Search

Best student paper award ICRAT 2020; International audience; Moving object analysis is a constantly growing field with numerous concrete applications in terms of traffic understanding, prediction and simulation. While many algorithms and analytic processes exist, there are still areas of investigation with novel trajectory analysis methods. As such, the geometric information analyses data with respect to its statistical distribution along extracted dimensions. This opens new ways of gaining a better understanding of large and complex trajectory data sets while providing flexible data manipulations. In this paper, we report our investigations with the development of an interactive methodology based on the geometric information analytic process where users can analyze trajectories sets, cluster and deform them maintaining the actual statistical properties of the investigated trajectories. As a contribution, this paper shows how geometric information can provide novel support for trajectory analyses taking into account the statistical properties of the investigated clusters. We also provide recommendations of good usage of such techniques with actual examples validated by a a domain expert of air traffic flow analysis

National audience

International audience

Accurate prediction of aircraft arrival times is one of the fundamental elements for air traffic controllers to manage an optimal arrival and departure sequencing on the runway, reduce flight delays, and achieve a good collaboration with airports and airlines. In this work, we analyze the feature engineering problem to predict Aircraft Landing Time (LDT) in Extended-TMA with machine learning models. Two main contributions are highlighted in this work. First, the impact of different features in LDT prediction is investigated. Second, a machine learning prediction model is presented to predict LDT. Our case of study is the E-TMA of Singapore Changi Airport (WSSS) with a radius of $100$NM. Firstly, data analysis is conducted to check the availability of different resource data, as well as cleaning the raw trajectory data. Then, feature construction and extraction are discussed in details, machine learning prediction models are proposed to address the LDT prediction. The experimental results show that 4 sets of features play a significant impact on LDT prediction for primary runway-in-use, they are: (1) Control intent: traffic demand, current traffic density, and adjacent flow; (2) Weather: surface wind; (3) Trajectory: the position of aircraft; (4) Seasonality: parts of a day and a week. Moreover, comparing three Machine Learning algorithms, in our study case, Extra-Trees is the best prediction algorithm compared with other machine learning models in terms of Root Mean Square Error (RMSE) and Mean Absolute Error (MAE). It is also found that Machine learning models perform much better than the current operational system. In summary, two main conclusions are drawn from the present work. First, predicting the aircraft LDT is strongly correlated with the TMA density at the flight operation time. Second, feature selection with domain knowledge and expert opinions is very important, and with good features, the model is less sensitive to the choice of machine learning algorithm. Civil Aviation Authority of Singapore (CAAS) Accepted version This research is supported by the Civil Aviation Authority of Singapore under the Aviation Transformation Program.

When designing a wind farm layout, we can reduce the number of variables by optimizing a pattern instead of considering the position of each turbine. In this paper we show that when reducing the problem to only two variables defining a grid, we can gain up to 3% of energy output on simple examples of wind farms dealing with many turbines (up to 1000) while dramatically reducing computation time. To achieve these results, we compared both a genetic algorithm and a differential evolution algorithm to previous results from the literature. These preliminary results should be extended to examples involving non-rectangular farm layouts and wind distributions that may require pattern deformation variables in order to increase solution diversity.

Best paper award in the trajectory prediction track; International audience; Ground-based aircraft trajectory prediction is a major concern in air traffic control and management. A safe and efficient prediction is a prerequisite to the implementation of new automated tools. In current operations, trajectory prediction is computed using a physical model. It models the forces acting on the aircraft to predict the successive points of the future trajectory. Using such a model requires knowledge of the aircraft state (mass) and aircraft intent (thrust law, speed intent). Most of this information is not available to ground-based systems. Focusing on the climb phase, we train neural networks to predict some of the unknown point-mass model parameters. These unknown parameters are the mass and the speed intent. For each unknown parameter, our model predicts a Gaussian distribution. This predicted distribution is a predictive distribution: it is the distribution of possible unknown parameter values conditional to the observed past trajectory of the considered aircraft. Using this distribution, one can extract a predicted value and the uncertainty related to this specific prediction. Using a physical model like BADA, this distribution could be used to derive a probability distribution of possible future trajectory ([1]). This study relies on ADS-B data coming from The OpenSky Network. It contains the climbing segments of the year 2017 detected by this sensor network. The 11 most frequent aircraft types are studied. The obtained data set contains millions of climbing segments from all over the world. Using this data, we show that despite having an RMSE slightly larger than previously tested methods, the predicted uncertainty allows us to reduce the size of prediction intervals while keeping the same coverage probability. Furthermore, we show that the trajectories with a similar predicted uncertainty have an observed RMSE close to the predicted one. The data set and the machine learning code are publicly available.

International audience; This paper aims at presenting a novel approach to airline sentiment analysis processing using Twitter data. By transforming trained sentiment classifiers into regressors, the daily sentiment distribution obtained can be represented as atrimodal Gaussian Mixture leading to a simple but efficient classification algorithm. These classes can be considered as daily sentiment scores. This classification applied to passenger generated tweets and airline generated tweets for five major US airlines highlights major difference in experience between passengers and airlines. This methodology also confirms the existing gap between flight performance and passenger experience and the necessity of considering and implementing passenger centric metrics

International audience; In this paper, we detail the steps that lead to optimized trajectories according to a selected criterion, in a low dimensional space. After presenting the main techniques for optimizing flight paths, as well as methods for reducing the size of the state space, we precise the modeling of our problem. We use the Karhunen-Loève transformation, or Functional Principal Components Analysis (FPCA), as our main tool to model the state space. We also select the constraints undergone by our airplane: here, we decide only to consider the impact of the wind. For its simplicity, the Simulated Annealing (SA) is chosen in order to find the optimized trajectory. Thus, once the modeling is finished, we launch our simulations and proceed to an analysis of our results

International audience; This work aims at developing an agent-based platform that allows to model and analyze decisions made by different stakeholders in an Airport Operations Centre. We will develop a methodology combining agent-based modelling and field/lab behavioral experiments for identifying the incentives behind the decisions of the stakeholders in an Airport Operations Centre environment. Once, the causal relationships have been identified, these will be translated into an agent-based environment so, it will be possible to have a virtual environment for identifying which incentives are the best for aligning the objectives of the center, considering the diversity of objectives present in the system. The causal-relationships identified in the study will be validated with a human-in-the-loop environment already developed under the SESAR program. This study is an interdisciplinary one which integrates simulation, decision making and behavioral science in the airport operations center environment.

