Control And Navigation Of Uavs Formation : Application To Quadrotors

Résumé: A cooperative Multi-Vehicles Systems/Multi-Agent Systems MVS/MASs can be characterized as a group of decisionmaking autonomous agents/vehicles operating together and seeking a common and collective objective (i.e., group behaviors) based on their sensed information and the shared inter-agents information. This thesis investigates the methods and techniques from a control point of view to maximize the overall gained performance of a robots-system when accomplishing cooperatively a given task in terms of stability, robustness, speed and tracking accuracy. The main objective is to design cooperative Time Varying Formation Tracking controllers for a group of networked vehicles, while tracking a formation reference trajectory. In the literature, classical MVS control approaches have been deeply investigated such as, behavioral based method, virtual structure approach and leader-Follower technique. However, these methods have been shown to be unified within the framework of Consensus techniques. Consensus technique (or protocol) refers to a control law designed for a group of agents (featured by a linear dynamics) to reach an agreement in some variable of interest (i.e., States). The consensus problems have been extended to a MVS formation control. In this thesis, we focus into designing distributed TVF tracking controllers for a networked MVS. The main contribution was into ensuring a distributed and a reduced network information exchange among the MVS individuals. In addition, practical assumptions have been relaxed such as the connectivity of the network. The MVS stability has been studied relying on Lyapunov theory. Numerical simulations have been provided to confirm the conjectures of the proposed developments with an application to multi-Quadrotors systems.

Publié dans la revue: