Archive ouverte de l'ENAC
Not a member yet
3458 research outputs found
Sort by
Departure Flight Scheduling in Multi-Airport Systems
No full textInternational audienceThis study introduces an advanced framework for optimising collaborative departure scheduling in Multi-Airport Systems (MAS), enhancing slot assignments to minimise delays, improving throughput, and managing airspace resources, ultimately improving overall airport operational performance
Trajectory optimization for fully actuated hexacopters: Enhancing maneuverability and applications
No full textInternational audienceSeveral factors need to be taken into account when generating a feasible optimized trajectory. The optimization process heavily relies on the dynamic model of the system. Currently, there are various drone designs available, categorized based on their actuation status. In this study, we apply a trajectory optimization technique to a fully actuated hexacopter (FA-Hex), which is a new application to the best of our knowledge. This type of vehicle has been successfully integrated into several practical applications. Unlike the under-actuated hexacopter (UA-Hex), the FA-Hex can perform maneuvers with minimal banking angles, significantly enhancing the drone's maneuverability. Our research focuses specifically on trajectory optimization for the FA-Hex and demonstrates the adaptability of our method to different scenarios. We discuss two specific applications: a drone filming without a gimbal joint and a drone with a cable-suspended pendulum. We compare the simulation results with the UA-Hex model to highlight the differences in maneuverability between the two systems. The trajectory optimization is performed offline using CasADi in the MATLAB framework
Unidirectional dielectric resonator antenna using 3-D-printed uniaxial anisotropic ceramic
No full textInternational audienceA Huygens source dielectric resonator antenna (DRA) with unidirectional radiation pattern is presented. It consists of a coaxial probe exciting a rectangular, homogeneous and uniaxial anisotropic dielectric resonator (DR). To obtain a Huygens source radiation pattern, a pair of quasi-TM and TE modes are combined by controlling the permittivity tensor of the DR. A prototype operating at 2.5 GHz has been designed. The DR is made up of periodic anisotropic unit cells on a subwavelength scale and fabricated using a three-dimensional (3-D) printer. The simulated and measured results are in reasonable agreement. A relative impedance bandwidth of 12.5% and a front-to-back ratio larger than 15 dB at operating frequency are finally measured
Tropical Mathematics and Applications to Theoretical Physics and Scientific Computing
No full textInternational audienceTropical Mathematics built on Idempotent Semi-Rings and Dioids permits an extension of the usual Linear methods to Non-Linear problems and provides powerful analyzing and computing in Theoretical Physics and Applied Mathematics. Until recently, solutions in mathematics and physics were organized around algebraic structures such as groups, rings, and fields. These techniques are not well-suited to modeling and solving non-linear problems.This book covers how Idempotent Mathematics when applied appropriately can be a versatile and powerful way to transform non-linear problems into linear ones and can provide solutions to complicated theoretical physics problems
Safety First ? Ou comment le déclin cognitif dû au vieillissement pourrait être pris en compte pour l'âge de départ à la retraite ?