International audience; This paper presents a new capacity management concept where Dynamic Airspace Configuration (DAC) and Flight Centric ATC (FCA) are dynamically applied together during the Air Traffic Flow and Capacity Management (ATFCM) pretactical phase. An airspace delineation methodology is also introduced aiming at identifying when and where DAC or FCA can perform better. This methodology entails the establishment of a dynamic vertical boundary that divides the airspace in two different parts where DAC and FCA are deployed. In addition, the geometrical complexity metric has been considered to measure the traffic demand and the sector capacity as an evolution of the current use of entry counts or occupancy counts.Three 24 hours scenarios over the Hungarian airspace have been simulated in order to validate the improvements, in terms of capacity and cost-effectiveness, that the dynamic integration of both capacity management solutions may provide in comparison with the two solutions deployed separately. Results shows that when DAC and FCA are dynamically applied, a significant reduction in the number of overloads and underloads detected is achieved, what might lead to a higher capacity since more aircraft can be handled. Furthermore, a reduction in the controlling hours is also registered in this situation, what it is translated into a better cost-effectiveness solution

This paper considers an e-commerce market wherein a vertically integrated marketplace competes downstream with a single retailer and upstream with an independent parcel delivery operator. Because of the information collected by the marketplace on customers' habits and preferences, the integrated parcel delivery operator has lower delivery costs than its competitor. Products are differentiated according to the retailer and the parcel operator who delivers them. The representation of product di§erentiation is inspired by the Anderson, De Palma and Thisse (2002) discrete choice model. We study several scenarios each representing a speciÖc policy implemented to regulate the marketplace. The Örst one is a data sharing policy. The integrated marketplace has to share its information with the other delivery operator which in turn will lower this operatorís cost of delivering the marketplaceís product. The second one is vertical separation under which the parcel delivery operator previously owned and managed by the marketplace becomes independent. Finally we consider a full dismantlement scenario under which there is both vertical and horizontal separation. We show that the optimal policy is either complete dismantlement or data sharing. The relative impacts on consumer surplus and total welfare of these two options involve a tradeo§ between the increased competition implied by complete dismantling and the data related delivery cost advantage achieved under data sharing. When this cost advantage is small, completely dismantling dominates, while data sharing is the best policy when the cost advantage is large.

In vector-field-based path following (VF-PF), a vector field is designed and utilized to guide a robot to follow a desired path. VF-PF algorithms have been shown to achieve high path-following precision under small control effort compared with many other path following algorithms. However, due to the existence of singular points where the vector field vanishes, the global convergence of VF-PF algorithms to the desired path cannot be guaranteed. Moreover, as it holds for most of the existing path following algorithms, VF-PF algorithms may fail to enable following a self-intersected desired path. In this paper, we propose a method to assist generic VF-PF algorithms in their global convergence. In particular, our method creates vector fields free of singular points in a higher-dimensional space containing the lower-dimensional desired path. Subsequently, the projection of our vector fields on the lower-dimensional space of the desired path can be exploited to guarantee the global convergence of the robot to the desired path, including those self-intersected. Finally, we show that our algorithm combines and fairly extends features from existing path following algorithms and a nonlinear trajectory tracking algorithm. The theoretical results have been validated by experiments and simulations.

International audience; The growing role of automation in modern cockpits allowed a drastic diminution of aircraft acci- dents and permitted to lower crews’ workload. At the same time, it introduced new challenges that necessitate new research, particularly on error management, and vigilance (Wickens et al., 1998). High level of automation is suspected to provoke a loss of hand-flying capabilities, an increased complacency, and a reduction of the situation awareness (Endsley et al., 1995). Loss of hand-flying capabilities is observed when automation takes on the tasks previously assigned to the operator. In this situation crew skills may atrophy as they go unexercised (Haslbeck et al., 2016). Crews tend to be “automation addict”, further reducing hand flying capacities. Thus, the ability to face a sudden breakdown of automation or an unexpected situation can be alleviated (e.g., Haslbeck et al., 2012). On the other side, complacency effects (e.g., Parasuraman et al., 2010) can promote the occurrence of a particularly prominent typology of error: the failure of the crew to properly monitor the flight instruments, particularly when using highest levels of automation. The present study compared visual scanning strategies of professional pilots with and without automation. Twenty airline pilots (10 Captain and 10 First Officers forming 10 crews) performed two approaches (once as pilot flying, once as pilot monitoring) with different levels of automation (with or without flight directors and autothrottle) in an Airbus 320 full flight simulator. Flying the aircraft without any automation is considered more complex because it requires frequent gaze tran- sitions between all instruments with no external references. Pilots were required to perform the approaches in poor visibility and in accordance with operator regulations, more specifically regard- ing stabilization requirements. Eye movements of both pilots were recorded with head-mounted eye trackers. We analyzed flight performance, eye fixations, and visual patterns to examine visual scanning strategies, associated with the different automation levels. We hypothesized that during the approach without automation, the visual scanning strategies should be different than during the approach with automation, associated to a higher number of visual patterns. In addition, we hypothesized that a sub-optimal gaze allocation could be observed during approaches without automation, with the Pilot Flying possibly highly focused on primary flight instrument (specifically the attitude indicator and ILS deviation scales) to the detriment of other flight parameters. On the opposite, we assumed that pilots could highly focus on the flight directors to the detriments of other flying instruments when flying with full automation level. Our results confirmed that full automation systems were associated to a monitoring strategy fo- cused on the flight directors to the detriment of basic flight instruments, and a lower number of visual patterns. On the opposite, without automation, visual scanning was more distributed among all primary flight instruments and the number of visual pattern was higher. Results con- firmed that cockpit monitoring can be impacted by variations of the level of automation, and that an inadequate monitoring can result from complacency when automation is high. We also suggest the importance of developing better training programs to enhance pilots’ monitoring skills.

International audience; The Aircraft Landing Problem (ALP) consists in sequencing aircraft on available runways, and scheduling their landing times taking into consideration various operational constraints. It is an NP-hard problem and an ongoing challenge for both researchers and air traffic controllers. A straightforward solution widely used in practice consists in scheduling aircraft using the simple “First-Come First-Served” (FCFS) sequence. However, it rarely provides optimal solutions, especially in large congested airports.In this work, we propose a heuristic approach based on optimistic planning to solve the problem. We model the ALP as an environment of states, actions, transitions and costs, then explore the resulting search tree so as to identify a near-optimal sequence of actions within a limited time budget. In a previous contribution, we used the “First-Come First-Served” (FCFS) rule in the computation of the costs, to estimate the cost of the cheapest path (sequence of actions) from a given state. Now, we investigate a baseline model based on linear regression, and two different machine learning (ML) models trained on a large number of optimized solutions. These models can quickly and accurately estimate the cheapest-sequence cost, which helps the search to identify a near-optimal branch more efficiently. Numerical experiments are performed on our publicly available data set, and show that using machine learning models in our heuristic search does not only ameliorate the previous results in terms of percentage improvement, but also reduces the optimality gap within a computation time that is compatible with on-line operations.

