Ecole Nationale de l'Aviation Civile
Toutes les publications de l'ENAC en direct.
All ENAC publications.
This paper focuses on Geodesic Principal Component Analysis (GPCA) on a collection of probability distributions using the Otto-Wasserstein geometry. The goal is to identify geodesic curves in the space of probability measures that best capture the modes of variation of the underlying dataset. We first address the case of a collection of Gaussian distributions, and show how to lift the computations in the space of invertible linear maps. For the more general setting of absolutely continuous probability measures, we leverage a novel approach to parameterizing geodesics in Wasserstein space with neural networks. Finally, we compare to classical tangent PCA through various...
En communications optiques spatiales satellite-sol, une problématique majeure est la traversée atmosphérique. La détection cohérente est aujourd’hui une solution privilégiée par rapport à la détection directe, car elle permet l’utilisation de modulations en phase et donc une meilleure efficacité énergétique et spectrale. Le projet CALICO (Compensation des pertes Atmosphériques de LIen laser satellite-sol en détection COhérente) a pour objectif de concevoir un récepteur cohérent capable d’assurer la démodulation homodyne d’un signal QPSK en présence d’un effet Doppler important. Le traitement réalisé doit alléger le besoin en contrôle ou correction d’erreur, et ainsi...
DVSim : un simulateur de vision événementielle pour l'apprentissage de réseaux de neurones à impulsions
DVSim : un simulateur de vision événementielle pour l'apprentissage de réseaux de neurones à impulsions
<div><p>We present an adaptive control scheme to enable the emergence of order within distributed, autonomous multi-agent systems. Past studies showed that under high-density conditions, order generated from traffic-following behavior reduces travel times, while under low densities, choosing direct paths is more beneficial. In this paper, we leveraged those findings to allow aircraft to independently and dynamically adjust their degree of traffic-following behavior based on the current state of the airspace. This enables aircraft to follow other traffic only when beneficial. Quantitative analyses revealed that dynamic trafficfollowing behavior results in lower aircraft...
<div><p>Runway capacity is one leading bottleneck in nearground flight operations due to various operational constraints, such as aircraft wake separation and runway occupancy time. With rising air travel demand and limited opportunities for infrastructure expansion, enhancing capacity and efficiency has become a critical priority. To address this, both the European Union Aviation Safety Agency (EASA) and the Federal Aviation Administration (FAA) have introduced aircraft type re-categorization to loosen the distance-based wake separation standards proposed by International Civil Aviation Organization (ICAO). Additionally, dynamic separation, which accounts for weather and...
<div><p>The measurement of arrival traffic operational complexity is essential in supporting decisionmaking processes, as it provides valuable insights into arrival traffic operations. Many studies have concentrated on measuring subjective complexity, achieving significant success in their respective areas; however, subjective complexity measurement requires the involvement of professional controllers, leading to higher costs during implementation. To reduce the cost for measuring subjective complexity, this paper analysed the collective dynamics of arrival traffic in both horizontal and vertical dimensions, and summarized several traffic patterns through a coupled...
<div><p>This study addresses flight allocation in a flightcentric air traffic management framework using two Mixed-Integer Linear Programming (MILP) models. The first model minimizes maximum complexity to balance workload, while the second minimizes interactions between flights assigned to different controllers. Tests over the Brest airspace show that the first model, despite achieving workload balance, leads to impractical allocations by assigning distant, non-interacting flights to the same controller. The second model, though more operationally feasible, results in workload imbalances, with some controllers managing high-complexity traffic while others remain...
<div><p>Neural networks have become a crucial element in modern artificial intelligence. However, despite their advancements, they often act as black boxes and can produce unexpected and incorrect results. This is why it is important to formally verify the properties of neural networks. Our work focuses on using quantization as an optimization method to accelerate the verification process. We propose a verification method for a quantized neural network (QNN) based on rational approximation of the neural network and set-based theory. We evaluate our method using the iris and HIGHWAY-ENV benchmarks, with z3 as a Satisfiability modulo theories (SMT) solver.</p></div>
<div><p>While the increasing frequency and severity of wildfires highlight the urgent need for innovative and efficient response strategies, Unmanned Aerial Vehicles (UAVs) are increasingly recognised as valuable assets in Search and Rescue (SAR) operations, particularly in natural disaster scenarios. This paper presents a novel approach that leverages a coordinated UAV team to guide inhabitants to safe locations in wildfireaffected areas. The proposed framework determines the optimal sequence of the locations for UAVs to follow by computing the fastest safe path, while considering real-world complexities, such as road conditions for evacuees and the dynamic spread of...
