Search

Multirotor aerial platforms have obtained growing attentions in industry and academia, for its simplicity in mechanical structure, agility in maneuverability and ability for vertical take-off and landing (VTOL). Conventional multirotor has underactuated dynamics, and can not be fully controlled in 6 Degree-of-Freedom (DoF). In fact, only its three-dimensional position and yaw angle, called the flat outputs, can be controlled independently. However, for certain applications, such as perching on a vertical vertical wall or flying in a narrow space, the the non-flat outputs, the roll and pitch angles, are independently specified from the position requirements at some particular time. These tasks require the independent control of position and attitude at least partially for certain instants, and are generally challenging for multirotor platforms.This dissertation addresses this issue in two aspects. Firstly, an algorithm is designed for the conventional quadcopter platforms to generate trajectories for tasks with requirements on both position and attitude. It is formulated as an optimization, and converted into a series of convex problems to solve. Constraints on dynamics, space limitations, inputs and states are explicitly included. The algorithm is verified numerically on the task of quadcopter perching at the specified location on a vertical wall.Secondly, a fully actuated multirotor aerial platform is proposed. Commercial quadcopters and passive hinges are used to generate tiltable thrust vectors during flight. This platform has a salient feature for mechanical simplicity, as it does not require additional actuators to control the directions of thrust vectors. A controller for the proposed multirotor platform is designed to enable independent control of position and attitude.The proposed multirotor platform has overactuation in dynamics, which renders a redundancy of 2 DoF for inputs. A new controller is proposed, under which the input allocation scheme searches within this redundancy for smaller thrust forces required to hover at different attitudes. The range of achievable attitudes is enlarged under this new scheme compared with the previously proposed controller, under the same thrust saturation limit for the platform actuators. These controllers are validated with both simulation and experiments and demonstrated by the proposed multirotor aerial platform hovering at non-horizontal attitudes, or tracking independent trajectories for position and attitude simultaneously.

International audience; The internal forces and torques arising in cooperative manipulators ensembles are the grasp forces/torques and it is obviously desirable to control them. We present a novel approach that describes the internal loading as the interaction forces/torques arising in a multi-body system formed by multiple manipulators that behave like a formation of robots. We show that these quantities belong to the null space of the grasp matrix, thus they do not affect the dynamics of the object. The main contribution of this paper is a decoupling control scheme for tracking the internal and the motion-inducing forces and torques in a physically consistent way. The scheme is based on a physically and mathematically consistent model of the dynamics of the constrained interaction.