International audience; This paper presents a novel air traffic structuration approach to maintain flows of air traffic and to adapt traffic situations, which can reduce the mental workload of air traffic controllers. We reformulate the optimization problem by reorganizing the aircraft trajectories in space (e.g. aircraft rerouting) or time dimension (e.g. rescheduling time of departure, flow crossing, time based metering) or both in some areas where the system identifies a high level of disorder in the traffic structure. To structure the traffic, an air traffic complexity metric based on linear dynamical systems is used for this optimization problem. To minimize the impact of traffic structure, we propose an adaptive metaheuristic approach with the integration of reinforcement learning for our resolution algorithm. The resolution algorithm is applied for short-term (flow crossing, time-based metering, and traffic encounters) trajectory planning applications and national scale planning under time uncertainty in French airspace. For short-term scenarios, our approach can restructure the traffic which allows controllers to take less effort for managing traffic situations. Our solution also improves the traffic structure with approximately 50 % reduction of air traffic complexity at national scale. Our re earch findings introduce further steps toward taking other trajectory structuration techniques into account and developing new search strategies to our resolution algorithm.

International audience; Identification and characterisation of air traffic flows is an important research topic with many applications areas including decision-making support tools, airspace design or traffic flow management. Trajectory clustering is an unsupervised data-driven approach used to automatically identify air traffic flows in trajectory data. Long-established trajectory clustering techniques applied for this purpose are mostly based on classical algorithms like DBSCAN, which depend on distance functions capturing only local relations in the data space. Recent advances in Deep Learning have shown the potential of using deep clustering techniques to discover hidden and more complex relations that often lie in low-dimensional latent spaces. The goal of this paper is to explore the application of deep trajectory clustering based on autoencoders to the problem of flow identification. Thus, we present two clustering techniques (artefact and DCEC) and show how they embed trajectories into the latent spaces in order to facilitate the clustering task.

Le développement des mini-drones dans le secteur civil est aujourd’hui un secteur en pleine expansion. Comment modéliser ces engins ? Comment concevoir les lois de commande permettant de les contrôler ?L’objectif de cet ouvrage est de présenter des méthodes d’ingénierie issues de l’automatique permettant de répondre à ces questions. Plus précisément, cet ouvrage est dédié à la modélisation, à l’identification et à la synthèse de lois de commandes spécifiques à l'avionique des mini-drones.

International audience; Analyzing the electroencephalographic (EEG) signal of epileptic patients as part of their diagnosis is a very long and tedious operation. The most common technique used by medical teams is to visualize the raw signal in order to find pathological events such as interictal epileptic spikes (IESs) or abnormal oscillations. More and more efforts are being adopted to try to facilitate the work of doctors by automating this process. Our goal was to analyze signal density fields to improve the visualization and automatic detection of pathological events. We transformed the EEG signal into images on which we applied a convolution filter based on a Kernel Density Estimation (KDE). This method that we propose to call CKDE for Convolutional Kernel Density Estimation allowed the emergence of local density fields leading to a better visualization as well as automatic detection of IESs. Future work will be necessary to make this technique more efficient, but preliminary results are very encouraging and show a high performance compared to a visual inspection of the data or some other automatic detection techniques.

International audience; The continued increase of air traffic, which doubles every 15 years, has produced economic benefits but poses environmental issues that put at risk the sustainable development of air transport. Factors such as jet fuel prices volatility, the introduction of new environmental regulations and in- tense competition in the airline industry incentive the research on ight e_ciency topics. Despite fuel costs in descent and approach phases are minor than those in cruise phase, flight crew high-workload during these phases often lead to energy mismanagement resulting in inefficient flight operations. This paper proposes an optimization algorithm that generates fuel-efficient arrival trajectories accounting for the current aircraft position. Results show 13% fuel savings compared with a best-in-class Flight Management System and, at the same time, descent time is reduced by 1%. A preliminary assessment has been conducted in the ight simulator with the aim of validating the operational feasibility of the computed trajectory. Data post-analysis suggest that the trajectory is flyable with current flight controls and guidance laws although idle margins shall be readjusted to ac- count for accelerations during descent. This paper may constitute a solid background for the likely development of flight efficiency functions for the future generation of avionic systems.

International audience; We present a new Constraint Programming (CP) model to optimize the transition cost of Fixed Job Scheduling (FJS), which improves our previous approach based on per-resource constraints by orders of magnitude. Our new model relies on a much tighter relaxation which encompasses all resources to directly propagate on the global cost, thanks to the MinWeightAllDiff optimization constraint. We also present several strategies which exploit the optimal matching computed by the MinWeightAllDiff constraint to efficiently guide the search. The resulting CP solver, using parallel cooperation between the strategies, consistently outperforms a state-of-the-art MIP solver on real instances of an FJS application, the Gate Allocation Problem, at Paris-Charles-de-Gaulle international airport.

Micro Air Vehicle (MAV)s with transitioning flight capabilities, or simply Hybrid Micro Air Vehicle(HMAV)s combine the beneficial features of fixed-wing configurations in terms of endurance,with vertical take-off and landing capabilities of rotorcraft to perform five different flight phasesduring typical missions : vertical takeoff, transitioning flight, forward flight, hovering, and verticallanding. This promising MAV class has a wider flight envelope than conventional MAVs,which implies new challenges for both control community and aerodynamic designers. One ofthe major challenges of HMAVs is the fast variation of aerodynamic forces and moments duringthe transition flight phase, which is difficult to model and control accurately. In this thesis, wefocus on the development of control laws for a specific class of HMAVs, namely tail-sitters.In order to stabilize the HMAV and overcome its modeling problem, we propose a flightcontrol architecture that estimates in realtime its fast nonlinear dynamics with an intelligentfeedback controller. The proposed flight controller is designed to stabilize the HMAV attitude,velocity and position during all flight phases. By using Model-Free Control (MFC) algorithms,the proposed flight control architecture bypasses the need for a precise HMAV model that iscostly and time consuming to obtain. A comprehensive set of flight simulations covering theentire flight envelope of the HMAV is presented, with the respective analysis for each of theflight phases. Furthermore, the control performance and the limitations of the MFC architecture are discussedin order to introduce further applications in real flight experiments. Flight tests clarifyand validate the proposed control methodology in a practical context, thus solving the principalissue of HMAVs; that is, the formulation of accurate HMAV dynamic equations to designcontrol laws. In addition, from simple mathematical algorithms, MFC is easily implemented ona microprocessor without the need for high computational costs, such as time processing andmemory resources. The results obtained provide a straightforward way in which to validatethe methodological principles presented in this thesis, to certify the designed MFC parametersand to establish a conclusion regarding MFC advantages and disadvantages in theoretical andpractical contexts related to aerospace systems., Les micro drones à capacités de vol de transition, ou simplement HMAVs (de l’anglais Hybrid MicroAir Vehicles), regroupent les principales caractéristiques aérodynamiques des configurationsà voilure fixe, en termes d’endurance, avec les capacités de décollage et d’atterrissage verticaldes voilures tournantes afin d’effectuer cinq phases de vol au cours de ses missions, telles quele décollage vertical, le vol de transition, le vol en croisière, le vol stationnaire et l’atterrissagevertical. Cette nouvelle classe de micro drones a un domaine de vol plus large que les microdrones conventionnels, ce qui implique de nouveaux défis pour les automaticiens et les concepteursaérodynamiques. L’un des principaux défis des HMAVs est la variation rapide des forces etdes moments aérodynamiques pendant la phase de vol de transition, qui est difficile à modéliseret à contrôle avec précision. Dans cette thèse, nous nous concentrons sur le développement delois de pilotage pour une classe spécifique des HMAVs, à savoir les tail-sitters.Afin de stabiliser la dynamique des tail-sitters et de surmonter leur problème de modélisation,nous proposons une architecture de contrôle de vol qui estime en temps réel leurs dynamiquesgrâce à un contrôleur à rétroaction intelligent. Le contrôleur de vol proposé est conçu pourstabiliser l’attitude du tail-sitter ainsi que sa vitesse, et sa position pendant toutes ses phasesde vol. En utilisant des algorithmes de la commande sans modèle, l’architecture de contrôle devol proposée contourne le besoin d’un modèle dynamique précis dont l’obtention est coûteuseet longue. Une série complète de simulations de vol couvrant l’ensemble du domaine de vol destail-sitters est présentée et, pour chaque phase de vol, son analyse respective.Avant d’introduire des expériences de vol réel, nous évaluons les performances et les limites del’architecture de commande sans modèle en simulation. Les essais en vol permettent de clarifier etde valider notre méthodologie de contrôle dans un contexte pratique, résolvant ainsi le principalproblème des tail-sitters, à savoir la formulation d’équations dynamiques précises pour concevoirles lois de commande. En outre, à partir d’algorithmes mathématiques simples, la commandesans modèle est facilement implémentée sur microprocesseurs sans nécessiter de coûts de calculélevés, tels que la fréquence de traitement et les ressources de mémoire. Les résultats obtenusfournissent un moyen simple de valider les principes méthodologiques présentés dans cette thèse,de certifier les paramètres obtenus lors de la conception de la commande sans modèle et d’établirune conclusion concernant ses avantages et ses inconvénients dans des contextes théoriques etpratiques liés aux systèmes aérospatiaux.

