Archive ouverte de l'ENAC
Not a member yet
3458 research outputs found
Sort by
EEG measures of cognitive workload: a meta-analysis
No full textInternational audienceCognitive workload (CWL) is a mental property emerging from the interaction between the individual’s limited resources and cognitive tasks. Band power analysis by electroencephalography (EEG) has been a privileged method in order to obtain an inexpensive measure (in terms of cost, implementation, and maintenance) of brain activity. Three meta-analyses were conducted to examine the effect of CWL on the three main frequency bands used in the literature: theta (4-8 Hz), alpha (8-12 Hz) and beta (13-30 Hz) bands. A total of 45 effect sizes from 24 studies involving 723 participants were computed. CWL was associated with significant effects on theta (g = 0.68, CI [0.41, 0.95]), alpha (g = -0.25, CI [-0.45, 0.04]) and beta (g = 0.50, CI [0.21, 0.79]) power. Our results suggest that theta, especially the frontal theta, can be the best index of CWL. However, heterogeneity and potential publication bias were present. These results are discussed based on the literature of cerebral oscillations and future research directions are suggested
Thermal Imaging of the Face: Mental Workload Detection in Flight Simulator
No full textInternational audienceThermography-based physiological measurement is a contact-free approach that can be particularly helpful for detecting pilots’ mental state in operational settings. In particular, thermal infrared imaging of the face is a powerful, unobtrusive and non-invasive tool that enables rapid and automatic analysis of changes in regional facial blood flow. These blood flow changes index sympathetic activity and are measured by capturing thermal imprints of particular facial regions such as nose or forehead. Although several studies suggest a relationship between mental workload and facial thermoregulation profile, evidence about this link has not yet been sufficiently investigated. In this work, we investigated how thermal measures can allow robust and continuous assessment of mental workload variations of pilots undergoing simulated flight tasks. We analyzed thermal data and heart rate of 20 participants in a flight simulator. Mental workload was modulated by the difficulty of the landing scenario or by an in-flight N-back task. Participants also performed a resting task (called cool-off) in the flight simulator. Thermal imprints did not vary significantly with landing difficulty or N-back difficulty. However, we found that the nose tip and nose area became significantly colder (signal slope was negative) during all piloting scenarios vs the rest period. Heat rate was slightly more sensitive to the piloting difficulty since it was marginally higher during the difficult vs easy landing. Results are promising but further analysis is needed to confirm that the thermal measures could identify fine-grained mental workload variations in a flight simulator setting
Modélisation du traitement visuel de l'information lors de la locomotion à l'aide de réseaux de neurones impulsionnels bio-inspirés
No full textArtificial intelligence has been booming for many years and defines the basis for all learning carried out by computer systems. Among the learning methods used are artificial neural networks. The latter's development was initially greatly motivated by the desire to understand the biological mechanisms underlying the treatments carried out by the human brain. In recent years, neural networks have become increasingly complex, with an ever-increasing number of parameters and energy consumption. While today's deep or convolutional neural networks perform excellently for many cognitive tasks, it isn't easy to use them to understand information processing in living organisms. However, some types of artificial neural networks remain more rooted in biology. These are the impulse neural networks (Spiking Neural Networks or SNN). They are based on communication mechanisms observed between neurons. The latter transmit information to each other through electrical impulses with reduced energy consumption. Learning within these SNNs can be carried out in an unsupervised way from plasticity rules observed in living organisms, such as Spike-Timing-Dependent Plasticity or STDP. This thesis aims to model visual processing during locomotion from SNs regulated by STDP-type learning and which receive data captured by asynchronous cameras whose operation is directly inspired by the retina. A first study (Fricker et al., VISAPP2022) looks at creating such a network for learning optical flow components. By interfacing this SNN with event data entries that were first simulated, we were able to highlight the network's performance for estimating the different components of the optical flow through different noise and speed conditions. The same network was later trained from actual navigation data captured using an event camera. While the task is more complicated