From a broad perspective, this thesis aims at exploring the extent to which microeconomic heterogeneity shapes the trends and fluctuations of aggregate outcomes, by focusing on bank credit, productivity, and the interaction between these two variables.The first part of the thesis is motivated by the weakness of the total factor productivity (TFP) growth observed post-crisis in most developed countries. It examines the evolution and characteristics of resource misallocation in the French manufacturing sector before, during, and after the Great Recession. The inefficiency of the input allocation dampened productivity growth in the lead-up to the crisis. It also accounts for a sizeable part of the disruptions observed during the Great Recession, with the interplay between labor and capital misallocation playing a major role. On the other hand, the post crisis slowdown appears to be mostly driven by the sluggishness of the firm-level TFP growth, rather than by a worsening of resource misallocation.The second part of the thesis examines how the granular structure of the loan distribution in France shapes the cyclicality of aggregate bank credit lent to non-financial corporations. Microeconomic credit shocks affecting the largest borrowers largely drive this comovement, while bank individual shocks do not contribute significantly. It suggests that at the macro level mechanisms specific to the granular borrowers dominate both the effect of the financial frictions constraining smaller firms and the bank lending channel. The high level of concentration on the borrower side also affects bank liquidity flows: it leads credit line takedowns to be less diversifiable and more synchronized.The third part of the thesis relates input allocation to credit allocation. It suggests that the propensity of banks to lend to healthy firms was significantly reduced during both the 2007-2009 crisis and the Eurozone crisis. As bank lending shocks affect firm-level real outcomes, this reduction contributed to decrease the investment gap between high-quality and low-quality firms, thereby directing capital input towards companies that were more risky and less productive. The surge in capital misallocation observed in time of crisis may therefore reflect disruptions affecting credit allocation.; De manière générale, cette thèse vise à évaluer dans quelle mesure l'hétérogénéité microéconomique influence les tendances et fluctuations des agrégats macroéconomiques, en particulier le crédit bancaire, la productivité et l'interaction entre ces deux variables.La première partie de la thèse est motivée par la faiblesse de la croissance de la productivité globale des facteurs observée post crise dans la plupart des pays développés. Cette partie étudie l'évolution et les caractéristiques de l'allocation des ressources dans le secteur manufacturier français avant, pendant et après la Grande Récession. L'inefficacité de l'allocation des facteurs a freiné la croissance de la productivité au cours de la décennie qui a précédé la crise. Elle explique également une part significative des fluctuations observées pendant la Grande Récession, l'interaction entre les inefficiences de l'allocation du facteur capital et du facteur travail jouant un rôle majeur. Par ailleurs, le ralentissement post crise semble être principalement dû à l'atonie de la croissance de la productivité individuelle des firmes, plutôt qu'à une détérioration de l'efficacité de l'allocation des ressources.La deuxième partie de la thèse étudie comment la structure granulaire de la distribution des prêts aux entreprises en France façonne le caractère cyclique du crédit bancaire agrégé. Les chocs de crédit microéconomiques affectant les plus gros emprunteurs sont pour une large part à l'origine de cette corrélation, alors que les chocs individuels spécifiques aux banques n'y contribuent pas de manière significative. Cela suggère qu'au niveau macroéconomique les mécanismes propres aux emprunteurs granulaires dominent l'effet des frictions financières qui pourraient contraindre les entreprises plus petites, ainsi que le canal du bilan des banques. La forte concentration observée dans la distribution des emprunteurs affecte également les flux de liquidité des banques: elle restreint la diversification et conduit à une synchronisation accrue des lignes de crédit.La troisième partie de la thèse relie la répartition des facteurs à l'allocation du crédit. Cette partie suggère que la propension des banques à prêter à des entreprises saines a été significativement réduite tant pendant la crise de 2007-2009 que pendant la crise de la zone Euro. Les chocs bancaires impactent l'activité réelle des entreprises; cette réduction a ainsi contribué à diminuer l'écart d'investissement entre les entreprises de bonne qualité et celles de qualité inférieure, ce qui a eu tendance à orienter le facteur capital vers des firmes qui étaient plus risquées et moins productives. L'augmentation soudaine des inefficiences liées à une mauvaise allocation du capital observée en temps de crise peut donc refléter des perturbations affectant l'allocation du crédit.

From a broad perspective, this thesis aims at exploring the extent to which microeconomic heterogeneity shapes the trends and fluctuations of aggregate outcomes, by focusing on bank credit, productivity, and the interaction between these two variables.The first part of the thesis is motivated by the weakness of the total factor productivity (TFP) growth observed post-crisis in most developed countries. It examines the evolution and characteristics of resource misallocation in the French manufacturing sector before, during, and after the Great Recession. The inefficiency of the input allocation dampened productivity growth in the lead-up to the crisis. It also accounts for a sizeable part of the disruptions observed during the Great Recession, with the interplay between labor and capital misallocation playing a major role. On the other hand, the post crisis slowdown appears to be mostly driven by the sluggishness of the firm-level TFP growth, rather than by a worsening of resource misallocation.The second part of the thesis examines how the granular structure of the loan distribution in France shapes the cyclicality of aggregate bank credit lent to non-financial corporations. Microeconomic credit shocks affecting the largest borrowers largely drive this comovement, while bank individual shocks do not contribute significantly. It suggests that at the macro level mechanisms specific to the granular borrowers dominate both the effect of the financial frictions constraining smaller firms and the bank lending channel. The high level of concentration on the borrower side also affects bank liquidity flows: it leads credit line takedowns to be less diversifiable and more synchronized.The third part of the thesis relates input allocation to credit allocation. It suggests that the propensity of banks to lend to healthy firms was significantly reduced during both the 2007-2009 crisis and the Eurozone crisis. As bank lending shocks affect firm-level real outcomes, this reduction contributed to decrease the investment gap between high-quality and low-quality firms, thereby directing capital input towards companies that were more risky and less productive. The surge in capital misallocation observed in time of crisis may therefore reflect disruptions affecting credit allocation.; De manière générale, cette thèse vise à évaluer dans quelle mesure l'hétérogénéité microéconomique influence les tendances et fluctuations des agrégats macroéconomiques, en particulier le crédit bancaire, la productivité et l'interaction entre ces deux variables.La première partie de la thèse est motivée par la faiblesse de la croissance de la productivité globale des facteurs observée post crise dans la plupart des pays développés. Cette partie étudie l'évolution et les caractéristiques de l'allocation des ressources dans le secteur manufacturier français avant, pendant et après la Grande Récession. L'inefficacité de l'allocation des facteurs a freiné la croissance de la productivité au cours de la décennie qui a précédé la crise. Elle explique également une part significative des fluctuations observées pendant la Grande Récession, l'interaction entre les inefficiences de l'allocation du facteur capital et du facteur travail jouant un rôle majeur. Par ailleurs, le ralentissement post crise semble être principalement dû à l'atonie de la croissance de la productivité individuelle des firmes, plutôt qu'à une détérioration de l'efficacité de l'allocation des ressources.La deuxième partie de la thèse étudie comment la structure granulaire de la distribution des prêts aux entreprises en France façonne le caractère cyclique du crédit bancaire agrégé. Les chocs de crédit microéconomiques affectant les plus gros emprunteurs sont pour une large part à l'origine de cette corrélation, alors que les chocs individuels spécifiques aux banques n'y contribuent pas de manière significative. Cela suggère qu'au niveau macroéconomique les mécanismes propres aux emprunteurs granulaires dominent l'effet des frictions financières qui pourraient contraindre les entreprises plus petites, ainsi que le canal du bilan des banques. La forte concentration observée dans la distribution des emprunteurs affecte également les flux de liquidité des banques: elle restreint la diversification et conduit à une synchronisation accrue des lignes de crédit.La troisième partie de la thèse relie la répartition des facteurs à l'allocation du crédit. Cette partie suggère que la propension des banques à prêter à des entreprises saines a été significativement réduite tant pendant la crise de 2007-2009 que pendant la crise de la zone Euro. Les chocs bancaires impactent l'activité réelle des entreprises; cette réduction a ainsi contribué à diminuer l'écart d'investissement entre les entreprises de bonne qualité et celles de qualité inférieure, ce qui a eu tendance à orienter le facteur capital vers des firmes qui étaient plus risquées et moins productives. L'augmentation soudaine des inefficiences liées à une mauvaise allocation du capital observée en temps de crise peut donc refléter des perturbations affectant l'allocation du crédit.

