Archive ouverte de l'ENAC
Not a member yet
3458 research outputs found
Sort by
Why and when do you look away when trying to remember? Gaze aversion as a marker of the attentional switch to the internal world during memory retrieval
No full textInternational audienceIt is common to look away while trying to remember specific information, for example during autobiographical memory retrieval, a behavior referred to as gaze aversion. Given the competition between internal and external attention, gaze aversion is assumed to play a role in visual decoupling, i.e., suppressing environmental distractors during internal tasks. This suggests a link between gaze aversion and the attentional switch from the outside world to a temporary internal mental space that takes place during the initial stage of memory retrieval, but this assumption has never been verified so far. We designed a protocol where 33 participants answered 48 autobiographical questions while their eye movements were recorded with an eye-tracker and a camcorder. Results indicated that gaze aversion occurred early (<1s) and predominantly during the access phase of memory retrieval-i.e., the moment when the attentional switch is assumed to take place. In addition, gaze aversion lasted a relatively long time (on average 6 seconds), and was notably decoupled from concurrent head movements. These results support a role of gaze aversion in perceptual decoupling. Gaze aversion was also related to higher retrieval effort and was rare during memories which came spontaneously to mind. This suggests that gaze aversion might be required only when cognitive effort is required to switch the attention towards the internal world to help retrieving hard-to-access memories. Compared to eye vergence, another visual decoupling strategy, the association with the attentional switch seemed specific to gaze aversion. Our results provide for the first time several arguments supporting the hypothesis that gaze aversion is related to the attentional switch from the outside world to memory.</div
Authenticating civil UAV communications with post-quantum digital signatures
No full textInternational audienceThe integration of Unmanned Aerial Systems (UAS) into public airspace requires a secure UAV Traffic Management (UTM) system. Providing information security to communications is crucial for the UTM’s operational safety. Ensuring the integrity and authentication of multicast data, such as emergency information and telemetry, is essential to prevent tampering and impersonation. Digital signatures provide an ideal solution for securing this data. However, implementing digital signature standards on small Unmanned Aerial Vehicles (UAVs) poses challenges due to hardware limitations. To address this, lightweight cryptographic algorithms have been developed for resource-constrained hardware. However, both lightweight and standard cryptographic algorithms are vulnerable to quantum computing attacks. To mitigate this vulnerability, post-quantum cryptographic primitives are being standardized. This study focuses on implementing post-quantum digital signature standards, Falcon and CRYSTALS-Dilithium, on constrained hardware resembling small UAVs. The performance of these algorithms in terms of computation time is evaluated considering real-time tasks during flights. The results demonstrate the feasibility of using unmodified standards on resource-constrained hardware alongside other priority tasks. Additionally, investigating algorithms with a reduced security level for telemetry broadcasts is considered. Falcon512 and CRYSTALS-Dilithium can both be used in this environment. Falcon512 is preferred due to its constancy in execution time and its shorter signatures and public keys
Easing the tuning of drone autopilots through a model-based framework
No full textInternational audienceOff-the-shelf open-source autopilots are customized by practitioners to satisfy their customer’s specific needs. When custom functions require low delays and/or fast frequency, in the order of magnitude of hundreds or a couple of thousand hertz, they may impact the behavior of the underlying stabilization loop. This paper proposes a tool chain able to extract a model using a Domain-Specific Language (DSL) based on AADL (Architecture Analysis & Design Language) semantics, extended with specific needs to capture the internal behavior of autopilots. This extraction is done directly during the compilation process of the autopilot. Then, we apply on the model of an autopilot a tool to assign offsets for offset-free systems
Poster - Sim2Learn: Simulation of an Electron Microscope to facilitate Mental to Conceptual Model convergence
No full textInternational audienceScientific instruments such as electron microscopes allow rapid development in many fields of science but are difficult to fully exploit. Even for experienced users, their internal state (shape of the electron beam), represented in conceptual models, remains fairly inaccessible. This prevents the operator from quickly understanding the complete causal link between the observed phenomenon and the resulting image. This paper investigates how a simulation and visualization application of such an instrument, may help users bridge the gap between their mental and the conceptual models of the instrument by displaying an approximation of its internal state in real time. This is supported by a study with 10 microscopists who had to perform a series of plausible tasks on the interface and answer questions. The study shows that users have a better understanding of the instrument’s inner state while using the user interface. These results will help software designers develop the next generation of scientific instrument tools