Wind information can provide an optimal estimate of the runway orientation by minimizing the crosswind component of the wind at airports, which severely affects aircraft take-off and landing performance. Additionally, a systematic geometric design requires information on wind speed, direction, duration, and specific information about latitude and longitude, temperature variation, and altitude of the airport site. In the present research, meteorological synoptic data has been precisely measured and collected over Genaveh unconstructed airport for a period of five years. Investigation of the gathered data leads to the selection of an optimal runway orientation using wind rose representation and other data analysis. Additionally, the required runway length has been estimated in order to be compatible with the standards and aircraft types considered to apply the Genaveh site. All analyses are executed for variation of temperature, altitude, landing, and take-off situations. The results demonstrate that the previously considered orientation of the runway is considerably different from the optimal direction by at least 10 degrees. Moreover, a longer runway length is required to cope with the standards to reduce the risk of accidents in the presence of crosswinds., Rüzgâr bilgisi, havalimanlarında rüzgârın uçağın kalkış ve iniş performansını ciddi şekilde etkileyen yan rüzgâr bileşenini en aza indirerek pist oryantasyonunun optimal bir tahminini sağlayabilmektedir. Ek olarak, sistematik bir geometrik tasarım ise rüzgâr hızı, yönü, süresi hakkında bilgi ve havalimanı sahasının enlem ve boylamı, sıcaklık değişimi ve rakımı hakkında özel bilgiler gerektirmektedir. Mevcut araştırmada, meteorolojik sinoptik veriler hassas bir şekilde ölçülmüş ve beş yıllık bir süre boyunca Genaveh'in yapılmamış havalimanı üzerinden ilgili veriler toplanmıştır. Toplanan verilerin temel amacı, rüzgâr gülü gösterimi ve diğer veri analizleri kullanılarak optimal bir pist oryantasyonunun seçilmesine yardımcı olmasıdır. Ek olarak, Genaveh sahasını uyguladığı düşünülen standartlar ve uçak tipleri ile uyumlu olması için gerekli pist uzunluğu tahmin edilmesidir. Tüm analizler sıcaklık, irtifa, iniş ve kalkış durumlarının değişimini temel almaktadır. Sonuçlar, pistin önceden dikkate alınan yönünün, optimum yönden en az 10 derece önemli ölçüde farklı olduğunu göstermektedir. Ayrıca, yan rüzgarların varlığında kaza riskini azaltmak için standartlarla başa çıkmak için daha uzun bir pist uzunluğu gereklidir.

International audience; This paper proposes a methodology to address the problem of international multi modal hubs crisis management. The analysis starts from the case study of the Brussel airport terrorist attacks to highlight multi-modal and international aspects of this kind of crisis. Then, focusing on railway and airway modes, the contextual and organizational diversity is explained. The need of a global crisis management system is justified, but the challenges ensuing from the different transportation modes are highlighted. The proposition of a modelling approach based on System of Systems formalism is studied. The possibility of building a correct by construction design of common functioning mode is justified using the scientific state of the art of formal methods providing a scientific framework for future technological and organizational innovations assessment.

International audience; Proceedings of the Interdisciplinary Workshop on Human-Drone Interactionco-located with the 2020 ACM CHI Conference on Human Factors in Computing Systems (CHI 2020)

Security threats have emerged in the past decades as a more and more critical issue for Air Transportation which has been one of the main ressource for globalization of economy. Reinforced control measures based on pluridisciplinary research and new technologies have been implemented at airports as a reaction to different terrorist attacks. From the scientific perspective, the efficient screening of passengers at airports remain a challenge and the main objective of this thesis is to open new lines of research in this field by developing advanced approaches using the resources of Computer Science. First this thesis introduces the main concepts and definitions of airport security and gives an overview of the passenger terminal control systems and more specifically the screening inspection positions are identified and described. A logical model of the departure control system for passengers at an airport is proposed. This model is transcribed into a graphical view (Controlled Satisfiability Graph-CSG) which allows to test the screening system with different attack scenarios. Then a probabilistic approach for the evaluation of the control system of passenger flows at departure is developped leading to the introduction of Bayesian Colored Petri nets (BCPN). Finally an optimization approach is adopted to organize the flow of passengers at departure as best as possible given the probabilistic performance of the elements composing the control system. After the establishment of a global evaluation model based on an undifferentiated serial processing of passengers, is analyzed a two-stage control structure which highlights the interest of pre-filtering and organizing the passengers into separate groups. The conclusion of this study points out for the continuation of this theme., Les menaces à la sécurité sont apparues au cours des dernières décennies comme un problème de plus en plus critique pour le transport aérien, qui est l'un des principaux levier de la mondialisation de l' Économie. Des mesures de contrôle renforcées reposant sur des recherches pluridisciplinaires et sur de nouvelles technologies ont été mises en œuvre dans les aéroports en réaction à différentes attaques terroristes. Du point de vue scientifique, le filtrage efficace des passagers dans les aéroports reste un défi et l’objectif principal de cette thèse est d’ouvrir de nouvelles pistes de recherche dans ce domaine en développant des approches avancées utilisant les ressources de la science informatique. Tout d’abord, cette thèse présente les principaux concepts et définitions de la sécurité dans les aéroports et donne un aperçu des systèmes de contrôle des terminaux de passagers, et plus précisément des postes d’inspection - filtrage qui sont identifiés et décrits. Un modèle logique du système de contrôle des départs des passagers d’un aéroport est proposé. Ce modèle est transcrit en graphique ment (Controlled Satisfiability Graph - CSG) , ce qui permet de tester le système de filtrage sous différents scénarios d’attaque. Ensuite, une approche probabiliste pour l’évaluation du système de contrôle des flux de passagers au départ est développée, conduisant à l’introduction des réseaux de Petri colorés Bayésiens (BCPN). Enfin, une approche d'optimisation est adoptée pour organiser au mieux les flux de passagers au contrôle de départ compte tenu de la performance probabiliste des éléments composant le système de contrôle. Après la mise en place d'un modèle d'évaluation global basé sur un traitement en série indifférencié des passagers, une structure de contrôle en deux étapes est analysée, qui met en évidence l'intérêt du pré - filtrage et de l'organisation des passagers en groupes distincts. La conclusion de cette étude indique la poursuite de ce thème