International audience; In this paper we consider fully-actuated and redundantly-actuated robots, whose saturated inputs can have high bandwidth or can be slowly varying (with dynamics). The slowly varying inputs can be considered as configurations for the system. The proposed strategy allows to find the optimal actuators' configuration to optimize a cost function as the manipulability or the energy consumption. The approach allows for both a static design, which can include actuators' parameters such as position, orientation, saturations, numbers of actuators, and for a dynamic design, where the configurations can be controlled by an input of the system. A generalized solution to the optimal problem is proposed with the use of genetic algorithms. The results are validated in two simulation scenarios: a reconfiguration of the actuators orientation of an redundantly-actuated planar robot for trajectory tracking and the design optimization of the orientation of the motors in a generalized hexa-rotor with arbitrary propeller orientation.

Efficiency analysis is used for assessing links between technical efficiency (TE) of livestock farms and animal diseases. However, previous studies often do not make the link with the allocation of inputs and mainly present average effects that ignore the often huge differences among farms. In this paper, we studied the relationship between exposure to gastrointestinal (GI) nematode infections, the TE and the input allocation on dairy farms. Although the traditional cost allocative efficiency (CAE) indicator adequately measures how a given input allocation differs from the cost-minimising input allocation, they do not represent the unique input allocation of farms. Similar CAE scores may be obtained for farms with different input allocations. Therefore, we propose an adjusted allocative efficiency index (AAEI) to measure the unique input allocation of farms. Combining this AAEI with the TE score allows determining the unique input-output position of each farm. The method is illustrated by estimating efficiency scores using data envelopment analysis (DEA) on a sample of 152 dairy farms in Flanders for which both accountancy and parasitic monitoring data were available. Three groups of farms with a different input-output position can be distinguished based on cluster analysis: (1) technically inefficient farms, with a relatively low use of concentrates per 100 l milk and a high exposure to infection, (2) farms with an intermediate TE, relatively high use of concentrates per 100 l milk and a low exposure to infection, (3) farms with the highest TE, relatively low roughage use per 100 l milk and a relatively high exposure to infection. Correlation analysis indicates for each group how the level of exposure to GI nematodes is associated or not with improved economic performance. The results suggest that improving both the economic performance and exposure to infection seems only of interest for highly TE farms. The findings indicate that current farm recommendations regarding GI nematode infections could be improved by also accounting for the allocation of inputs on the farm.