Etude du détournement du regard comme marqueur oculaire de la bascule attentionnelle du monde extérieur vers le monde mental interne
No full textAttention can be either directed to the external world-for example, while reading this manuscript (external attention)-or to the internal mental world containing memories and thoughts (internal attention). Internal attention is involved in various cognitive tasks such as memory retrieval and mind-wandering among others. Since attentional resources are limited, the two forms of attention compete and attention must switch from one world to the other a considerable number of times per day. An attentional switch towards the internal world is accompanied by a disengagement from the external world-a phenomenon known as perceptual decoupling. There has been a growing desire to discover an objective behavioral or physiological marker that can help detect the attentional switch. Such a marker would be useful in various domains, for instance being able to monitor attention in tasks where it needs to be maintained towards the external world. Eye movements, as indicators of the direction of attention, seem to be promising candidates: internal attention is associated with ocular behaviors that reduce the processing of visual information, in particular gaze aversion. To investigate this issue, this thesis involves a collaboration between Brain and Cognition Research Center-interested in a better understanding of the attentional switch, its role in the cognitive cascade during autobiographical memory retrieval, and ocular markers associated with it-on one hand, and the National School of Civil Aviation-interested in an application aiming to monitor the attention of aeronautical operators-on the other. This work starts with a literature review on the attentional switch, which lies at the interface between multiple domains: attention, autobiographical memory, mind-wandering, and eye movements. The convergence of the different domains allows for a proposal of a new hypothetical framework suggesting gaze aversion-the direction of the gaze to a neutral space during memory recall-as a behavioral marker of the attentional switch for detection at a single-trial level. Although common and widespread, this behavior has rarely been studied scientifically. Thus, we first conducted a study based on three online experiments (n = 160) which showed that the direction of the gaze, when averted, is used as a social cue to distinguish the states of internal and external attention. We followed up with a second behavioral study, where eye movements were recorded during autobiographical memory retrieval (n = 32). First, the results supported a relationship between gaze aversion and the attentional switch. Second, since the detection of the attentional switch would require a fine definition of eye markers, the study also characterizes gaze aversion which: lasts an average of 6 sec, is not accompanied by head movement, and occurs in various directions. This characterization however remains incomplete due to technical constraints. Since no eye-tracker is designed to specifically measure gaze aversion, we also tested other recording methods in our third preliminary study (n = 4), which suggested the potential of electrooculography. Finally, to promote the study of the attentional switch in aeronautics, a fourth pilot study (n = 16), inspired by the literature on involuntary autobiographical memories, proposes a protocol during which memory cues are presented during an air traffic control simulation task to induce memories. Given the similarities between autobiographical retrieval and mind-wandering, we consider that this is an appropriate approach to indirectly study mind-wandering episodes while avoiding the biases of self-report methods. This study also made it possible to observe gaze aversions in a simulation environment, a promising milestone encouraging its utility in an ecological setting. In conclusion, this thesis has highlighted the strong potential of gaze aversion and opens up perspectives for future fundamental and applied research.L'attention peut être orientée vers le monde extérieur, par exemple pendant la lecture de ce manuscrit (attention externe), mais elle peut aussi être orientée vers le monde mental interne (attention interne) pour diverses tâches cognitives telles que le rappel en mémoire ou l'errance mentale. Les ressources attentionnelles étant limitées, les deux formes d'attention entrent en compétition entraînant une bascule entre les mondes externe et interne de nombreuses fois par jour. La bascule attentionnelle vers le monde interne s'accompagne d'un désengagement qui rend "aveugle" au monde extérieur, un phénomène bien documenté appelé découplage perceptif. Il