Friday July 3^rd, 2026 - 4pm - Auditorium BARB91 - Place Sainte-Barbe, 1 - 1348 Louvain-la-Neuve

Reciprocal structures (RS) are often likened to a game of pick-up sticks because of their lack of a clear hierarchy. An RS consists of one or more modules, each comprising at least three beams, enabling an infinite number of associations and, thus, geometric freedom. It combines architectural appeal with the ability to span long distances using relatively short structural elements, which is particularly relevant when built with logs, as they are naturally limited in length. However, wider adoption remains limited due to uncertainties about their structural efficiency, geometric stability, and the numerical predictability of the connections. The objective of this thesis is to establish a reliable design framework for reciprocal structures by addressing these barriers.

The structural efficiency of RS is demonstrated by comparing RS log floors with conventional timber floor patterns. The research shows that a reciprocal pattern yields a material-efficient, thin floor solution for square layouts, sometimes outperforming conventional patterns.

Geometric stability is then examined using existing connection systems and stability assessment methods. A universal approach is identified to detect unstable configurations at the early design stage, thereby enabling consistent preliminary design rules.

Finally, modelling reliability is addressed through a novel spherical hinge connection for spatial RS. Experimental tests on large-scale prototypes show strong agreement with numerical predictions, confirming that accurate modelling is possible when the connection behaviour is properly represented.

Jury members

Prof. Luca Sgambi (UCLouvain), Supervisor
Prof. Pierre Latteur (UCLouvain), Supervisor
Prof. Denis Zastavni (UCLouvain), Chairperson
Prof. Joao Almeida (UCLouvain), Secretary
Prof. Elsa Garavaglia (Politecnico di Milano, Italy)
Prof. Cyril Douthe (Ecole nationale des ponts et chaussées, France)

Pay attention : the public defense of Martin Steinmetz will also take place in the form of a videoconference

Monday June 29^th, 2026 - 5pm - Auditorium BARB93 - Place Sainte-Barbe 1, 1348 Louvain-la-Neuve

With the increasing ubiquity of safety-critical autonomous systems operating in uncertain environments, there is a need for methods that provide behavioural guarantees in the presence of epistemic uncertainty, i.e. lack of knowledge about the system, and potentially aleatoric uncertainty, i.e. intrinsic uncertainty of the system. Toward formally verifying properties of complex systems, methods based on system approximations - abstractions - have surged in recent years.

This thesis studies memory-dependent abstraction-based methods for both deterministic and stochastic systems (modelling aleatoric uncertainty) where only partial observations are available (modelling epistemic uncertainty). Instead of representing the system through memoryless discrete states, the proposed approach builds abstractions whose states encode observation histories. For deterministic systems, two frameworks are developed: uniform memory, where abstractions encode fixed-length output sequences, and adaptive memory, a more flexible setting in which memory is allocated locally through a refinement procedure. The thesis then extends memory-dependent abstractions to stochastic systems, a technically challenging extension that necessitates different theoretical tools.

Additionally, we introduce the Cantor-Kantorovich distance, a novel metric between labelled Markov chains. This distance arises from the natural topology over infinite symbolic traces, and is useful for our memory refinement procedure. Beyond its direct role in this thesis, the Cantor-Kantorovich distance exhibits strong theoretical properties and holds potential for broader applications, making it a transversal yet central contribution of this thesis. 

Jury members

Prof. Raphaël Jungers (UCLouvain), Supervisor
Prof. Alessandro Abate (Oxford University, UK), Supervisor
Prof. Philippe Lefèvre (UCLouvain), Chairperson
Prof. Gianluca Bianchin (UCLouvain), Secretary 
Prof. Sadegh Soudjani (University of Birmingham, (UK) & Max Planck Institute, Germany)
Prof. Manuel Mazo Jr. (TU Delft, The Netherlands)
Prof. Duarte Antunes (TU Eindhoven, The Netherlands)

Pay attention : the public defense of Adrien Banse will also take place in the form of a videoconference

Friday June 26^th, 2026 - 4pm - Faculté d’architecture, d’ingénierie architecturale, d’urbanisme - Salle AV-02 - Rue Henri Wafelaerts 51, 1060 Saint-Gilles (Bruxelles) 

This doctoral research investigates the expanding self-help urbanisation of environmental reserves in peripheral São Paulo, which this study describes as socio-ecological margins. 

Situated at the intersection of housing exclusion and environmental crisis, this research questions how everyday inhabitation processes in São Paulo’s ecological reserve co-produced and are transformed by contemporary conditions of socio-ecological marginalisation. 

Grounded in four years of immersive fieldwork between 2022 and 2025 in three self-built communities —Comunidade KM47, Comunidade Linha do Trem, and Comunidade Terra de Deus— the research combines ethnographic cartographies, oral narratives, spatial documentation, and documentary filmmaking to build evidence of inhabitation struggles that often face erasure through eviction before being archived.

Building from latin american debates in critical urbanism, social geography and political ecology the thesis unfolds through three analytical lenses: everyday inhabitation practices, processes of (de/re)territorialisation, and grassroots governance. 

Across these lenses, the research develops three interrelated conceptual contributions. Under the lens of everyday inhabitation, the thesis conceptualises a marginalised environmentalism of subsistence as a conflicted form of environmentalism in which low-income families simultaneously protect and yet urbanise the ecologies they depend on, instrumentalising the natural environment as a transitional strategy toward urban citizenship. 