International audience; Considering the most widely spread configuration of actuators for low orbit satellites, namely a set of reaction wheels and a set of magnetorquers, we revisit the classical " cross product control law " solution for achieving attitude stabilization and momentum dumping. We show how the classical solution has a quasi-cascade structure that, under a suitable input-to-state (ISS) assumption, can be stabilized by high gain, thereby making the actuators more inclined to saturate. Motivated by this, we propose a revisited version of this control law that transforms the quasi-cascade into a real cascade. Then we show that both strategies are such that the attitude control is affected by the momentum dumping, and that they both require a suitable ISS property. To overcome these drawbacks, we propose a new allocation-based controller which makes the attitude dynamics completely independent of the momentum dumping and induces global asymptotic stability without any ISS requirement. Several formal statements and simulation results support our discussions and highlight the pros and cons of the different control strategies.

The Current Limit Avoidance System (CLA) has been recently deployed at the JET tokamak to avoid current saturations in the poloidal field (PF) coils when the eXtreme Shape Controller is used to control the plasma shape. In order to cope with the current saturation limits, the CLA exploits the redundancy of the PF coils system to automatically obtain almost the same plasma shape using a different combination of currents in the PF coils. In the presence of disturbances it tries to avoid the current saturations by relaxing the constraints on the plasma shape control. The CLA system has been successfully implemented on the JET tokamak and fully commissioned in 2011. This paper presents the first experimental results achieved in 2011–2012 during the restart and the ITER-like wall campaigns at JET.

This article aims to show that if the farm sample under consideration is characterized by significant heterogeneity, recovering output-specific input uses or modeling farm-level input uses may be more complicated than previously recognized. The article theoretically investigates the problems implied by farm heterogeneity in input allocation equations and multicrop econometric models, proposes an empirically tractable solution for overcoming these problems, and assesses the empirical relevance of the theoretically identified problems. The results demonstrate that heterogeneity matters, but that the revealed heterogeneity bias has a limited impact on the empirical results. Copyright 2012, Oxford University Press.

In this paper we employ and extend the dynamic allocation theory first presented in a recent work by the last author, L. Zuccarian, to guarantee asymptotic tracking of a prescribed elongation of the plasma cross section in the Frascati Tokamak Upgrade (FTU). This task is hard to accomplish because it can only be achieved using the so-called poloidal coil, a high bandwidth actuator needed to perform high performance horizontal plasma position regulation. Another actuator, the poloidal coil, could be used for this positioning task, but its bandwidth is insufficient. Using dynamic input allocation it is possible to hierarchically achieve the two goals above by way of a nonlinear scheme suitably exploiting the two actuators: the high priority (fast) goal is the horizontal position regulation task, while the low priority (slow) goal is the elongation regulation. We present theoretical results supporting the proposed scheme, as well as simulations and experiments showing the effectiveness of the proposed solution.

China’s rural reforms expose farm households to the risk of administrative land reallocation and adjustment. The possibility of land reallocation gives rise to the problem of tenure insecurity which reduces farm households' incentives to invest in the land and to use the labor forces efficiently and hence negatively affect farmers' income. In this study, the normalized quadratic profit function is used to analyze profit maximization problems in farm households in the Zhejiang and Hubei provinces of China from 1995 to 2002. Additional variables have been introduced to capture the effects of a series of institutional environment and factor market constraints, including land insecurity, crop cultivation structure, labor input and capital input allocations between agricultural and non-agricultural productions. Our results indicate that, although the official controls on rural labor mobility have been relaxed, the rural labor market has not yet reached the optimal level, and a less-than-optimal level of labor input is devoted to non-agriclutural activities for farm households in both provinces. Furthermore, the negative effects of land tenure insecurity on farmhousehold income through the interactions with the other three input allocations are observed in the Hubei province.

Cet article propose l'utilisation de la méthode de l'Entropie Maximale Généralisée (GME) pour estimer la répartition des inputs (et des coûts de production) entre différents produits en utilisant des sources de données hétérogènes (données comptables agricoles et données de l'enquête pratiques culturales). L'objectif est d'explorer le rôle d'information préalable fiable (bien-définie) dans l'amélioration de la précision de l'estimation du GME. La performance de la méthode GME est comparée par la suite à une approche bayésienne -- Haute Densité Postérieure (HPD)-- afin d'évaluer leur performance lorsque d'information préalable fiable est utilisée et d'examiner leur utilité pour concilier des sources de données hétérogènes. Les deux approches sont appliquées à un réseau de données comptables agricoles qui contient des informations sur la répartition des inputs entre différents produits. Les résultats d'estimation montrent que l'utilisation d'information préalable fiable provenant de source externe à l'échantillon améliore les estimations GME même si cette performance n'est pas toujours significative. Il apparaît également que l'approche bayésienne (HPD) pourrait être une bonne alternative à l'estimateur GME. HPD fournit des résultats qui sont proches de la méthode GME avec l'avantage d'une mise en œuvre simple et transparente de l'information préalable.