despite convincing optical flow decoding, it highlights the need for a dataset halfway between simulated event data captured in actual conditions to get closer to natural navigation tasks while maintaining control over the data. A second study focuses on creating such a dataset for network learning. This dataset is obtained from dedicated software to extract and then model the 3D structure within different environments. This software can then simulate the spikes generated by a virtual camera moving according to varying trajectories within these environments. This approach makes it possible to control the data received by the network and thus improve learning. The first results show that the neurons of the network become selective to the different components of the optical flow while remaining invariant to the spatial properties after learning. The proposed method, therefore, provides a solid basis for processing optical flow from impulse neural networks and within different environmental contexts.Le domaine de l'intelligence artificielle est en plein essor depuis de nombreuses années et définit la base de tout apprentissage s'effectuant par des systèmes informatiques. Parmi les méthodes d'apprentissage utilisées se retrouvent les réseaux de neurones artificiels. Le développement de ces derniers était à l'origine grandement motivé par le désir de comprendre les mécanismes biologiques sous-tendant les traitements effectués par le cerveau humain. Au cours des dernières années, les réseaux de neurones se sont de plus en plus complexifiés avec un nombre de paramètres et une consommation énergétique toujours plus élevés. Si les réseaux de neurones profonds ou convolutionnels actuels ont d'excellentes performances pour de nombreuses tâches cognitives, il est désormais difficile de les utiliser comme outil pour comprendre le traitement de l'information chez le vivant. Certains types de réseaux de neurones artificiels restent néanmoins plus ancrés dans la biologie, il s'agit des réseaux de neurones impulsionnels (Spiking Neural Networks ou SNN en anglais). Ils reposent sur des mécanismes de communication observés entre les neurones. Ces derniers se transmettent des informations sous la forme d'impulsions électriques dont la consommation énergétique est très réduite. L'apprentissage au sein de ces SNN peut s'effectuer de façon non-supervisée à partir de règles de plasticité observées chez le vivant comme la Spike-Timing-Dependent Plasticity ou STDP. Cette thèse a pour objectif de modéliser le traitement visuel lors de la locomotion à partir de SNNs régulés par un apprentissage de type STDP et qui reçoivent des données captées par des caméras asynchrones dont le fonctionnement s'inspire directement de celui de la rétine. Une première étude (Fricker et al., VISAPP2022) se penche sur la création d'un tel réseau pour l'apprentissage des composantes de flux optique. En interfaçant ce SNN avec des entrées de données événementielles d'abord simulées, nous avons pu mettre en évidence les performances du réseau pour l'estimation des différentes composantes du flux optique à travers différentes conditions de bruit et de vitesse. Le même réseau a par la suite été entrainé à partir de données de navigation réelles capturées à l'aide d'une caméra événementielle. Si la tâche s'avère plus difficile malgré un décodage du flux optique convaincant, elle met en évidence le besoin d'un jeu de données à mi-chemin entre des données événementielles simulées et capturées en condition réelle afin de se rapprocher de tâches de navigation naturelle en gardant un contrôle sur les données. Une seconde étude s'intéresse à la création d'un tel jeu de données pour l'apprentissage du réseau. Ce jeu de données est obtenu à partir d'un logiciel dédié permettant d'extraire puis de modéliser la structure 3D au sein de différents environnements. Ce logiciel peut ensuite simuler les spikes générés par une caméra virtuelle se déplaçant selon différentes trajectoires au sein de ces environnements. Cette approche permet de bien contrôler les données reçus par le réseau et d'améliorer ainsi l'apprentissage. Les premiers résultats montrent que les neurones du réseau après apprentissage deviennent sélectifs aux différentes composantes du flux optique tout en restant invariant aux propriétés spatiales. La méthode proposée constitue donc une base solide pour le traitement du flux optique à partir de réseau de neurones impulsionnels et au sein de différents contextes environnementaux
Cognitive effects of prolonged continuous human-machine interaction: The case for mental state-based adaptive interfaces
