pr\u00e9sentent<\/p>\n\n\n\n
Autoris\u00e9e \u00e0 pr\u00e9senter ses travaux en vue de l\u2019obtention du Doctorat de l’Institut Polytechnique de Paris, pr\u00e9par\u00e9 \u00e0 T\u00e9l\u00e9com SudParis en :<\/p>\n\n\n\n
le LUNDI 1 D\u00e9CEMBRE 2025 \u00e0 9h30<\/p>\n\n\n\n
\u00e0<\/p>\n\n\n\n
Amphith\u00e9\u00e2tre Sophie Germain, B\u00e2timent Turing
Centre Inria Saclay 1 Rue Honor\u00e9 d’Estienne d’Orves, 91120 Palaiseau<\/p>\n\n\n\n
Membres du jury :<\/strong><\/p>\n\n\n\n M. Anis LAOUITI<\/strong>, Professeur, T\u00e9l\u00e9com SudParis, FRANCE – CoDirecteur de these R\u00e9sum\u00e9 :<\/strong><\/p>\n\n\n\n UAV ou drones sont des syst\u00e8mes a\u00e9riens qui peuvent \u00eatre pilot\u00e9s \u00e0 distance ou fonctionner de mani\u00e8re enti\u00e8rement autonome. Leur polyvalence permet leur d\u00e9ploiement dans de nombreux secteurs. Dans l’\u00e9levage de pr\u00e9cision, les UAV autonomes peuvent faciliter la surveillance en temps r\u00e9el en collectant des donn\u00e9es sur la sant\u00e9 du b\u00e9tail. Pour collecter ces mesures, les UAV doivent localiser le b\u00e9tail et se d\u00e9placer \u00e0 proximit\u00e9. Cependant, il est difficile de localiser des animaux sur des terrains vastes et encombr\u00e9s, et les m\u00e9thodes de localisation traditionnelles (par exemple, colliers GPS et RFID) sont co\u00fbteuses et consomment beaucoup d’\u00e9nergie. Cette th\u00e8se aborde donc le probl\u00e8me de la recherche de b\u00e9tail par drone (UCS) : concevoir des trajectoires de recherche qui minimisent le temps de recherche pr\u00e9vu (EST) pour localiser le b\u00e9tail dans le cadre de contraintes op\u00e9rationnelles. Sur la base d’une analyse approfondie de la litt\u00e9rature, nous d\u00e9finissons l’UCS comme un probl\u00e8me d’optimisation combinatoire de trajectoire. L’UCS est un probl\u00e8me probabiliste d\u00e9pendant du temps qui pr\u00e9sente des similitudes avec des probl\u00e8mes d’optimisation de trajectoire bien connus tels que le probl\u00e8me du voyageur de commerce (TSP), mais qui diff\u00e8re en termes d’objectif principal. Nous sommes donc les premiers \u00e0 introduire et \u00e0 r\u00e9soudre le probl\u00e8me de l’UCS. Nous r\u00e9solvons ce probl\u00e8me en suivant trois axes principaux: i) Formulations exactes. Tout d’abord, nous d\u00e9veloppons des mod\u00e8les de programmation lin\u00e9aire pour trois instances UCS avec cible stationnaire (UCS-ST), avec cible mobile unique (UCS-SMT) et avec deux cibles mobiles (UCS-TMT). Ces mod\u00e8les fournissent des solutions optimales prouv\u00e9es et constituent des r\u00e9f\u00e9rences. Cependant, leur \u00e9volutivit\u00e9 est limit\u00e9e aux petites instances. ii) Deuxi\u00e8mement, nous \u00e9tudions les approches bas\u00e9es sur l’apprentissage, car elles sont plus adapt\u00e9es aux environnements dynamiques et ne n\u00e9cessitent pas de conception experte. De plus, elles am\u00e9liorent l’\u00e9volutivit\u00e9. Nous concevons un cadre d’apprentissage profond par renforcement \u00e0 l’aide de l’algorithme d’optimisation de politique proximale (PPO) pour UCS-ST, d\u00e9montrant ainsi la viabilit\u00e9 de la recherche bas\u00e9e sur les politiques. iii) Troisi\u00e8mement, nous introduisons un Transformer supervis\u00e9 et uniquement bas\u00e9 sur un encodeur qui traite l’UCS-ST comme une pr\u00e9diction du n\u0153ud suivant et peut g\u00e9n\u00e9rer efficacement des trajectoires \u00e0 partir de parcours optimaux. Globalement, la th\u00e8se contribue aux domaines des formulations exactes et des mod\u00e8les \u00e9volutifs bas\u00e9s sur l’apprentissage pour la recherche de b\u00e9tail par drone, tout en d\u00e9limitant les compromis entre l’optimalit\u00e9 et l’efficacit\u00e9 computationnelle. Les m\u00e9thodes sont \u00e9galement applicables au-del\u00e0 de la surveillance du b\u00e9tail et s’\u00e9tendent \u00e0 d’autres probl\u00e8mes dans de larges classes de CPOP sous incertitude.