No full textInternational audienc
Foundation Model ColorMap : A Framework for Extracting and Visualizing theFoundation Models’ Color Knowledge
No full textInternational audienceRecent advancements in AI have led to the development of powerful models capable of processing text, images, and multimodal data. These models, often referred to as “foundation models,” include Large Language Models (LLMs) and Visual Question Answering (VQA) systems.To evaluate their capabilities, benchmarks using stimulus-response paradigms have been designed to test performance across various domains. But researchers have not thoroughly tested color knowledge. Given the role of color in data visualization and the increasing integration of natural language interaction in applications, it is essential to investigate how these models understand and represent color.The proposed framework systematically extracts and visualizes models’ internal representations of color names. It comprises two key components: a structured probing protocol and a visualization technique that maps color understanding. This method applies to LLMs through text-based analysis and to vision models via image-to-text processing. By comparing different models, we can assess their conceptual integration of color
Velocity Planning with Multi-Objectives in Displacement- Time Graphs Using Deep Reinforcement Learning
No full textInternational audienceThis paper presents a novel velocity planning method in displacement-time graphs with multiple constraints and optimization goals using deep reinforcement learning. The method formulates the velocity planning problem as a reinforcement learning task with state representation including time, position, velocity, acceleration, and distances to each obstacle triangle representative. The action space is discretized within allowable accelerations, and the kinematics ensure velocity constraints during state transitions. The advantage of this method lies in its independence from scene-specific tuning, and exhibiting robustness in various complex scenarios. Comparative analysis demonstrates a 100% success rate, along with superior computational efficiency when contrasted with the baseline approach, while also exhibiting better comfort performance. It offers a valuable alternative for velocity planning in robotics and autonomous vehicles, showcasing deep reinforcement learning's potential in practical robotics applications
When Buildings Blur the Lines: Revealing the Hidden Performance Equivalences in MANET Routing Protocols *
No full textInternational audienceThe study finds that in urban environments, the performance of non-cross-layer routing protocols (AODV, OLSR, DSDV) becomes nearly identical, contradicting earlier research that ranked their effectiveness. Using a new building-aware loss model in ns-3, we show that realistic urban conditions, unlike traditional open-field models, cause end-to-end delay to increase up to tenfold, with high variability due to unstable links. We also investigate whether adjusting parameters in open-field models could mimic urban connectivity. While this tuning matched some network connectivity metrics, it failed to replicate more complex concepts linked with propagation dynamics of cities, proving that open-field models cannot accurately simulate urban networks without fundamental changes
Role of Multi-modal Machine Learning, Explainable AI and Human-AI Teaming in Trusted Intelligent Systems for Remote Digital Towers
No full textInternational audienceRemote digital towers (RDTs) represent a transformative advancement in air traffic management (ATM), leveraging cutting-edge technology to enable remote operation by air traffic controllers (ATCOs) while improving efficiency and safety. In the context of RDTs, artificial intelligence (AI), Multimodal Machine Learning (MML) and eXplainable AI (XAI) are playing an increasingly pivotal role in enhancing operational efficiency and safety. However, several challenges need to be addressed, including the development of AI, MML and XAI, research into functional requirements, and the identification of inputs for user and machine interfaces, as well as customization options. This study explores the use of XAI in addressing specific air traffic control challenges and by offering transparent, comprehensible, and actionable insights, XAI fosters resilience, efficient, and closer collaboration between human operators and AI systems. Here, the study defines the specifications for taxiway and runway monitoring and decision support within the RDT domain. It outlines the functional requirements for customized solutions, including XAI, human-centred XAI, human-machine interfaces (HMI), and human-AI teaming (HAIT). A systematic literature review is conducted to assess transparency in AI, with a focus on explainability, HMI, and graphical user interfaces (GUI) within human-centred XAI for RDTs. Additionally, the research identifies state-of-the-art techniques for interactive data visualization, human-centric AI model development, hAIi interfaces, and HAIT, providing a multi-modal agent framework for future development in the RDT domain
Deviation results for Mandelbrot's multiplicative cascades with (stretched) exponential tails
No full textLet be a nonnegative random variable with expectation . For all , we consider the total mass of the associated Mandelbrot multiplicative cascade in the -ary tree. For all , we also consider the total mass of the measure at height in the -ary tree. Liu, Rio, Rouault \cite{lrr,liu2000limit,Rouault04} established large deviation results for for all n \in \intervallentfo{1}{\infty} (resp.\ for ) in the case has an everywhere finite cumulant generating function (resp.\ is bounded). Here, we extend these results to the case that is only assumed finite on a neighborhood of zero, and even to the case that has a stretched exponential tail. In addition, we study deviations of all orders. It is noticeable that we obtain recursive definitions of rate functions and that we resort to the moments bound instead of the standard Chernoff bound to establish the upper bounds of deviation in the infinite tree