Description du sujet de thèse

In nature, animals - particularly those living in colonies, such as ants - develop complex networks that are optimally adapted to their function. Individuals in these groups communicate directly or indirectly via pheromones. They form an elaborate social network, whose purpose is greater than individual interest. On the other hand, man also builds networks for transport, energy, flows and information exchange. Varying in form, often centered, these networks support flows, in particular the mobility of city dwellers, who adopt individual, solistic behaviors. The aim is to compare these biological vs. anthropic networks and to investigate, in a nature-inspired approach, which structural and functional characteristics of biological networks are beneficial to the deployment of efficient, sustainable, resilient, frugal urban networks, i.e. optimal in terms of flows and mobilities and low emitters of atmospheric pollutants (thanks to flow optimization). More specifically, by observing a biomimetic analogue of ant colonies, we are seeking to resolve the open (and unresolved) question of the factors triggering Braess's paradox, in conjunction with Wardrop's equilibrium (derived from congestion game theory) and graph theory applied to transport geography.
Biomimicry and, more generally, approaches inspired by nature, represent a promising avenue for scientific innovation (Passino & Kevin, 2005; Benyus, 2011). When it comes to movement, we often cite the burdock, a plant whose mode of seed dissemination by zoochory (hooking into animals' coats with highly efficient hooks) led to the invention of the sturdy Velcro tape, or the kingfisher, whose analysis of posture and beak position in mid-flight contributed to substantial improvements in the air penetration of Japan's high-speed train (the Shinkansen). Over the course of evolutionary time, species have adapted to more or less rapid changes in their physical environment, accumulating “skills” to optimize access to resources and, ultimately, their probability of survival.
These skills, when they relate to flows and networks, can be translated into specific reticular forms which, by virtue of their structure, take on optimal accessibility properties. An emblematic scientific example is the Physarum mold, which has been used to virtually reconstitute urban transport networks, linking flow generators (i.e. that of Tokyo, Tero et al., 2007). Another example on which the team proposing this project has worked and published is that of orbital spider webs (Josselin et al., 2015, 2016). Using graph theory and the development of a multi-agent system (MAS), it has been shown that the topological properties of this radio-circular shape have been integrated by the spider, which builds its web and deposits its glue in optimal ways to capture its prey and avoid self-gluing.
In 2013, the left bank of the Seine in Paris was closed. It was expected that the surrounding boulevards would be more congested, but this was not the case. In 2012, Rouen's Pont Mathilde, which crosses the Seine, caught fire following an accident involving an oil truck. Carrying 80,000 vehicles a day, it was closed until summer 2014, when it was rebuilt identically. During this period, some vehicles took other bridges, others bypassed the city, without ever reaching the initial flows; some vehicles had literally “disappeared” (Razemon, 2016). Similar events have been observed in various conurbations around the world (Bazzan & Klügl, 2005): improved urban traffic following the closure of a freeway in the Cheonggyecheon district of Seoul, 42nd Street in New York, worsening congestion on the Via Aurelia following the addition of a new traffic circle in Pisa. These observations are due to Braess' paradox (2005).
This states that adding one or more roads to a network, instead of making traffic flow more smoothly, can actually slow it down. This can happen under certain conditions, but in a rather uncertain way. The effect of traffic randomness on network delays was described in 1952 by Wardrop (1952), based on two principles: (i) the path used is at least as efficient as the others (individual rationality) (ii) average travel time tends towards a minimum (collective rationality; network users cooperate by choosing paths that would ensure optimal use of the network). This research, in mathematics and economics, is linked to Nash equilibrium and congestion game theory (Acemoglu, 2018). Wardrop's equilibrium and Braess's paradox, problems that have been observed but whose ins and outs are still poorly understood, determine the ability of networks to optimally distribute flows. We hypothesize that observing and modeling the behaviors of ants in colonies, deployed in networks of controlled shape and size can help us better understand the causalities of urban congestion due to flow imbalances.
The interdisciplinary team (geography, animal biology, mathematics and computer science) will be working together to solve several problems linked to the occurrence of Braess' paradox in networks. The expected results are as follows:
- a better understanding of the functioning of the biological networks studied, notably via exchanges between individuals in colonies (pheromones) ;
- application and construction of metrics and mathematical models dedicated to reticular structures, highlighting the importance of network topology in explaining flows, within the broader framework of (congestion) game theory;
- the search for global optimality in urban systems, by integrating means of regulation that go beyond individual behavior;
- identification of the extent to which biomimetic analogy can be used to transfer knowledge from biological to anthropogenic urban networks;
- explicit modeling of individual behaviors and regulators in a Multi-Agent System.
The PhD thesis will focus on the following topic, at the heart of the problem: the contribution of biomimicry to the understanding of flow equilibria in urban road networks and the resolution of congestion due to Braess' paradox.
The thesis will be co-directed by at least 2 team members (from different disciplines). The student's profile will be open and interdisciplinary. The question of biomimicry will orient the profile towards one of the 2 disciplines of the CNRS laboratories involved, depending on the training of the student recruited for the thesis:
- quantitative geography: focus on the application of models deduced from ant observation to urban networks, with a strong geomatics and spatial analysis dimension; integration of graph theory and, to a lesser extent, game theory;
- ecology: focus on the biomimetic device and the resulting agent modeling; in-depth ethological analyses; integration of game theory and, to a lesser extent, graph theory.

Contexte de travail

The PhD student will complete a thesis in geography at the Doctoral School 537 of Avignon University and will be assigned to the 7300 ESPACE Joint Research Unit. He/she will benefit from the site's infrastructure and services (IT equipment, on-site catering). The candidate should have a keen interest in biomimetics, be able to observe nature and transfer its functions/capabilities to a transport/mobility problem.
With a Master's degree and experience in a similar field, he/she is strongly open to interdisciplinarity.
He/she will be co-supervised by Didier JOSSELIN (quantitative geography, geomatics - ESPACE Joint Research Unit, CNRS INSHS) and Olivier Blight (ecologists, ant specialist - The Mediterranean Institute of Biodiversity
and Marine and Continental Ecology (IMBE), CNRS INEE), integrated into an interdisciplinary team and working at the interface of the disciplines involved. Exchanges with foreign laboratories working on similar themes are envisaged.
The thesis will be carried out within the framework of the BANJO project (Biomimetic Anthropic Network for Journey Optimisation) funded by the MITI (Mission for Interdisciplinarity) of the CNRS.

Contraintes et risques

Doctoral work involves many hours in front of a computer screen. Some business travel is also expected, for possible field missions or working meetings with project members, or even foreign laboratories. In addition, the person recruited will be involved in the laboratory life of the ESPACE Joint Research Unit, and in particular in scientific events (seminars, conferences, etc.).