HAL Portal UTC Université de Technologie de Compiègne
Not a member yet
11652 research outputs found
Sort by
Jouer à cache-cache avec les algorithmes dans le "GayTok" : du shadowbanning aux affordances de la plateforme
No full textInternational audienceThis chapter investigates algorithmic bypass strategies that sexual and gen- dered communities set up on TikTok to post sexual content on the platform. Focusing on the gay community, the chapter proposes a typology of platform affordances through sound, text and image. The analysis draws on a digital ethnography method consisting of “growing a niche algorithm” to study content recommendation and in- visibilisation on social media. Based on an epistemology of the closet, the chapter concludes with a discussion on online visibility and virality as matters of adjustment and negotiation
Optimization of Liquid-State Anaerobic Digestion by Defining the Optimal Composition of a Complex Mixture of Substrates Using a Simplex Centroid Design
No full textInternational audienceThis study aimed to define the optimal composition of three heterogeneous substratesof the anaerobic digestion process to maximize methane production. The investigated substrateswere sewage sludge (SS), the organic fraction of municipal solid waste (OFMSW), and horse waste(HW). The optimal composition of these substrates was defined using the mixture design and, morespecifically, the simplex–centroid mixture design. Customized methods and materials were employedto study the complex mixture design of these substrates. The findings revealed that the optimalmixture involved all three substrates with the composition 0.17 HW, 0.66 SS, and 0.17 OFMSW, whichdemonstrated the highest methane yield at 269 NmL·gVS−1. In addition, a mathematical model wasdeveloped to predict methane production based on a specific composition of co-substrates. Theresults were validated at the small pilot scale
Operational Design Domain Monitoring with Uncertain Measurements
No full textInternational audienceIntelligent vehicles are constantly increasing their automation levels; thus, it is necessary to ensure that such vehicles operate under conditions designed to guarantee correct and safe operation. This paper proposes a method to address this challenge using the Operational Design Domain (ODD) concept and vehicle onboard real-time monitoring. The proposed method comprises a framework that estimates the state of the Operational Domain (OD) while considering measurement uncertainty to determine whether the vehicle is within its ODD. This method utilizes fuzzy sets to represent the vagueness of real-world driving environments. We also introduce a taxonomy-based approach to formally define ODDs with respect to vehicle capabilities, independent of the architecture. The approach is demonstrated through simulation use cases, by monitoring the vehicle's OD at runtime and determining when it exits its defined ODD. The results pave the way for the introduction of verifiable safety rules based on OD membership degrees
Navigation partagée dans un système autonome cybernétique multi-agents
No full textMy thesis is based on shared navigation between the autonomous system and the human. In our research, we focus on command fusion. In our approach, both entities, the human and the autonomous system, simultaneously control the vehicle, and a module acquires their commands and performs the command fusion. This approach involves studying the intentions of both the human and the autonomous system to ensure the most appropriate fusion of their choices and to evaluate the decision-making of each entity. The intention of the autonomous system is calculated using a visual servoing controller. The implementation of visual servoing relies on a deep learning network detecting lanes. For the human driver, who actively drives and cannot express their intention simultaneously, we use a deep learning-based model to predict their intention. The construction of this model required the creation of a driving dataset using our vehicles and the development of a recurrent model that integrates data of various types. Each of these intentions is then evaluated according to specific criteria, including safety, comfort, and context, to guide the fusion process towards the selection of the highest quality intention. This quantification is based on a state analysis derived from the realization of these intentions. We then use game theory to facilitate the fusion process, where each entity, human and autonomous system, aims to steer the final command towards their choice.Ma thèse porte sur la navigation partagée entre le système autonome et l'humain. Dans notre recherche, nous mettons l'accent sur la fusion des commandes. Dans notre approche, les deux entités, l'humain et le système autonome, pilotent simultanément le véhicule, et un module acquiert leurs commandes et effectue la fusion des commandes. Cette approche implique l'étude des intentions de l'humain et du système autonome afin d'assurer la fusion la plus appropriée de leurs choix et d'évaluer la prise de décision de chaque entité. L'intention du système autonome est calculée à l'aide d'un contrôleur d’asservissement visuel. La mise en œuvre de l'asservissement visuel repose sur un réseau de deep learning capable de détecter les voies de circulation. Pour le conducteur humain, qui conduit activement et ne peut exprimer son intention en même temps, nous utilisons un modèle basé sur le deep learning pour prédire son intention. La construction de ce modèle a nécessité la création d'un ensemble de données de conduite à l'aide de nos véhicules et le développement d'un modèle récurrent qui intègre des données de divers types. Chacune de ces intentions est ensuite évaluée selon des critères spécifiques, y compris la sécurité, le confort et le contexte, dans le but de guider le processus de fusion vers la sélection de l'intention de la plus haute qualité. Cette quantification est basée sur une analyse de l'état dérivée de la réalisation de ces intentions. Ensuite, nous utilisons la théorie des jeux pour faciliter le processus de fusion, au chaque entité, humain et système autonome, souhaite orienter la commande finale vers son choix
Exploring Growth Plate Morphological Evolution: A Computational Mechanobiological Perspective
No full textNational audienceThis research primarily investigates the relationship between trabecular bone patterns within the vicinity of the growth plate and associated morphological changes, specifically regarding the shape modifications and adaptations to mechanical stimuli. To address this, a finite element model was constructed and compared using medical images. The model was tested first with a generic growth model case and subsequently extended to a clinical scenario involving pathologies such as hip dysplasia
Finite element models for laminated timber structures using solid, solid-beam and solid-shell approaches
No full textInternational audienceIn the context of timber construction in particular for buildings, Adhesive-Free Cross-Laminated Timber (AFCLT) panels aim to replace glued Cross-Laminated Timber (CLT) panels as an eco-friendly alternative. This study introduces a numerical approach aimed at efficiently assessing the vibration performance of AFCLT panels assembled using wood dowels. Predictive finite element models are developed through the application of the Verification and Validation methodology. Solid-beam and solid-shell approaches based on standard solid elements are developed with first-order or higher-order theories. A modification of the system of algebraic equations, based on the concept of independent and dependent nodes, is developed. Dependent nodes are eliminated, significantly reducing the number of degrees of freedom and floating-point operations. The methodology is evaluated for the calculation of frequencies and mode shapes of AFCLT panel with a mix of solid elements, solid-shell and solid-beam approaches
A design modification of a quadrotor frame based on fused deposition modeling
No full textInternational audienceQuadrotor UAVs have mass variations in their performance and cost, or these are made of materials such as carbon fiber, which implies a higher manufacturing cost. Therefore, an alternative is to resort to manufacturing parts using 3D printing to reduce the mass and improve strength of materials. This work focuses on the structural analysis of a quadrotor UAV previously manufactured using 3D printing technology with polylactic acid (PLA) filament as an alternative to higher cost materials. Buckling problems are presented in the original design of the frame, so its design is reinforced by considerably increasing the mass. The main objective is to reduce the mass without compromising the structural integrity. Experimental tensile tests are performed on PLA specimens to determine their mechanical properties, as well as rotor thrust tests to evaluate the maximum loads on the frame. Three frame alternatives are proposed and designed, with the second proposal being the most successful, with a 41% reduction in mass. The structural simulation shows that this proposal complies with the structural limits of the aircraft, highlighting the feasibility of using PLA and optimizing designs using the finite element technique to reduce mass without compromising strength, representing an economical and efficient option in terms of aeronautical engineering
Quaternion-based observer control for multirotor UAVs, an application to unactuated grasping
No full textInternational audienceA novel approach for aerial drone control in object pickup tasks is presented. The methodology integrates quaternion-observer control to address the challenge of variable mass during object interaction. A specialized non-actuated gripper designed explicitly for aerial drones enhances their ability to grasp objects efficiently. Real-time tests validated the effectiveness and feasibility of the proposed solution. The experiments demonstrated the robustness and adaptability of quaternion-observer control in compensating for variable mass during object pickup tasks. Additionally, the practical utility of the non-conventional gripper design under real-world conditions emphasized its relevance in aerial manipulation scenarios
Trajectory tracking for aerobatics maneuvers in quadrotors vehicle
No full textInternational audienceSome inspection missions for quadrotors require a previous definition of the trajectories before being tracked. The well design of the trajectory can contribute for the success of the mission when aggressive maneuvers are demanded. In this paper, we propose a solution for designing and tracking unconventional trajectories requiring aerobatics maneuvers in quadrotors. The trajectory is based on the Hopf bifurcations and the sliding mode approach to form spiral loops in 3D around a pillar that need to be inspected by a quadrotor. The control architecture is based on the sliding mode methodology adding new parameters for achieving a desired angular velocity around the stable limit cycle solution. The proposed solution is validated in simulations and numerical results illustrate the good performance of the proposed trajectory-tracking control scheme