No full textInternational audienceOperators of complex systems across multiple domains (e.g., aviation, automotive, and nuclear power industry) are required to perform their tasks over prolonged and continuous periods of time. Mental fatigue as well as reduced cognitive flexibility, attention, and situational awareness all result from prolonged continuous use, putting at risk the safety and efficiency of complex operations. Mental state-based adaptive systems may be a solution to this problem. These systems infer the current mental state of an operator based on a selection of metrics ranging from operator independent measures (e.g., weather and time of day), to behavioral (e.g., reaction time and lane deviation) as well as physiological markers (e.g., electroencephalography and cardiac activity). The interaction between operator and system may then be adapted in one of many ways to mitigate any detected degraded cognitive state, thereby ensuring continued safety and efficiency. Depending on the task at hand and its specific problems, possible adaptations -usually based on machine learning estimations- e.g., include modifications of information, presentation modality or stimuli salience, as well as task scheduling. Research on adaptive systems is at the interface of several domains, including neuroergonomics, human factors, and human-computer interaction in an applied and ecological context, necessitating careful consideration of each of the aforementioned aspects. This article provides an overview of some of the key questions and aspects to be considered by researchers for the design of mental state-based adaptive systems, while also promoting their application during prolonged continuous use to pave the way toward safer and more efficient human-machine interaction
Représentation des systèmes dynamiques : Notes de cours
No full textEngineering schoolC'est Henri Poincaré (1854 - 1912) qui est très vraisemblablement à l'originede la théorie des systèmes dynamiques grâce à ses travaux sur le traitementdu problème des trois corps en 1889. Rappelons que ce problème consiste àdéterminer les trajectoires de trois corps mutuellement soumis à l'attractiongravitationnelle et qu'il n'est pas possible de le résoudre analytiquement engénéral, seules des solutions approchées peuvent être calculées.De manière très générale un système dynamique est un système qui évolueavec le temps et la théorie des systèmes dynamiques propose des techniquespermettant de prévoir le comportement dans le temps d'un système à partir deslois d'évolution innitésimal
Du déchiffrement des hiéroglyphes à l'interprétation de la mécanique quantique
No full textNational audienceOn the occasion of the bicentary of the deciphering of hieroglyphics and the centuty (99 years) of the introduction of Louis de Broglie's function, this article show that there is an analogy between the difficulties of deciphering hieroglyphics by Thomas Young and Jean-François Champollion and the difficulties of interpreting quantum mechanics. These difficulties arise in both cases from the existence of a double solution. Taking into account this double solution in quantum mechanics (external wave function and internal wave function) allows to understand simply the wave-particle duality and to synthesize the points of view of Louis de Broglie, Erwin Schrödinger, Marx Born and Albert Einstein.A l'occasion du bicentenaire du déchiffrements des hiéroglyphes et du centenaire (99 ans) de l'introduction de la fonction d'onde de Louis de Broglie, nous montrons qu'il existe une analogie entre les difficultés du déchiffrement des hiéroglyphes par Thomas Young et Jean-François Champollion et les difficultés d'interprétation de la mécanique quantique. Ces difficultés proviennent dans les deux cas de l'existence simultanée de deux solutions demandant deux clés de lecture différentes. On montre que la prise en compte en mécanique quantique de cette double solution (fonction d'onde externe et fonction d'onde interne) permet de comprendre simplement la dualité onde-particule et de faire la synthèse des points de vue de Louis de Broglie, d'Erwin Schrödinger, de Marx Born et d'Albert Einstein
How to increase efficiency and capacity in a single runway airport? Study of the case: Viracopos Campinas Airport-VCP
No full textThe Viracopos-Campinas International Airport (ICAO code: SBKP, IATA code: VCP) is an international airport in the city of Campinas, in São Paulo State. The airport settled in a region of the leading technological, logistical, and economic development. The Viracopos Cargo Terminal was elected in 2017 the second-best in the World by ACE (Air Cargo Excellence Awards) in the category up to 400 thousand tons/year and is one of the most agile and safe in Brazil. Viracopos has a new and modern passenger terminal with a capacity of 25 million people per year. Viracopos has verified growth in passengers handled since 2008, with the start of operations of Azul Airlines. In 2012, the airport was conceded to Aeroportos Brasil Viracopos to manage, expand and modernize the airport complex for 30 years. In the first year of private operation, the airport transported 8.8 million passengers from the old terminal. In 2014, there were almost 10 million, and in 2019 Viracopos reached 10.5 million passengers after a prolonged recession in Brazil's economy. The airport always seek a continuous and constant growth of passengers and cargo. However, in 2020 the COVID19 pandemic devastated the aviation industry, generating economic loss due to passengers and aircraft movements' downturn. More than ever, improve operations efficiency and capacity is one of the way outs for the crises. This study aims to understand the VCP terminal airspace procedures, runway configuration, apron configuration, operations characteristics and study best practices applied in the single-runway global benchmark. After that, make recommendations to possible improvements in approach and departure procedure and airside processes and infrastructure to increase operations efficiency and runway throughput in VCP
