Archive ouverte de l'ENAC
Not a member yet
3458 research outputs found
Sort by
Usage of more transparent and explainable conflict resolution algorithm: air traffic controller feedback
No full textEdited by Mickaël CausseInternational audienceRecently, Artificial intelligence (AI) algorithms have received increasable interest in various application domains including in Air Transportation Management (ATM). Different AI in particular Machine Learning (ML) algorithms are used to provide decision support in autonomous decision-making tasks in the ATM domain e.g., predicting air transportation traffic and optimizing traffic flows. However, most of the time these automated systems are not accepted or trusted by the intended users as the decisions provided by AI are often opaque, non-intuitive and not understandable by human operators. Safety is the major pillar to air traffic management, and no black box process can be inserted in a decision-making process when human life is involved. To address this challenge related to transparency of the automated system in the ATM domain, we investigated AI methods in predicting air transportation traffic conflict and optimizing traffic flows based on the domain of Explainable Artificial Intelligence (XAI). Here, AI models’ explainability in terms of understanding a decision i.e., post hoc interpretability and understanding how the model works i.e., transparency can be provided for air traffic controllers. In this paper, we report our research directions and our findings to support better decision making with AI algorithms with extended transparency
Make-A-Morph: Exploring the design space of inflatable devices made from planar fabric
No full textInternational audienceDeveloping inflatable devices from planar fabric is a new versatile fabrication process that allows the development of complex geometric shapes with a beneficial mass to robustness ratio. However, designing and fabricating with this matter is complex, and the existing design primitives for shape change can constrain designers' creativity. We present a pipeline that allows users and designers to explore and compose with various shape-change primitives. To this extent, we rely on digital simulation combined with a simple digital fabrication tool. This pipeline allows to explore and visualize deformation and develop new application cases for shape-changing interfaces. We propose a workshop around manipulating these tools to foster discussion between designers and researchers around the future of shape-changing interface fabrication
A numerical proof by reliable Global Optimization for a problem of covering a rectangle with circles
No full textInternational audienceIn this paper, we show how a reliable global Branch and Bound optimization method based on interval arithmetic can be used efficiently to numerically prove a conjecture in geometry about how to cover a rectangle by 6 circles of equal radius
Hierarchy in the cockpit: How captains influence the decision-making of young and inexperienced first officers
No full textInternational audienceThe present study aimed at investigating the extent to which Captains' risky decisions influence young and inexperienced First Officers. Participants (i.e., student pilots who had almost completed their training) had to decide, alone or in a crew configuration, whether to continue or abort the landing according to four risk levels (safe, moderately risky, highly risky and extremely risky). In the lone pilot configuration, they made their decisions by themselves, while in the crew configuration they were paired with a Captain who acted as a risk taker and almost always chose to land (except in extremely risky situations). The Captain's mere presence led participants to increase their risk-taking in moderately risky situations (before they even knew the Captain's decision), supposedly in an attempt to look competent and impress their superior. In reaction to the Captain's decision to land, participants also increased their risk-taking in highly risky situations. This tendency was positively correlated to the perceived authority of the Captain. Surprisingly, some participants sometimes insisted on continuing the landing in extremely risky situations after the Captain asked for a go-around, suggesting that some pilot students may greatly overestimate their piloting skills (i.e., Dunning Kruger effect). Some applications of the present experimental protocol as training for student pilots are proposed
Use of 5G and mmWave radar for positioning, sensing, and line-of-sight detection in airport areas