Under the territorial lens, this work discusses the concept of grounded relationalities to describe the fragile yet politically significant affective and symbolic bonds to territory produced through this inhabitation process which are systematically targeted and erased before they can be mobilised as collective resistance.
Lastly, through a governance lens, the study develops greyness as governance to expose the condition of legal indeterminacy and selective enforcement through which the State governs the socio-ecological margins, and which low-income populations and housing movements navigate as both a constraint and a survival resource.

Together, these three lenses expose how contemporary socio-ecological marginalisation operates by systematically targeting what this thesis calls the ecologies of the marginalised: the socio-environmental relations produced through cohabitation between low-income families and their biodiverse environments.
This study argues that material dispossession, legitimacy denial, and governance untranslatability collectively target and erode these ecologies, which are the precise terrain on which marginalisation is produced and through which resistance and more just socio-ecological futures remains possible.

Jury members

Prof. Chiara CAVALIERI (UCLouvain), Supervisor
Prof. Jeroen STEVENS (KULeuven), Supervisor
Prof. Gérald LEDENT (UCLouvain), Chairperson
Prof. Elisabetta CINZIA ROSA (UCLouvain), Secretary
Prof. Luciana NICOLAU FERRARA (Universidade Federal do ABC, Brazil)
Prof. Francisco DE ASSIS COMARU (Universidade Federal do ABC, Brazil)
Prof. Hanna HILBRANDT (University of Zurich, Switzerland)

Pay attention : the public defense of Lucas Lerchs will also take place in the form of a videoconference
 

Vendredi 19 juin 2026 à 14h00 - Auditoire SUD11 - Croix du Sud - 1348 Louvain-la-Neuve -

Recurrent events of severe ocean deoxygenation during the Devonian Period represent major disruptions of the marine environment, yet many mysteries still surround them. To address specific aspects of these events, we use the Earth system model of intermediate complexity cGENIE to investigate the controls on ocean oxygenation through a series of targeted numerical experiments. Our results highlight the dominant role of continental configuration in shaping ocean circulation and modulating the response of oxygenation to other environmental and climatic forcings. We further show that astronomical forcing can strongly influence ocean oxygen through its control on continental nutrient weathering fluxes and biological productivity, producing substantial spatially contrasted responses, and intervals of eccentricity maxima associated with increased anoxia. Finally, model–data comparisons for the Kellwasser events indicate that increased nutrient availability can adequately reproduce the extent and seafloor fraction of anoxia, providing quantitative benchmarks against which different mechanisms and hypotheses can be tested.

Jury members :
Prof. Michel Crucifix (UCLouvain) (Supervisor)
Prof. Anne-Christine Da Silva (ULiège) (Supervisor)
Prof. Qiuzhen Yin (UCLouvain) (President)
Prof. Hugues Goosse (UCLouvain) (Secretary)
Dr. Alexandre Pohl (Biogéosciences Dijon)
Prof. Georg Feulner (Potsdam Institute for Climate Impact Research)
Dr. Pam Vervoort (University of Birmingham)

Pay attention : the public defense of Justin Gérard will also take place in the form of a videoconference
 

Thursday June 18^th, 2026 - 4:00pm - Auditorium BARB93 - Place Sainte-Barbe 1, 1348 Louvain-la-Neuve

Les opérations autonomes de rendez-vous et de proximité nécessitent pour un satellite d’estimer la pose 6D d’un satellite cible à partir d’images monoculaires. Bien que les méthodes d’apprentissage profond aient démontré de bonnes performances, elles reposent sur des données d’entraînement synthétiques générées à partir de modèles CAO et présentent souvent une généralisation limitée aux conditions réelles en orbite. En outre, ces méthodes apprises s’appuient sur des réseaux neuronaux dont le processus de décision demeure opaque, limitant de ce fait la confiance pouvant leur être accordée.

Cette thèse propose de lever ces limitations au travers de Neural Radiance Fields (NeRFs), interprétés comme la fonction inverse de l’estimation de pose via la synthèse de nouvelles vues. Une première contribution consiste en une méthode de reconstruction 3D basée sur les NeRFs permettant d’inférer la géométrie et l’apparence de la cible à partir d’un nombre limité d’images, rendant possible un apprentissage sans modèle CAO. Une deuxième contribution exploite cette représentation pour générer des images géométriquement cohérentes sous des points de vue et des conditions d’apparence variés, réalisant ainsi une augmentation de données et améliorant de ce fait la robustesse de l’estimateur face à des changements d’illumination.  Enfin, une dernière contribution permet de visualiser les caractéristiques 3D du satellite cible exploitées par l’estimateur de pose, apportant un éclairage sur son processus de décision et améliorant son interprétabilité.

Cette thèse étend donc l’estimation de pose à des cibles inconnues, améliore la généralisation hors domaine et renforce la transparence des modèles, en vue de leur déploiement dans des missions spatiales autonomes.

Jury members

Prof. Christophe De Vleeschouwer (UCLouvain), Supervisor
Prof. Renaud Detry (KULeuven), Supervisor
Prof. Laurent Francis (UCLouvain), Chairperson
Prof. Benoît Macq (UCLouvain), Secretary
Prof. Jean Berlamont (KULeuven)
Prof. Dirk Vandepitte (KULeuven)
Prof. Yang Gao (Hong Kong University Of Science And Technology, Hong Kong)
Dr. Jonathan Denies (Aerospacelab)

Pay attention : the public defense of Antoine Legrand will also take place in the form of a videoconference