International audience; Closest Point of Approach (CPA) is one of the main problems in aircraft Conflict Detection (CD). It aims to find out the minimum distance and the associated time between two aircraft. Conventional CPA calculation algorithm generally assumes that the speed and heading of aircraft are constant. But the uncertainties in real operational scenarios lead to inaccuracy of CPA calculation. This project presents a novel data-driven CD framework with Machine Learning (ML) algorithms. The proposed framework provides a promising solution for improving the CPA prediction accuracy with the help of real trajectory data. It contributes to not only reduce the number of fault Short-mid term conflict alert for air traffic controllers, but also support the implementation of future free flight concept, so as to reduce fuel consumption and emission.

International audience; Light Detection and Ranging (LIDAR) sensors are nowadays widely used for multiple applications such as autonomous driving, wind estimation and robots, however their use in airports has not yet been fully explored. Nevertheless, LIDAR sensors are decreasing in cost, and there are detection algorithms that could be useful in small airports. We present a system to visualize and detect aircraft in small airfields. The system enables an interactive calibration of the algorithm, and exploration of the results. The system has proven useful for the detection and positioning of airplanes in the ENVISION Sesar project.

International audience; In aviation, safety has always been a key issue to reduce the number of incidents and accidents. Nowadays, it is even more important since the air traffic increases every year, and is predicted to reach 7.8 billion passengers worldwide in 2036. Flight safety offices aim at enhancing safety, analyzing past events, and preventing potential critical occurrences. This paper presents TAPAS, a post-operational aircraft approach analysis tool. The software offers an interactive analysis of aircraft approach energy management. The software uses and validates an existing atypical approach detection algorithm on flight data record and flight safety office events from airlines. Various correlations and analysis are conducted to illustrate the potential benefits of this methodology.

In an effort to simplify human resource management and reduce operational costs, control towers are now increasingly designed to not be implanted directly on the airport but remotely. This concept, known as remote tower, offers a “digital”working context: the view on the runways is broadcast remotely using cameras located on site. Furthermore, this concept could be enhanced to the control of several airports simultaneously from one remote tower facility, by only one air traffic controller (multiple remote tower). These concepts offer designers the possibility to develop novel interaction forms. However, the most part of the current augmentations rely on sight, which is largely used and, therefore, is sometimes becoming overloaded. In this Ph.D. work, the design and the evaluation of new interaction techniques that rely onnon-visual human senses have been considered (e.g. hearing, touch and proprioception). Two experimental campaigns have been led to address specific use cases. These use cases have been identified during the design process by involving experts from the field, appearing relevant to controllers due to the criticality of the situation they define. These situations are a) poor visibility (heavy fog conditions, loss of video signal in remote context), b) unauthorized movements on ground (when pilots move their aircraft without having been previously cleared), c) runway incursion (which occurs when an aircraft crosses the holding point to enter the runway while another one is about to land), and d) how to deal with multiple calls associated to distinct radio frequencies coming from multiple airports. The first experimental campaign aimed at quantifying the contribution of a multimodal interaction technique based on spatial sound, kinaesthetic interaction and vibrotactile feedback to address the first use case of poor visibility conditions. The purpose was to enhance controllers’ perception and increase overall level of safety, by providing them a novel way to locate aircraft when they are deprived of their sight. 22 controllers have been involved in a laboratory task within a simulated environment.Objective and subjective results showed significantly higher performance in poor visibility using interactives patial sound coupled with vibrotactile feedback, which gave the participants notably higher accuracy in degraded visibility.Meanwhile, response times were significantly longer while remaining acceptably short considering the temporal aspect of the task. The goal of the second experimental campaign was to evaluate 3 other interaction modalities and feedback addressing 3 other critical situations, namely unauthorized movements on ground, runway incursion and calls from a secondary airport. We considered interactive spatial sound, tactile stimulation and body movements to design3 different interaction techniques and feedback. 16 controllers’ participated in an ecological experiment in which they were asked to control 1 or 2 airport(s) (Single Vs. Multiple operations), with augmentations activated or not. Having no neat results regarding the interaction modalities into multiple remote tower operations, behavioural results shown asignificant increase in overall participants’ performance when augmentation modalities were activated in single remotecontrol tower operations. The first campaign was the initial step in the development of a novel interaction technique that uses sound as a precise means of location. These two campaigns constituted the first steps for considering non-visual multimodal augmentations into remote tower operations., Afin de simplifier la gestion des ressources humaines et de réduire les coûts d’exploitation, certaines tours de contrôle sont désormais conçues pour ne pas être implantées directement sur l’aéroport. Ce concept, connu sous le nom de tour de contrôle distante (remote tower), offre un contexte de travail “digital” : la vue sur les pistes est diffusée via des caméras situées sur le terrain distant. Ce concept pourrait également être étendu au contrôle simultanés de plusieurs aéroports à partir d’une seule salle de contrôle, par un contrôleur seul (tour de contrôle distante multiple). Ces notions nouvelles offrent aux concepteurs la possibilité de développer des formes d’interaction novatrices. Cependant, la plupart des augmentations actuelles reposent sur la vue, qui est largement utilisée et, par conséquent, parfois surchargée.Nous nous sommes ainsi concentrés sur la conception et l’évaluation de nouvelles techniques d’interaction faisant appel aux sens non visuels, plus particulièrement l’ouïe, le toucher et la proprioception. Deux campagnes expérimentales ont été menées. Durant les processus de conception, nous avons identifié, avec l’aide d’experts du domaine, certaines situations pertinentes pour les contrôleurs aériens en raison de leur criticité: a) la mauvaise visibilité (brouillard épais,perte de signal vidéo), b) les mouvements non autorisés au sol (lorsque les pilotes déplacent leur appareil sans y avoir été préalablement autorisés), c) l’incursion de piste (lorsqu’un avion traverse le point d’attente afin d’entrer sur la piste alors qu’un autre, simultanément, s’apprête à atterrir) et d) le cas des communications radio simultanées provenant de plusieurs aéroports distants. La première campagne expérimentale visait à quantifier la contribution d’une technique d’interaction basée sur le son spatial, l’interaction kinesthésique et des stimuli vibrotactiles, afin de proposer une solution au cas de perte de visibilité sur le terrain contrôlé. L’objectif était d’améliorer la perception de contrôleurs et d’accroître le niveau général de sécurité, en leur offrant un moyen différent pour localiser les appareils. 22 contrôleurs ont été impliqués dans une tâche de laboratoire en environnement simulé. Des résultats objectifs et subjectifs ont montré une précision significativement plus élevée en cas de visibilité dégradée lorsque la modalité d’interaction testée était activée. Parallèlement, les temps de réponse étaient significativement plus longs relativement courts par rapport à la temporalité de la tâche. L’objectif de la seconde campagne expérimentale, quant à elle, était d’évaluer 3 autres modalités d’interaction visant à proposer des solutions à 3 autres situations critiques : les mouvements non autorisés au sol,les incursions de piste et les appels provenant d’un aéroport secondaire contrôlé. Le son spatial interactif, la stimulation tactile et les mouvements du corps ont été pris en compte pour la conception de 3 autres techniques interactives. 16contrôleurs aériens ont participé à une expérience écologique dans laquelle ils ont contrôlé 1 ou 2 aéroport(s), avec ou sans augmentation. Les résultats comportementaux ont montré une augmentation significative de la performance globale des participants lorsque les modalités d’augmentation étaient activées pour un seul aéroport. La première campagne a été la première étape dans le développement d’une nouvelle technique d’interaction qui utilise le son interactif comme moyen de localisation lorsque la vue seule ne suffit pas. Ces deux campagnes ont constitué les premières étapes de la prise en compte des augmentations multimodales non visuelles dans les contextes des tours de contrôles déportées simples et Multiples.

