Laboratoire Matière et Systèmes Complexes

Thèse de Raphaël Ponthieu effectuée sous la direction de Pascal Hersen (MSC) et José Halloy (LIED).

Soutenance le jeudi 29 novembre 2018 à 14h00.

Lieu : bâtiment Condorcet, amphithéâtre Pierre-Gilles de Gennes (PGG, niveau -1). La soutenance sera suivie d’un pot au 6e étage.

Les fourmis et les machines : interfacer systèmes vivants et systèmes artificiels

Abstract:

Within the animal kingdom social insects fascinate humans. Especially ants, which are capable of adapting to various environments and taking advantage of their biotopes. Observing practices that we would have thought to be exclusive to humans – such as farming (of mushrooms) or breeding (of aphids) – spurs the will to understand by which mean ants operate. It turns out that ants achieve certain tasks in different ways than humans. For instance, to find the shortest path to a food source, or the best nest to migrate to; the colony can make a collective decision, decentralized and without the need of direct comparisons by individuals. Those mechanisms, once uncovered, have inspired computer scientists to create shortest-path algorithms based on the same principles.

My work has consisted of the elaboration and use of new methods to study ants. Two complementary objectives have been chosen to drive the work: implementation of a biohybrid system and assessing the means of and limits to controlling the behaviour of ants.

The first objective consists of asking ourselves what will happen if we connected an ant colony with a computer. Would new kinds of behaviours emerge? Could ants adapt to the computer? And vice versa? For this I have designed apparatus that permit interaction between those two entities. It involved giving sight to the computer into what the colony was doing, then providing ways for the computer to act on the colony.

To do so, I designed and tried out tools to monitor the activity of colonies as well as modular environments which ants can inhabit. I designed an embedded system that records and analyses activity of ant colonies on a micro-computer. I designed and 3D printed modules of various sizes that can be assembled together to form a structurally varied environment. For the computer to be able to act upon ant behaviour, I chose to endow it with the capability of modifying environmental conditions, at particular temperatures. Once the system was built, I conducted experiments to assess how local changes of temperature affect ant behaviour. This characterisation of behaviour under temperature changes sheds light on the relation of ants with their environment.

This permitted the exploration of questions of control of behaviour of ants by a machine. The use of temperature having proved its capacity to modulate the local occupancy density, I then have been able to show that it was possible to confine an ant in a set location. Finally, I implemented a closed feedback loop system, in which the temperature used to confine the ant is dependent on the ant activity in real time. This last experiment shed light on the complexity of the dynamic relation between ants and their environments and opened new perspectives for future investigations.

This thesis established new approaches applied to ants – namely biohybrid systems and control of living organisms – and brings to light accessible directions for future works. During this research many tools have been designed and documented for use by others.