plesiadapis (Plesiadapiformes) led to question a recent hypothesis on early Primates’ phylogeny. In addition, a quantitative analysis of manual and pedal postures relatively to substrate type used during locomotion, followed by a morphological study of hand and foot metapodials and phalanges were also conducted on series of primate and non-primate species. The results were analyzed in an integrative way to relate morphological features to functional attributes, along with assessing their phylogenetic importance. Among many results, this work allowed proposing alternative hypotheses regarding two key characters of primates, the primary function of nails: more linked to sensitivity than to a mechanical advantage; and the environmental scenario that may have driven the evolution of hallucal grasping capabilities: small vertical substrates instead of the fine branch niche. Moreover, in an effort to better understand biomechanical constraints at play during arboreal locomotion, a novel spatially-resolved force sensor was created, which has potential applications in various fields such as robotics.

No description available.

This study compared four different water main failure prediction models: a statistically simple model, a statistically complex model, a statistically complex model with modifications termed the 2019 model, and an age-based model. The statistically complex models compute the probability of failure based on age, size, internal pressure, length of pipe in corrosive soil, land use, and material of the. These two values are then used to prioritize a water main rehabilitation program to effectively use the municipality’s funds. The 2019 model calculates the probability of failure and consequence of failure differently than the statistically complex model by considering corrosive soil data instead of assuming all the pipes are in highly corrosive soil and average daily traffic volume data instead of using street classifications. The statistically simple model only uses the pipe age and material for probability of failure. The age-based model relies purely on the age of the pipe to determine its probability of failure. Consequences of failure are determined by the proximity of the pipe to highly trafficked streets, critical services, pipe replacement cost, and the flow capacity of the pipe. Risk of failure score is the product of the consequence of failure score and probability of failure score. Pipes are then ranked based on risk of failure scores to allow municipalities to determine their pipe rehabilitation schedule. The results showed that the statistically complex models were preferred because results varied between all four models. The 2019 model is preferred for long-term analysis because it can better account for future traffic growth using the average daily traffic volume. Corrosive soil data did not have a significant impact on the results, which can be attributed to the relatively small regression parameter for corrosive soil. The age-based model is not recommended because results of this study shows it places a significantly high number of pipes in the high and critical risk categories compared to the other models that account for more factors. This could result in the unnecessary replacement of pipes leading to an inefficient allocation of funds. Keywords: Risk of Failure, Consequence of Failure, Probability of Failure