This paper proposes and applies an innovative estimation approach for farm group programming models using Generalised Maximum Entropy. The proposed set-up simultaneously determines calibrating cost function parameters and input allocations to production activities. The methodology is applied to Belgium FADN data of arable farms for which available input allocations allow to validate the estimation approach. Results outperform separate estimates of input allocations previously applied in the literature and this finding is robust with respect to support point design.

Ces travaux ont également fait l'objet d'une présentation lors du 2009 Annual Meeting, July 26-28, 2009, Milwaukee, Wisconsin; voir la notice suivante: http://prodinra.inra.fr/record/37368; Standard farm-level data used for crop choice modelling are farm accountancy data providing detailed data with respect to production (crop areas and production) but only providing variable input use (pesticides, fertilizers, …) data at the farm level. Allocation of variable input uses to crops appears useful for several reasons: it is necessary to compute gross margins at the crop level and may be used to investigate the empirical validity of multi-crop econometric models. The contribution of this research is threefold. First it shows that the standard regression based approaches for allocating variable input uses to crops are potentially biased due to the (partial) simultaneity of the (expected) crop variable input and acreage choices. Second, it proposes a econometric multi-crop model which is consistent with respect to its deterministic and random parts, for defining these biases and for providing potential solutions. Third, it proposes approaches to eliminate these biases. These approaches are based on the use of correction or control functions (Imbens and Wooldridge 2007; Wooldridge 2008).

Standard farm-level data used for crop choice modelling are farm accountancy data providing detailed data with respect to production (crop areas and production) but only providing variable input use (pesticides, fertilizers, …) data at the farm level. Allocation of variable input uses to crops appears useful for several reasons: it is necessary to compute gross margins at the crop level and may be used to investigate the empirical validity of multi-crop econometric models. The contribution of this research is threefold. First it shows that the standard regression based approaches for allocating variable input uses to crops are potentially biased due to the (partial) simultaneity of the (expected) crop variable input and acreage choices. Second, it proposes a econometric multi-crop model which is consistent with respect to its deterministic and random parts, for defining these biases and for providing potential solutions. Third, it proposes approaches to eliminate these biases. These approaches are based on the use of correction or control functions (Imbens and Wooldridge 2007; Wooldridge 2008)., Replaced with revised version of paper 07/10/09.

Diplôme : Dr. d'Université; L’objectif de cette thèse est de proposer une modélisation des choix des agriculteurs qui soit à la fois cohérente avec la théorie économique et utilisable dans le cadre de collaborations avec des agronomes. Cette thèse apporte trois contributions principales dans la modélisation micro-économétrique des choix de production des agriculteurs. L’utilisation de l’approche primale constitue une première originalité de ce travail de thèse. L’approche primale, ou approche par les fonctions de production, permet d’intégrer de façon simple des contraintes de type agronomique et par conséquent de produire des résultats lisibles et facilement interprétables par des non-économistes. Une deuxième originalité de ce travail de thèse est de combiner l’approche micro-économétrique et la programmation mathématique (positive), et en particulier la représentation implicite des contraintes agronomiques, de capital et de travail, à travers une fonction de coût implicite. La troisième originalité de ce travail de thèse réside dans le traitement des termes aléatoires. Les modèles proposés contiennent des termes aléatoires qui font partie intégrante de la spécification choisie. Ils représentent soit l’hétérogénéité des conditions de production entre les exploitations, soit les aléas inhérents à la production agricole (aléas climatiques ou sanitaires).

When building an economic model for supply analysis the aim is to model a decision making process of one or more agents which fits the observed practice as good as possible. Hereby the modeller is often confronted with incomplete information about the production process; particular crop specific input data are rarely available. The problem of defining activity related technology inputs coefficients is not new. A good deal of literature comes from the mathematical programming perspective, where input coefficients were estimated using a standard linear regression function to fully represent the mathematical program. However this approach is a pure technical device and may result in an inconsistent model. The author of the paper wants to investigate whether it is possible, employing proper estimation techniques, to simultaneously estimate all unknown coefficients of a mathematical farm supply model. This includes the estimation of parameters of the non linear cost function, used to calibrate and catch the simulation behaviour and the crop specific input coefficients. It is shown that a simultaneous estimation of all parameters improves the goodness of fit of the estimated parameters and that such an approach is technically feasible.