<\/p>\n\n\n\n Abstract :<\/strong><\/p>\n\n\n\n Unmanned Aerial Vehicles (UAVs, or drones) are aircraft systems that can be remotely piloted or operate fully autonomously. Their versatility enables deployment across many sectors. In precision livestock farming, autonomous UAVs can support real-time monitoring by collecting health data about the cattle. To collect these measures, UAVs need to locate the cattle and travel near them. However, locating animals over vast, cluttered lands is difficult, and traditional localization methods (e.g., GPS collars, RFID, camera traps) are costly and energy-intensive. This dissertation therefore, addresses the UAV Cattle Search (UCS) problem: designing search paths that minimize the Expected Search Time (EST) to localize cattle under operational constraints. Based on an extensive literature review, we frame UCS as a Combinatorial Path Optimization Problem (CPOP). UCS is a probabilistic time-dependent problem that shares similarities with well-known path optimization problems like the Traveling Salesman Problem (TSP), but it differs in terms of the major objective. Thus, we are the first to introduce and solve the UCS problem. We solve this problem following three main axes: i) First, we develop exact Linear Programming formulations for three UCS instances: Stationary Target (UCS-ST), Single Moving Target (UCS-SMT), and Two Moving Targets (UCS-TMT). These models deliver provably optimal solutions and provide benchmarks. However, their scalability is limited to small instances. ii) Second, we investigate learning-based approaches as it is more suitable for dynamic environments and do not require expert design. In addition, it improves scalability. We design a Deep Reinforcement Learning framework using the Proximal Policy Optimization algorithm (PPO) for UCS-ST, demonstrating the viability of policy-based search. iii) Third, we introduce a supervised and encoder-only Transformer framework that treats UCS-ST as next-node prediction and can efficiently generate trajectories from optimal tours. Overall, the thesis contributes to the areas of exact formulations and scalable learning-based models for UAV cattle searching, while delineating the trade-offs in optimality and computational efficiency. The methods are applicable beyond livestock monitoring as well, and extend to other problems within broad classes of CPOP under uncertainty.<\/p>\n","protected":false},"excerpt":{"rendered":" L’Ecole doctorale : Ecole Doctorale de l’Institut Polytechnique de Paris et le Laboratoire de recherche SAMOVAR – Services r\u00e9partis, Architectures, Mod\u00e9lisation, Validation, Administration des R\u00e9seaux pr\u00e9sentent l\u2019AVIS DE SOUTENANCE de Madame Najoua BENALAYA Autoris\u00e9e \u00e0 pr\u00e9senter ses travaux en vue de l\u2019obtention du Doctorat de l’Institut Polytechnique de Paris, pr\u00e9par\u00e9 \u00e0 T\u00e9l\u00e9com SudParis en : […]<\/p>\n","protected":false},"author":4,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ocean_post_layout":"","ocean_both_sidebars_style":"","ocean_both_sidebars_content_width":0,"ocean_both_sidebars_sidebars_width":0,"ocean_sidebar":"","ocean_second_sidebar":"","ocean_disable_margins":"enable","ocean_add_body_class":"","ocean_shortcode_before_top_bar":"","ocean_shortcode_after_top_bar":"","ocean_shortcode_before_header":"","ocean_shortcode_after_header":"","ocean_has_shortcode":"","ocean_shortcode_after_title":"","ocean_shortcode_before_footer_widgets":"","ocean_shortcode_after_footer_widgets":"","ocean_shortcode_before_footer_bottom":"","ocean_shortcode_after_footer_bottom":"","ocean_display_top_bar":"default","ocean_display_header":"default","ocean_header_style":"","ocean_center_header_left_menu":"","ocean_custom_header_template":"","ocean_custom_logo":0,"ocean_custom_retina_logo":0,"ocean_custom_logo_max_width":0,"ocean_custom_logo_tablet_max_width":0,"ocean_custom_logo_mobile_max_width":0,"ocean_custom_logo_max_height":0,"ocean_custom_logo_tablet_max_height":0,"ocean_custom_logo_mobile_max_height":0,"ocean_header_custom_menu":"","ocean_menu_typo_font_family":"","ocean_menu_typo_font_subset":"","ocean_menu_typo_font_size":0,"ocean_menu_typo_font_size_tablet":0,"ocean_menu_typo_font_size_mobile":0,"ocean_menu_typo_font_size_unit":"px","ocean_menu_typo_font_weight":"","ocean_menu_typo_font_weight_tablet":"","ocean_menu_typo_font_weight_mobile":"","ocean_menu_typo_transform":"","ocean_menu_typo_transform_tablet":"","ocean_menu_typo_transform_mobile":"","ocean_menu_typo_line_height":0,"ocean_menu_typo_line_height_tablet":0,"ocean_menu_typo_line_height_mobile":0,"ocean_menu_typo_line_height_unit":"","ocean_menu_typo_spacing":0,"ocean_menu_typo_spacing_tablet":0,"ocean_menu_typo_spacing_mobile":0,"ocean_menu_typo_spacing_unit":"","ocean_menu_link_color":"","ocean_menu_link_color_hover":"","ocean_menu_link_color_active":"","ocean_menu_link_background":"","ocean_menu_link_hover_background":"","ocean_menu_link_active_background":"","ocean_menu_social_links_bg":"","ocean_menu_social_hover_links_bg":"","ocean_menu_social_links_color":"","ocean_menu_social_hover_links_color":"","ocean_disable_title":"default","ocean_disable_heading":"default","ocean_post_title":"","ocean_post_subheading":"","ocean_post_title_style":"","ocean_post_title_background_color":"","ocean_post_title_background":0,"ocean_post_title_bg_image_position":"","ocean_post_title_bg_image_attachment":"","ocean_post_title_bg_image_repeat":"","ocean_post_title_bg_image_size":"","ocean_post_title_height":0,"ocean_post_title_bg_overlay":0.5,"ocean_post_title_bg_overlay_color":"","ocean_disable_breadcrumbs":"default","ocean_breadcrumbs_color":"","ocean_breadcrumbs_separator_color":"","ocean_breadcrumbs_links_color":"","ocean_breadcrumbs_links_hover_color":"","ocean_display_footer_widgets":"default","ocean_display_footer_bottom":"default","ocean_custom_footer_template":"","ocean_post_oembed":"","ocean_post_self_hosted_media":"","ocean_post_video_embed":"","ocean_link_format":"","ocean_link_format_target":"self","ocean_quote_format":"","ocean_quote_format_link":"post","ocean_gallery_link_images":"on","ocean_gallery_id":[],"footnotes":""},"categories":[286,402],"tags":[],"class_list":["post-6951","post","type-post","status-publish","format-standard","hentry","category-fractualites-ennews-fr","category-seminaires-ness-2013-fr","entry"],"_links":{"self":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/6951","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/comments?post=6951"}],"version-history":[{"count":1,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/6951\/revisions"}],"predecessor-version":[{"id":6952,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/6951\/revisions\/6952"}],"wp:attachment":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/media?parent=6951"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/categories?post=6951"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/tags?post=6951"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
M. Omar sami OUBATTI<\/strong>, Ma\u00eetre de conf\u00e9rences, Universit\u00e9 Gustave Eiffel Paris, FRANCE – Rapporteur
M. Amine DHRAIEF<\/strong>, Professeur, Ecole Sup\u00e9rieure d\u2019\u00c9conomie Num\u00e9rique, Tunisie, TUNISIE – Rapporteur
M. Sami SOUIHI<\/strong>, Ma\u00eetre de conf\u00e9rences, Universit\u00e9 Paris-Est Creteil, FRANCE – Examinateur
Mme Leila NASRAOUI<\/strong>, Professeure, Ecole nationale des sciences de l’informatique, Tunisie, TUNISIE – Examinateur
M. Nadjib ACHIR<\/strong>, Ma\u00eetre de conf\u00e9rences, Universit\u00e9 Sorbonne Paris Nord, FRANCE – Examinateur
M. Paul M\u00fcHLETHALER<\/strong>, Directeur de recherche, Inria Saclay, FRANCE – Examinateur
Mme Leila AZOUZ SAIDANE<\/strong>, Professeur \u00e9m\u00e9rite, Ecole nationale des sciences de l’informatique, Tunisie, TUNISIE – CoDirecteur de these<\/p>\n\n\n\n\u00ab planification des trajectoires de recherche par drone pour la surveillance du b\u00e9tail : de la programmation lin\u00e9aire aux approches bas\u00e9es sur l\u2019apprentissage \u00bb<\/h2>\n\n\n\n
pr\u00e9sent\u00e9 par Madame Najoua BENALAYA<\/h2>\n\n\n\n