Archive ouverte de l'ENAC
Not a member yet
3458 research outputs found
Sort by
Voronoi diagrams and Simulated Annealing for airspace block optimization
No full textInternational audienceThis paper investigates the design of airspace blocks using Voronoi diagrams and simulated annealing. This approach aims to optimize the layout of airspace blocks by minimizing the complexity gap between them. The algorithm is tested with different complexity metrics. By using Voronoi diagrams, which partition the airspace into regions around specified points, and simulated annealing, which iteratively refines solutions to find near-optimal configurations, the algorithm provides a systematic method for airspace design. The study focuses specifically on the French airspace, providing a real-world application of the proposed methodology. Through experimentation and evaluation, the algorithm demonstrates its ability to generate airspace block configurations that balance complexity. This research contributes to ongoing efforts in airspace management and optimization by providing insights and techniques for designing airspace structures that meet the evolving needs of air traffic control systems.</div
On the Detection of Aircraft Single Engine Taxi using Deep Learning Models
No full textInternational audienceThe aviation industry is vital for global transportation but faces increasing pressure to reduce its environmentalfootprint, particularly CO2 emissions from ground operations such as taxiing. Single Engine Taxiing (SET) has emerged as apromising technique to enhance fuel efficiency and sustainability. However, evaluating SET’s benefits is hindered by the limitedavailability of SET-specific data, typically accessible only to aircraft operators. In this paper, we present a novel deep learningapproach to detect SET operations using ground trajectory data. Our method involves using proprietary Quick Access Recorder(QAR) data of A320 flights to label ground movements as SET or conventional taxiing during taxi-in operations, while using onlytrajectory features equivalent to those available in open-source surveillance systems such as Automatic Dependent Surveillance-Broadcast (ADS-B) or ground radar. This demonstrates that SET can be inferred from ground movement patterns, paving theway for future work with non-proprietary data sources. Our results highlight the potential of deep learning to improve SETdetection an
Polynomial Regression on Lie Groups and Application to SE(3)
No full textInternational audienceIn this paper, we address the problem of estimating the position of a mobile such as a drone from noisy position measurements using the framework of Lie groups. To model the motion of a rigid body, the relevant Lie group happens to be the Special Euclidean group SE(n), with n=2 or 3. Our work was carried out using a previously used parametric framework which derived equations for geodesic regression and polynomial regression on Riemannian manifolds. Based on this approach, our goal was to implement this technique in the Lie group SE(3) context. Given a set of noisy points in SE(3) representing measurements on the trajectory of a mobile, one wants to find the geodesic that best fits those points in a Riemannian least squares sense. Finally, applications to simulated data are proposed to illustrate this work. The limitations of such a method and future perspectives are discussed
Toward a consolidation of the European Airline Sector: the potential merger between ITA and LuftHansa
No full textInternational audienceThe purpose of this study is to assess the impact of Lufthansa's bid to acquire the Italian airline ITA Airways. On the basis of different scenarios, we aim to estimate the impact on the supply of air products and on consumers. To simulate the impact of such a merger on the European market, we rely on standard structural models of demand and supply used in the empirical IO literature. Since Berry (1994), many papers have used them to estimate demand in various sectors, including the airline sector, mostly at the US level. In particular, Berry et al. (2006) use a random coefficients model to study the role of hubs, while Berry and Jia (2010) compare the years 1999 and 2006. We want to compare the variation in consumer surplus and equilibrium fares under different scenarios. In particular, we need to model different possibilities for the range of products offered by the new merged entity. We also want to compare these scenarios with others in which ITA Airways could have merged with another airline, such as Air France. In our paper, in our attempt to simulate the impact of the merger on consumer surplus and fares, we face an additional challenge due to the possibility of repositioning the products offered by the competitors of ITA Airlines and LuftHansa. A particular feature of the European market is the presence of many low-cost carriers. These airlines are more likely to react to a reduction in the number of competitors and we need to model their strategies too
Adversarial attacks on neural networks through canonical Riemannian foliations
No full textInternational audienceDeep learning models are known to be vulnerable to adversarial attacks. Adversarial learning is therefore becoming a crucial task. We propose a new vision on neural network robustness using Riemannian geometry and foliation theory. The idea is illustrated by creating a new adversarial attack that takes into account the curvature of the data space. This new adversarial attack, called the two-step spectral attack, is a piece-wise linear approximation of a geodesic in the data space. The data space is treated as a (pseudo) Riemannian manifold equipped with the pullback of the Fisher Information Metric (FIM) of the neural network. In most cases, this metric is only semi-definite and its kernel becomes a central object to study. A canonical foliation is derived from this kernel. The curvature of transverse leaves gives the appropriate correction to get a two-step approximation of the geodesic and hence a new efficient adversarial attack. The method is first illustrated on a 2D toy example in order to visualize the neural network foliation and the corresponding attacks. Next, we report numerical results on the MNIST and CIFAR10 datasets with the proposed technique and state of the art attacks presented in [1] (OSSA) and [2] (AutoAttack). The results show that the proposed attack is more efficient at all levels of available budget for the attack (norm of the attack), confirming that the curvature of the transverse neural network FIM foliation plays an important role in the robustness of neural networks. The main objective and interest of this study is to provide a mathematical understanding of the geometrical issues at play in the data space when constructing efficient attacks on neural networks
Urban Air Mobility Guidance with Panel Method: Experimental Evaluation Under Wind Disturbances
No full textInternational audienceIn this paper, a nature-inspired guidance algorithm based on the panel method is proposed. The panel method is a numerical tool borrowed from the aerodynamics domain to calculate the potential field of a fluid flow around arbitrarily shaped objects. The proposed algorithm has little computational load and generates guidance vectors in real time that can guide multiple vehicles through smooth and collision-free paths. Panel-method-based guidance is a promising candidate for air mobility applications in urban environments where multiple aerial vehicles are expected to operate simultaneously without colliding with architectural structures and other vehicles in the airspace. In this study, the effectiveness and feasibility of the proposed guidance method is evaluated through a test campaign conducted in Toulouse, France, using multiple quadrotors in a scaled urban environment. Furthermore, the robustness of the guidance method under wind disturbances is tested in both indoor and outdoor experiments. Experimental results suggest that the panel-method-based guidance algorithm is an effective and robust tool for real-time, collision-free guidance of multiple aerial vehicles in complex urban environments
Robust pre-departure scheduling for a nation-wide air traffic flow management
No full textInternational audienceAir traffic flow management has been a major means for balancing air traffic demand and airport or airspace capacity to reduce congestion and flight delays. However, unpredictable factors, such as weather and equipment malfunctions, can cause dynamic changes in airport and sector capacity, resulting in significant alterations to optimized flight schedules and the calculated predeparture slots. Therefore, taking into account capacity uncertainties is essential to create a more resilient flight schedule. This paper addresses the flight pre-departure sequencing issue and introduces a capacity uncertainty model for optimizing flight schedules at the airport network level. The goal of the model is to reduce the total cost of flight delays while increasing the robustness of the optimized schedule. A chance-constrained model is developed to address the capacity uncertainty of airports and sectors, and the significance of airports and sectors in the airport network is considered when setting the violation probability. The performance of the model is evaluated using real flight data by comparing them with the results of the deterministic model. The development of the model based on the characteristics of this special optimization mechanism can significantly enhance its performance in addressing the pre-departure flight scheduling problem at the airport network level
Foreword
No full textThe 18th edition of the annual conference of the EUROPT working group of EURO, the EUROPT Workshop on Advances in Continuous Optimization –EUROPT 2021– has been held virtually from July 7 to 9, 2021, organized from Toulouse, France, and hosted at ENAC-–École Nationale de l'Aviation Civile: https://europt2021.recherche.enac.fr
Une critique de l’hylémorphisme dans les théories du design d’interaction : recherche de concepts dans la philosophie pour appréhender la forme et la matérialité
No full textInternational audienceThe search for an understanding of the emergence of design in terms of form and matter is a long-standing one. A rise of form as a conceptual tool has been observed in the scientific literature on interaction design, in parallel with a shift towards materiality, as a way of expressing the forces at work in design. In this article, we highlight how current theories seem to us to be implicitly built on the hylomorphic paradigm, in an inconsistent way. To give a more detailed account of the mechanisms of elaboration and conception in interaction design, we propose to use the explicit concepts of transduction, metastability, individuation, and concretization, taken from Gilbert Simondon’s critical approach to the hylomorphic schema. To the hylomorphic paradigm of Aristotelian shaping, we will oppose the transductive paradigm of taking form. For a better understanding, we present an illustration of these concepts in an example: a shape-changing touch display surface for airline pilots.La recherche d’une compréhension de l’émergence d’un design sur le plan de la forme et de la matière est ancienne. On observe une montée en puissance du concept de forme dans la littérature scientifique sur le design d’interaction, en parallèle d’un tournant matériel, pour exprimer les forces à l’œuvre dans la conception. Dans cet article, nous soulignons en quoi les théories actuelles nous semblent implicitement construites sur le paradigme hylémorphique, et ce, de manière peu cohérente. Pour rendre compte plus finement des mécanismes d’élaboration et de conception dans le design d’interaction, nous proposons d’utiliser les concepts explicites de transduction, de métastabilité, d’individuation et de concrétisation, tirés de l’approche critique du schéma hylémorphique de Gilbert Simondon. Au paradigme hylémorphique de la mise en forme aristotélicienne, nous opposons ainsi celui, transductif, de la prise de forme . Pour une meilleure compréhension, nous présentons une illustration de ces concepts dans un exemple : un dispositif tactile à changement de forme pour les pilotes d’avion de ligne
An Improved ML Method to Speed Up the Trajectory Prediction: Taking Melbourne Airport as a Study Case
No full textInternational audienceThe safety and efficiency of airspace operations largely depend on the accurate prediction of 4D trajectories in dense air traffic. Traditional methods are progressively giving way to more accurate machine learning (ML) techniques, among which the Long Short-Term Memory (LSTM) neural network emerges as an exceptionally promising tool and has been successfully applied, especially for time-series prediction tasks. In this study, we introduce an LSTM-based adjustable interpolation algorithm designed to significantly reduce computational time while maintaining accuracy at an acceptable level to meet operational constraints. The algorithm applies adjustable time intervals to input data based on ascent and descent rates, providing different data densities for different flight phases. A case study focusing on flight trajectories from Melbourne to Sydney is conducted, and the findings reveal that our proposed method can reduce computation time by half without significantly sacrificing prediction accuracy compared to the traditional linear interpolation method. Furthermore, it achieves accuracy improvements of at least 50% compared to raw data processing, with no substantial increase in computational time. Proven to be effective, our proposed algorithm can be an ideal solution for training dense air traffic data when regular training is required to meet accuracy and safety requirements. This includes applications in Urban Air Mobility (UAM) and unmanned aircraft operations, as well as airport management and airspace sector handovers