Values in a graphical interface such as a spreadsheet can be manipulated indirectly by means of a feature whereby representative values, such as unique values are identified and presented separately. Manipulations of these representative values are then passed on to the corresponding original values., Des valeurs dans une interface graphique, telle qu'un tableur, peuvent être manipulées indirectement au moyen d'une caractéristique selon laquelle des valeurs représentatives, telles que des valeurs uniques, sont identifiées et présentées séparément. Les manipulations de ces valeurs représentatives sont ensuite transmises aux valeurs originales correspondantes.

Traffic status feedback is provided by a driveradapted to interface with a human perceptible vibrationtransducer either as a haptic or audio signal, by causingthe transducer to emit a vibration having one or morecharacteristics defined in proportion to traffic data. Thetraffic data may relate to a particular space, and be mod-ulated as a function of their position in that space. Bycompiling a signal concerning multiple traffic entities orother traffic data, a user can be provided with a concen-trated representation of complex data, which the user willcome instinctively to interpret

This thesis focuses on Machine Learning and information extraction for aeronautical loads and stress data. In the first time, we carry out a study for the prediction of aeronautical loads curves. We compare regression trees based models and quantify the influence of dimension reduction techniques on regression performances in an extrapolation context. In the second time, we develop a deformation model acting simultaneously on the inputand the output space of the curves. We study the asymptotic properties of the estimators of the deformation parameters. This deformation model is associated to the modeling and predicting process of aeronautical loads.Finally, we give a simple and efficient method for predicting critical loads., Cette thèse s'intéresse à l'apprentissage et à la représentation de données de charges et contraintes aéronautiques. Nous réalisons dans un premier temps une étude préliminaire pour la prédiction des courbes de charges aéronautiques. Nous comparons des méthodes de régression à base d'arbres et quantifions l'influence de techniques de réduction de dimension sur les performances en régression dans un cadre d'extrapolation. Dans un second temps, nous développons un modèle de déformation agissant simultanément sur les entrées et sorties des courbes. Nous étudions les propriétés asymptotiques des estimateurs des paramètres de déformation. Ce modèle de déformation est associé au processus de modélisation et prédiction des charges aéronautiques. Dans un troisième temps, nous donnons une méthodes simple et efficace de prédiction de charges critiques.

International audience; This article presents the design of a mini UAV dedicated to atmospheric research with tight operational constraints coming from the end-users, which are the meteorologists associated to the project. Several aspects are covered in addition to the conceptual design of the frame itself and its manufacturing process. This includes the innovative launching system based on water rocket, the design of a 5-hole probe for wind and turbulence measurements, the new version of the on-board autopilot and finally the evaluation of a long range communication system. Preliminary results are presented to conclude the paper.

Aircraft upset conditions have, in the past, led to rare but fatal accidents and to this day remain a severe threat to civil aviation. In response, considerable efforts contributed towards the prevention and detection of upset conditions as well as recovery of the aircraft into its desired flight regime. Despite these efforts made, comprehensive strategies including analysis of aircraft dynamics, recovery control, and verification of the implemented control over a large flight envelope have been missing for future aircraft safety. In this thesis, we apply and extend recently developed tools such as bifurcation theory, sum of squares analysis and control synthesis, and nonlinear model-predictive control, to a set of newly developed piecewise polynomial aircraft models, which provide a trade-off between modeling accuracy and computational complexity. Thus, we contribute to the development as well as certification of safer flight control systems., Dans le passé, le décrochage d’aéronefs a entraîné des accidents rares mais mortels et reste à ce jour une préoccupation importante pour l’aviation civile.Malgré les efforts déployés pour la prévention et la détection du décrochage ainsi que la récupération de l'avion, des stratégies telles que l'analyse de la dynamique, le contrôle de la récupération au décrochage et la vérification des lois de commande mises en oeuvre constituent un chaînon manquant pour la sécurité des avions à venir.Dans cette thèse, nous appliquons et étendons des outils récents de l'automatique, tels que la théorie de la bifurcation, les programmes de « sum of squares » et le contrôle prédictif modèle non-linéaire aux modèles d’avions polynomiaux par morceaux développés, qui offrent une bonne précision de modélisation et une faible capacité de calcul. Ainsi, nous contribuons au développement et à la certification de systèmes de contrôle de vol plus sûrs.

International audience; The objective of this paper is to determine the ranging performance of the upcoming fifth generation (5G) signal. In order to do so, it is required to define 5G correlator outputs mathematical models. 5G systems will use OFDM (Orthogonal Frequency Division Multiplexing) signals; in the literature, mathematical models of OFDM signals are developed at the different receiver signal processing stages. These models assumed that the propagation channel is constant over an OFDM symbol; nevertheless, an in-depth study of QuaDRiGa, a 5G compliant propagation channel simulator, invalidates this hypothesis. Therefore, in this paper, mathematical models are developed that take into account the channel evolution. The focus is given on correlator outputs and results are applied to the computation of 5G based pseudo range accuracy.