No full textInternational audienceThis paper explores innovative low-cost technologies, widely used outside of Air Traffic Management (ATM), for use in airport surface surveillance. These technologies consist of a 5G-signal-based surveillance solution and a millimeter wave (mmWave) radar augmented with artificial intelligence (AI). The 5G solution is based on the combination of 3D Vector Antenna, innovative signal processing techniques, and hybridization techniques based on time-of-arrival and angle-ofarrival estimates with uplink and downlink 5G signals, as well as Machine Learning (ML)-based Line of Sight (LOS) detection algorithms. The mmWave solution is based on mmWave radar for non-cooperative target's positioning and sensing, combined with deep learning for objects classification. Standalone 5G positioning accuracy reaches m-level accuracy in LOS scenarios and it is better with downlink reference signals than with uplink ones, while it deteriorates quite drastically in NLOS scenarios. LOS detection accuracies above 84% average accuracy can be achieved with ML. The mmWave radar is tested in different scenarios (short, medium and long range) and it provides cost-effective surface surveillance up to few hundred meters (depending on the object radar cross section RCS) with ±60°field of view. The work is being conducted within the H2020 European-funded project NewSense and it delves into the 5G, Vector Antennas, mmWave, and ML/AI capabilities for future ATM solutions
Modèle paramétrique pour estimer la dégradation de la fiabilité de démarrage d’une turbine à gaz à partir de données d’essais
No full textNational audienceL’objectif de cette étude est d’estimer la probabilité de démarrage d’un turbomoteur en tout point de l’espace des facteurs jugés influents. En l’absence de modèle physique du démarrage et de nombre limité de données, une méthode générale d’estimation de la probabilité de démarrage à travers un modèle de dégradation paramétrique est proposée. Le modèle est régi par la distance à un point où la probabilité de démarrage est 1. La sélection du modèle revient à estimer la matrice définissant la distance dans l’espace des facteurs influents. Cette matrice est estimée par maximum de vraisemblance. Une application industrielle à partir des données d’essais de turbomoteur d’hélicoptère est présentée. Les résultats numériques sont prometteurs mais ont montré qu’il est nécessaire de considérer une classe plus générique de modèle. Enfin à travers une étude de robustesse, le modèle a aussi montré une stabilité à des faibles perturbations des données d’essais
The Nearest Is Not The Fastest : On The Importance Of Selecting In/Out Routing Hops Over A Satellite LEO Constellation
No full textInternational audienceThis study investigates the importance of choosing the first (respectively last) hop to access (respectively to exit) a Low Earth Orbit (LEO) satellite constellation, which is of upmost importance for the LEO routing performance. Usually, basic routing strategies connect a ground station to its nearest satellite, and this strategy does not always lead to the optimal routing path. We propose to select this first/last satellites within a subset of knearest satellites. After performing routing simulations over one of the next-generation satellite constellations, preliminary results show that this in/out hop selection strategy leads to a better link capacity usage and a lower data loss rate, allowing a faster TCP bulk data transfer
On local-global hysteresis-based hovering stabilization of the DarkO convertible UAV
No full textInternational audienceWe characterize an input-affine model of the UAV DarkO: a convertible drone designed and developed at the Ecole Nationale de L'Aviation Civile (ENAC) in Toulouse (France). Starting from a nonlinear model available in the literature, we present an approximate input-affine nonlinear model, whose dynamics simplifies the control design task. For this simplified model, we characterize the hovering equilibria in the absence of wind, and we derive the corresponding linearized dynamics. Then present a hysteresis-based switching mechanism combining a nonlinear feedback (providing a large basin of attraction) with a linearized feedback (providing improved performance but a smaller basin of attraction). Simulation results, using the original nonlinear model, confirm the effectiveness of the proposed feedback design
Complexity Optimal Air Traffic Assignment in Multi-layer Transport Network for Urban Air Mobility Operations
No full textInternational audienc
Aircraft Conflict Resolution Using Convolutional Neural Network on Trajectory Image
No full textInternational audienceA situation between several moving aircraft is a conflict when their position is less than the internationally specified distance. To solve aircraft conflicts, air traffic controllers consider many parameters including the positioning coordinate, speed, direction, weather, etc. of the involved aircraft. This is a complex task, specifically considering the increase of the traffic. Assisting systems could help controllers in their tasks. Most conflict resolution models are based on trajectory data of a fixed number of input aircraft. Under this constraint, it is possible to resolve conflicts using machine learning models, including convolutional neuron network models. Such models cannot resolve conflicts that imply a variable number of aircraft because the input size of the model is fixed. To solve this challenge, we transformed the trajectory data into images which size does not depend on the number of planes. We developed a multi-label conflict resolution model that we named ACRnet, based on a convolutional neural network to classify the obtained images. ACRnet model achieves an accuracy of 99.16% on the training data and of 98.97% on the test data set for two aircraft. For both two and three aircraft, the accuracy is 99.05% (resp. 98.96%) on the training (resp. test) data set