Algorithms for estimation and control of quadrotors with physical interaction with their environnement
In recent years, the field of aerial robotics has been improved, allowing the UAVs to apply a controlled wrench on their environment or on an object while flying. Inspired by the use of contacts in legged robots, in this Thesis we propose the idea of exploiting physical contact with the environment for the purpose of ‘locomotion’ during flight, with the goal of going beyond the common thought that the surrounding environment is a constraint to avoid. These ideas are studied and demonstrated on a novel aerial platform consisting of a quadrotor with a 1-dof arm that realizes maneuvers by leveraging contacts with pivot points.
Additionally, we also study the problem of generating trajectories that are most insensitive to variations in the model parameters. This problem has a general validity for any robot, and it is particularly relevant for UAVs because of the high uncertainty in their inertial parameters and actuation. In order to address these issues, we define and leverage the novel notion of « closed-loop state sensitivity » for generating trajectories that are minimally-sensitive to parameters with high robustness guarantees.
The jury is composed of:
– Nicolas Marchand, Directeur de Recherche au GIPSA-lab, Université de Grenoble Alpes (Rapporteur)
– Pascal Morin, Professeur à l’ISIR, Université Pierre et Marie Curie (Rapporteur)
– Isabelle Fantoni, Directrice de Recherche CNRS, LS2N Nantes (Examinatrice)
– Fabio Morbidi, Maître de Conférences au MIS, Université de Picardie Jules Verne (Examinateur)
– Paolo Robuffo Giordano Directeur de Recherche CNRS, IRISA/Inria Rennes (Directeur)
– Antonio Franchi, Chargé de Recherche CNRS, LAAS, Toulouse (Co-directeur)