International audience

International audience; For the sake of Air Traffic Management modernization , civil aviation organizations are currently developing IPS for Aeronautical Safety Services in the new ATN/IPS infrastructure. This includes to define new airborne and ground-based communication systems capable of managing both air traffic services (ATS) and aeronautical operational communications (AOC) safety services. One of the main challenges in this new ATN/IPS network is the IPv6 mobility problem. This paper proposes a solution which takes both advantages of ground based Locator/Identifier Separation Protocol and Proxy Mobile IPv6 to manage all the aircraft mobility scenarios. A dedicated OMNeT++ simulation model is also provided and shows the performances of our solution.

A mechanism for authenticating a website is provided whereby an additional authentication value which is inherently and verifiably associated with the entity operating the website such as address, business registration number, owner name or the like. The additional authentication value is included in a signed field of a certificate provided to visitors of the website. The browser interfacing with the website can then compare the additional authentication value with the same information obtained from another source, such as a dedicated authentication server, or user input. If the information is found to match, the information can be deemed to be authenticated., L'invention concerne un mécanisme permettant d'authentifier un site Web, une valeur d'authentification supplémentaire associée de manière inhérente et vérifiable à l'entité, telle qu'une adresse, un numéro d'enregistrement d'entreprise, un nom de propriétaire ou similaire exploitant le site Web. La valeur d'authentification supplémentaire est incluse dans un champ signé d'un certificat fourni aux visiteurs du site Web. Le navigateur en interface avec le site Web peut ensuite comparer la valeur d'authentification supplémentaire aux mêmes informations obtenues à partir d'une autre source, telle qu'un serveur d'authentification dédié, ou une entrée d'utilisateur. Si les informations s'avèrent correspondre, les informations peuvent être considérées comme authentifiées.

International audience; Departure metering has the potential to mitigate airport surface congestion and decrease flight delays. This paper considers several candidate departure metering techniques, including a trajectory-based optimization approach using a node-link model and three aggregate queue-based approaches (a scheduler based on NASA's ATD-2 logic, an optimal control approach, and a robust control approach). The outcomes of these different approaches are compared for two major airports: Paris Charles De Gaulle airport (CDG) in Europe and Char-lotte Douglas International airport (CLT) in the United States. Stochastic simulations are used to show that the robust control approach best accommodates operational uncertainties, while all the approaches considered yield higher taxi-out time savings at CLT compared to CDG.

With the continuous growth in the air transportation demand, air traffic controllers will have to handle increased traffic and consequently more potential conflicts. This gives rise to the need for conflict resolution advisory tools that can perform well in high-density traffic scenarios given a noisy environment. Unlike model-based approaches, learning-based or machine learning approaches can take advantage of historical traffic data and flexibly encapsulate the environmental uncertainty. In this study, we propose an artificial intelligent agent that is capable of resolving conflicts, in the presence of traffic and given uncertainties in conflict resolution maneuvers, without the need for prior knowledge about a set of rules mapping from conflict scenarios to expected actions. The conflict resolution task is formulated as a decision-making problem in a large and complex action space, which is applicable for employing the reinforcement learning algorithm. Our work includes the development of a learning environment, scenario state representation, reward function, and learning algorithm. As a result, the proposed method, inspired by Deep Q-learning and Deep Deterministic Policy Gradient algorithms, can resolve conflicts, with a success rate of over 81%, in the presence of traffic and varying degrees of uncertainties. Civil Aviation Authority of Singapore (CAAS) This research is partially supported by Air Traffic Management Research Institute (NTU-CAAS) Grant No. M4062429.052

International audience

International audience

International audience

Let G be a graph, for which each vertex is assigned one of k colors in {1, …, k}. A legal coloring requires that two adjacent vertices have two different colors. The minimum sum coloring problem (MSCP) consists of finding such a legal coloring with the smallest sum value, knowing that to each color is assigned an integer value from 1 to k. This paper presents a new memetic approach for this NP-hard problem. The proposed hybrid algorithm is based on a very small population, composed of only two individuals. Thanks to this small population, no selection operator needs to be defined, nor any replacement strategy. In order to prevent a premature convergence of the algorithm, alternative mechanisms are introduced based on an elitist approach. The local search introduced in the population based algorithm is split in two phases. The first one is based on the well known TabuCol algorithm and aims to reduce the number of conflicting vertices. The second one, which is used to improve the total sum value, implements an efficient 2-move local operator. The proposed approach was successfully applied on many graphs from the reference COLORS02 and DIMACS benchmarks. It allowed to achieve most of the best known results, and to overpass the best known results for 15 challenging graphs.

International audience; This paper aims at presenting a novel way of predicting and analyzing air traffic delays using publicly available data from social media with a focus on Twitter data. Three different machine learning regressors have been trained on this2017 passenger-centric dataset and tested for the prediction up to five hours ahead of air traffic delays and cancellations for the first two months of 2018. Comparing and analyzing different accuracy measures of their prediction performances show that this dataset contains useful information about the current state and short-term future state of the air traffic system. The resulting methods yield higher prediction accuracy than traditional state-of-the-art and off-the-shelf time-series forecasting techniques performed on flight-centric data. More over a post-training feature importance analysis conducted on the Random Forest regressor allowed a simplification and a refining of the model, leading to a faster training time and more accurate predictions. This paper is a first step in predicting and analyzing air traffic delays leveraging a real-time publicly available passenger-centered data source. The results of this study suggest a method to use passenger-centric data-sources both as an estimator of the current state of air traffic delays as well as an estimator of the short-term state of air traffic delays in the United States in real-time.

International audience; In this study, we propose an aircraft 4D trajectory optimization model based on Bezier curve. Many real-world factors (such as winds, obstacles, uncertainties) and actions (the modification of departure time, the trajectory shape, aircraft speed and altitude) are taken into account. To solve the model, an improved simulated annealing algorithm with two phases was proposed: the first phase for reducing the number of conflicts and the second phase for decreasing the total flight time. A national-size dataset for France which is provided by a fast time simulator (ΠRATS) is used as a case study. The experimental results show that the algorithm provides conflict-free trajectories within a very short time for all instances. For the objective to deconflict aircraft, the algorithm is almost linearly scalable for large-scale instances. For a given limited run time (such as 6 hours), the algorithm provides good solutions with small values of objective function (total flight time, changes of aircraft speed and obstacles encounters).