serait donc utile de pouvoir détecter cette bascule attentionnelle. Il y a, en ce sens, une volonté grandissante de découvrir un marqueur comportemental ou physiologique objectif de celle-ci. Le mouvement des yeux, indicateur de la direction de l'attention, semble être un candidat potentiel : l'attention interne est en effet associée à des comportements oculaires qui réduisent le traitement visuel, en particulier le détournement du regard, c'est-à-dire, le fait de diriger le regard vers un endroit neutre de l'espace, Pour investiguer cette problématique, ce travail de thèse implique une collaboration entre le Centre de Recherche Cerveau et Cognition -intéressé par la compréhension de la bascule attentionnelle, des marqueurs oculaires qui lui sont associés ainsi que de son rôle dans la cascade cognitive pendant le rappel en mémoire autobiographique- d'une part, et l'École Nationale d'Aviation Civile -intéressée par une application pour monitorer l'attention d'opérateurs aéronautiques- d'autre part. Cette thèse débute par une revue de la littérature sur la bascule attentionnelle située à l'interface entre différents domaines : l'attention, la mémoire autobiographique, l'errance mentale et les mouvements des yeux. La convergence des différents domaines permet de proposer un nouveau cadre de travail hypothétique suggérant que le détournement du regard pourrait être un marqueur comportemental de la bascule qui soit détectable en temps réel sur un essai unique. De manière surprenante, ce phénomène fréquent et répandu n'avait jusqu'alors été que peu étudié scientifiquement. Nous avons donc mené une première étude basée sur trois expériences sur internet (n = 160) qui a mis en évidence que la direction du regard, quand il est détourné, est un indice social utilisé pour distinguer les états d'attention interne et externe. Nous avons poursuivi avec une deuxième étude comportementale, où les mouvements des yeux ont été enregistrés pendant le rappel en mémoire autobiographique (n = 32). En premier lieu, les résultats supportent la relation entre détournement du regard et bascule attentionnelle. Ensuite, cette étude établit également pour la première fois les caractéristiques du détournement du regard : il dure en moyenne 6 sec, n'est pas suivi de mouvement de la tête et apparaît dans diverses directions. Cette caractérisation demeure incomplète en raison de contraintes techniques: puisqu'aucun oculomètre n'a été conçu pour mesurer les détournements du regard, nous avons testé d'autres méthodes d'enregistrement. Notre troisième étude, préliminaire à ce stade (n = 4), suggère l'intérêt de l'électro-oculographie. Enfin, pour promouvoir l'étude de la bascule attentionnelle dans le milieu aéronautique, une quatrième étude (n = 16), inspirée de la littérature sur les souvenirs autobiographiques involontaires, propose un nouveau protocole durant lequel des indices de mémoire sont présentés pendant une tâche de simulation de contrôle aérien pour induire des souvenirs. Nous considérons qu'il s'agit d'un bon modèle pour étudier indirectement les épisodes d'errance mentale en évitant les biais des méthodes d'auto-évaluation. En conclusion, cette thèse a mis en exergue le fort potentiel du détournement du regard et a ouvert des perspectives pour de futurs travaux fondamentaux et applicatifs
Improving AI Monitoring of Early Life Satellites Using Transfer Learning
No full textInternational audienceIn the last decades, many space domain actors such as the Centre National d'Etudes Spatiales (CNES) have begun to use Artificial Intelligence to monitor spacecraft housekeeping telemetry. These novel techniques are able to identify atypical behaviours and potential satellite anomalies that cannot be detected by more standard monitoring approaches. However, AI methods have an important drawback: they need a significant amount of data to be able to "learn" the nominal behaviour of a spacecraft and then detect novelties in new telemetry, which is not suitable for a satellite in the beginning of life where in-flight telemetry is very scarce. One way to bypass the scarcity of data is Transfer Learning (TL). Depending on the use case, operators may have already-available telemetry either from on-the-ground Assembly, Integration, and Test (AIT) of the spacecraft, from full-digital or hybrid simulators, or from in-flight telemetry of one or multiple "twin-spacecraft" in case of a constellation with already-launched units. This already-available telemetry is often close, but not perfectly similar, to in-flight telemetry of the newly-launched spacecraft to be monitored. The idea of TL is therefore to use this large and existing database (the source database), coupled with the first in-flight telemetry from the new spacecraft (the target database), to be able to mathematically-design a relevant AI learning model. In 2022, CNES and TéSA laboratory have worked together and have identified two TL methods to detect anomalies in telemetry of early life satellites with few data, by working directly on the telemetry dataset (the learning domain) or on the model learned from the target database. The first TL method consists in mathematically modifying the decision boundary estimated by a One-Class Support Vector Machine (OC-SVM) algorithm applied to the source database to match the target database. The second method based on "Domain Transfer" consists in building an "extended" learning domain made up with the relevant data from both the source and target databases, which is used to build a learning model. These two algorithms have been evaluated with real Earth Observation satellite telemetry. The preliminary outcomes of this research show promising results. Further work will consist in implementing these methods operationally so that AI monitoring methods can be used from the very beginning-of-life of CNES satellites. The main conclusion of this work is that TL can be an interesting tool to monitor spacecraft housekeeping telemetry during the first 6 months after the launch of a satellite