Mitigation measures and proposed alternatives to allow small Brazilian regional airports with reduced runway strip width to conduct IFR Non-Precision Operations
No full textRegional aviation in Brazil is driving new opportunities for airlines, not only because they can serve a new market, but also to attract new customers and thus feed their traditional domestic and international routes. It has been proven that, in Brazil, economic growth in the interior regions is now faster than what happens in large cities, which creates a favorable environment for commercial airlines, as they are a vector of development for these places. However, the costs to solve infrastructure problems in small Brazilian regional airports, especially on Amazon and Northeast regions, to comply with national civil aviation regulations, are so expensive that sometimes it is impossible to provide services in some places. This mainly happens with non-precision runway strips provisions, which usually require large earthworks, environmental licenses, and expropriations, raising the expenditure and time to open small airports for commercial flights. Address the situation presented above required, in addition to many technical discussions, a review of applicable national regulation, the assessment of possible flexibilizations already considered, and an evaluation on essential infrastructure and possible mitigation measures, as well as the assessment of related studies on the topic. It was also collected the perception on current situation and what is considered possible by aviation sector, to allow a more qualitative approach and then proceed with possible solutions, including actions planned in short and medium term, in order to allow commercial operations (aircrafts up to code 3C) at small regional airports that do not meet non-precision runway strips provisions
Regulatory outcome evaluation of rescue and fire fighting deregulation for small airports
No full textAircraft Rescue and Fire Fighting (ARFF) is an important component of an airport safety system. It is also costly, especially for small airports that process less than 200 thousand passengers per year. In December 2017 ANAC, the Brazilian Civil Aviation Authority, enacted a new rule that disengaged Rescue and Fire Fighting as a requirement for small airports that intend to have scheduled flight operations. As in any safety rule, there are rationales that led to such regulatory effort and targets to be reached by these means. The aim of this paper is to assess whether the objectives of such deregulation have been achieved as of December 2020
COVID-19 and state aid and subsidies for airlines engages on international air transport: a legal approach on fair competition
No full textSubsidies on aviation has been a practiced observed since the beginning of the activity but has gained more relevance since the outbreak of COVID-19, when aviation was brutally hit and the future of air transport became jeopardized. Aiming to help airlines to modulate impacts of the crisis, governments engaged on subsidies programs in many different forms of governmental aid. While international trade of goods counts on a developed system of subsidies discipline to avoid tilting on the level playing field, international air transport does not. It is then foreseeable that in a near future claims on “harmful subsidies” would be proposed. The study review core concepts of subsidies and fair competition on international fora in order to check whether the existing legal framework of WTO, ICAO and Domestic Competition Law of Brazil would be able to address such claims, especially when subsidies were granted on global scale, and observe if different systems would offer a consistent answer to a hypothetical claim proposed. The main findings of the study are that the WTO system is not applied to international air transport, that the absence of multilateral regulation under ICAO about the minimum on the subject may lead to an impossibility of the Organization to deliver a useful solution to similar cases, and finally, domestic regulation, at least in case of Brazil, could be a possibility of jurisdiction in limited cases but with a more restorative answer to plaintiffs than International Organizations