National audience; Dans de nombreuses applications, comme les statistiques directionnelles, l’imagerie de tenseurs de diffu-sion (DTI), le traitement du signal radar ou encore l’analyse de formes, les objets d’ ́etude ne sont pasfacilement repr ́esentables par des ́el ́ements d’un espace vectoriel. La g ́eom ́etrie riemannienne est alorssouvent utilis ́ee pour mod ́eliser cette absence de lin ́earit ́e. L’ ́etude de ce type de donn ́ees r ́eelles n ́ecessitedes outils statistiques et de traitement de donn ́ees bien souvent introduits dans un cadre vectoriel. Lag ́en ́eralisation aux vari ́et ́es riemanniennes de concepts statistiques usuels [7] et de m ́ethodes standardd’analyse de donn ́ees comme l’analyse en composantes principales [10] ou la quantification [5], est parcons ́equent un d ́efi important. L’absence d’unicit ́e de g ́en ́eralisation de lois usuelles comme la loi gaussi-enne, selon que l’on veuille privilgier le respect de certaines invariances [3, 1] ou la concidence entremaximum de vraisemblance et centre de masse riemannien [9], rend cette problmatique non triviale.Les lois de probabilit ́es elles-mˆemes peuvent tre analyses via la gomtrie riemannienne. Une familleparamtre de lois peut ˆetre munie d’une structure de vari ́et ́e riemannienne grce `a la m ́etrique de Fisher, quid ́efinit ainsi une g ́eom ́etrie dite de l’information. Les cas gaussiens unidimensionnel, multidimensionnelcentr ́e, et multidimensionnel centr ́e stationnaire ont t trs tudis, notamment en DTI [8] et en traitementdu signal radar [2], et font merger une gomtrie courbure ngative. Dans le cas non stationnaire, il fautchanger l’espace de repr ́esentation et se tourner vers le groupe de Lie sp ́ecial orthogonal [4]. D’autresapplications comme l’imagerie m ́edicale font ́emerger d’autres lois comme la loi gamma g ́en ́eralis ́ee, pourla g ́eom ́etrie de laquelle peu de r ́esultats sont connus. Dans cette session, nous pr ́esenterons diff ́erentespistes pour aborder ces questions, pour des applications allant du traitement du signal au contrˆole dutrafic a ́erien, en passant par l’imagerie m ́edicale.

National audience; Un enjeu majeur en contrôle du trafic aérien est la capacitéà estimer la complexité dune situation de trafic donnée. Cette complexité doitêtre inféréeà partir d'une image indiquant les positions et vitesses des avions présents dans l'espace aérien au temps considéré. Pour l'évaluer localement, on peut extraire en chaque point dobservation une matrice de covariance révélatrice du "désordre local" du champ des vitesses des avions dans un petit voisinage. On se retrouve alors avec un ensemble de matrices de covariance, c'est-à-dire un nuage de points dans l'espace des matrices symétriques définies positives, qui peutêtre vu comme une distribution de probabilité. Suivant l'approche présentée dans [2], nous verrons comment construire un résumé dune situation de trafic aérien en quantifiant la distribution empirique des matrices de covariance correspondante. Pour cela, nous munirons l'espace des matrices symétriques définies positives de la métrique de Fisher associée au vecteurs gaussiens centrés,également appelée métrique affine invariante [3], et nous nous intéresseronsà lapproximation discrète de lois de probabilité, en adaptant les outils de la quantification vectorielle [1] au cadre riemannien. Nous verrons que cette approche fournit un clustering de l'espace aérien en zones de complexité homogène, ainsi qu'un résumé de la situation de trafic qui peut ensuiteêtre comparéà d'autres par transport optimal discret.

Nowadays, stand-alone Global Navigation Satellite System (GNSS) positioning accuracy is not sufficient for a growing number of land users. Sub-meter or even centimeter accuracy is becoming more and more crucial in many applications. Especially for navigating rovers in the urban environment, final positioning accuracy can be worse as the dramatically lack and contaminations of GNSS measurements.To achieve a more accurate positioning, the GNSS carrier phase measurements appear mandatory. These measurements have a tracking error more precise by a factor of a hundred than the usual code pseudorange measurements. However, they are also less robust and include a so-called integer ambiguity that prevents them to be used directly for positioning.While carrier phase measurements are widely used in applications located in open environments, this thesis focuses on trying to use them in a much more challenging urban environment. To do so, Real-Time-Kinematic (RTK) methodology is used, which is taking advantage on the spatially correlated property of most code and carrier phase measurements errors. Besides, the thesis also tries to take advantage of a dual GNSS constellation, GPS and GLONASS, to strengthen the position solution and the reliable use of carrier phase measurements. Finally, to make up the disadvantages of GNSS in urban areas, a low-cost MEMS is also integrated to the final solution.Regarding the use of carrier phase measurements, a modified version of Partial Integer Ambiguity Resolution (Partial-IAR) is proposed to convert as reliably as possible carrier phase measurements into absolute pseudoranges. Moreover, carrier phase Cycle Slip (CS) being quite frequent in urban areas, thus creating discontinuities of the measured carrier phases, a new detection and repair mechanism of CSs is proposed to continuously benefit from the high precision of carrier phases.Finally, tests based on real data collected around Toulouse are used to test the performance of the whole methodology., Récemment, la précision qu’on peut obtenir avec le GNSS-autonome positionnement est insatisfaisant pour de plus en plus d’utilisateurs au terrain. Une précision au-dessous de mètre ou encore de niveau centimètre est devenue plus cruciale pour pleine d’applications. Surtout pour des véhicules dans le milieu urbain, la précision finale est considérablement pire comme les manques et les contaminations des GNSS signaux sont présentes. Afin d’avoir un positionnement plus précis, les mesures carrier phases sont dispensables. Les erreurs associés à ces mesures sont plus précises de factor centaine que les mesures pseudoranges. Cependant, un paramètre de nature entière, appelée ambiguïté, empêche les mesures phases de comporter comme les pseudoranges ‘précises’. Pendant que les mesures carrier phases sont largement utilisées par les applications localisées dans un environnement ouvert, cette thèse s’intéresse sur les exploitations dans un environnement urbain. Pour cet objectif, la méthodologie RTK est appliquée, qui est principalement basée sur les caractéristiques que les erreurs sur les mesures pseudoranges et phases sont corrélées spatialement. De plus, cette thèse profite de la double GNSS constellation, GPS et GLONASS, pour renforcer la solution de position et l’utilisation fiable des mesures carrier phases. Enfin, un low-cost MEMS est aussi intégré pour compenser des désavantages de GNSS dans un milieu urbain. A propos des mesures phases, une version modifiée de Partial-IAR (Partial Ambiguity Resolution) est proposée afin de faire comporter de la façon la plus fiable possible les mesures phases comme les pseudoranges. Par ailleurs, les glissements de cycle sont plus fréquents dans un milieu urbain, qui introduisent des discontinuités des mesures phases. Un nouveau mécanisme pour détecter et corriger les glissements de cycle est du coup mise en place, pour bénéficier de la haute précision des mesures phases. Des tests basés sur les données collectées autour de Toulouse sont faits pour montrer la performance.

International audience