Optimized multi-agent bird formation: how much preliminary knowledge would help?

Details

Supervisors

Hendrickx, Julien

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil en mathématiques appliquées, à finalité spécialisée

Abstract

Migratory birds are known for achieving great energy efficiency during long flights. This performance may happen thanks to the formation used by the flock of birds during its flight. The reduction of energy expenditure is very important for birds to survive during long-distance and sustained migration journeys and can therefore be a factor that influence them towards specific flight formations. Even if it seems natural to consider birds selfish, they could also have evolved through time to become cooperative. Results show that line formations cannot emerge purely based on energy-saving mechanisms for both selfish and cooperative birds. Considering only these mechanisms, there is no stable formation for egoistic birds. For cooperative birds, there exists a stable formation configuration but which leads to collision. Avoiding collision with other birds is also an important factor affecting birds' trajectories and their formations, as two birds cannot overlap. Experience gained over time, evolution or social pressure could also lead animals to reach optimal behaviors. Another non-aerodynamical factor of interest could thus be some preknowledge that birds would have acquired through evolution. This is what this master thesis will focus on. Considering that birds have some preknowledge about long migratory flights, they could use it to influence their behavior and thus helps them to reach flight formations. Also, the fact that birds may be subject to receive and process imperfect information will be investigated, as well as the impact of their desire to avoid colliding into each other. In this master thesis, the base model describing the aerodynamics of the considered birds is given. It also explains how their wake benefits are computed and how their positions are updated. After that we argue that birds could have a preferred position relatively to their front neighbor due to some preknowledge. They could thus make a compromise between satisfying their affection for their preferred position and satisfying their desire to optimize their instant wake benefits. We thus want to determine if considering this affection can help birds to reach a flight formation. If it is the case, we also wonder what kinds of compromise the birds have to make. Next, we consider that birds are not infallible and that they may thus not be able to process perfect information received by their sensors. Also, birds are subject to environmental factors which may have effect on their paths. We thus want to know if birds need to process perfect information to reach stable flight formation. After, we take a collision avoidance system into account. As two birds cannot overlap, it is important to make sure that the flight formation they could reach do not imply them colliding into each others. Also, it seems reasonable to consider birds having an aversion for collision and willing to avoid being too close to each others. We therefore determine the impact of this aversion for collision on the behavior of birds trying to attain stable flight formations. Lastly one will find a conclusion summarizing all the main results obtained during this master thesis. Some suggestions for improvements and further work are also given.