Labeling mental fatigue for passive BCI applications: Accuracy vs applicability tradeoff
No full textInternational audienc
Dynamic Air Traffic Flow Coordination for Flow-centric Airspace Management
No full textInternational audienceThe air traffic control paradigm is shifting from sectorbased operations to cross-border flow-centric approaches to overcome sectors' geographical limits. Under this paradigm, effective air traffic flow coordination at flow intersections is crucial for efficiently utilizing available airspace resources and avoiding inefficiencies caused by high demand. This paper proposes a dynamic air traffic flow coordination framework to identify, predict, assess, and coordinate the evolving air traffic flows to enable more efficient flow configuration. Firstly, nominal flow intersections (NFI) are identified through hierarchical clustering of flight trajectory intersections and graph analytics of daily traffic flow patterns. Secondly, spatialtemporal flow features are represented as sequences of flights transiting through the NFIs over time. These features are used to predict the traffic demand at the NFIs during a given future period through a transformer-based neural network. Thirdly, for each NFI, the acceptable flow limit is determined by identifying the phase transition of the normalized flight transition duration from its neighboring NFIs versus the traffic demand. Finally, when the predicted demand exceeds the flow limit, by evaluating the available capacity at different NFIs in the airspace, the flow excess is alternated onto other NFIs to optimize and re-configure the air traffic demand to avoid traffic overload. An experimental study was carried out in French airspace using the proposed framework base on the ADS-B data in December 2019. Results showed that the proposed prediction model approximated the actual flow values with the coefficient of determination (R 2) above 0.9 and mean absolute percentage error (MAPE) below 20%. Acceptable flow limit determination showed that for above 68% NFIs, the flight transition duration increases sharply when the demand exceeds a certain level. The flow excess at an NFI whose demand was predicted to exceed its limit was coordinated, and the potential increase in the flight transition duration caused by the flow excess was avoided
On the Network Characterization of Nano-Satellite Swarms
No full textInternational audienceLow-frequency radio interferometry is crucial to understanding the universe and its very early days. Unfortunately, most of the current instruments are ground-based and thus impacted by the interferences massively produced by the Earth. To alleviate this issue, scientific missions aim at using Moonorbiting nano-satellite swarms as distributed radio-telescopes in outer space, keeping them out of Earth interference range. However, swarms of nano-satellites are systems with complex dynamics and need to be appropriately characterized to achieve their scientific mission. This paper presents a methodology based on graph theory for characterizing the swarm network system by computing graph theory metrics around three properties: the node density, network connectivity and ISL availability. We show that these properties are well-suited for highlighting a possible heterogeneity in a network and adapt a routing strategy accordingly. This work is the first milestone in defining the best-suited routing strategy within the swarm from the derived network properties
Forecasting YouTube QoE over SATCOM
No full textInternational audienceWe investigate the feasibility of using machine learning methods for predicting the Quality of Experience (QoE) of end users in the context of video streaming over satellite networks. To achieve this, we analyzed QoE and traffic data from 2,400 YouTube video sessions over emulated geosynchronous (GSO) satellite links. The objective is to determine whether existing learning methods, originally developed for wired or mobile networks, can be adapted to accurately predict key QoE factors over SATCOM. We particularly investigate a specific existing framework, which achieves outstanding performance in predicting resolution and initial delay. However, we point out some discrepancies in their hypothesis, leading to optimistic forecasting results. We then refine their methodology to ensure a complete independence between training and test datasets, leading to a fairer QoE video streaming forecast over satellite networks