Abordagens nebulosas no controle de drone quadrotor

Abstract

The interest in research of unmanned aerial vehicles (UAVs) has increased significantly in recent years. The quadcopter is a type of UAV that has several advantages and applications. The dynamic coupling and the highly nonlinear behavior impose a challenging character to the control of these aircrafts. In this project, fuzzy systems are used to obtain non-linear controllers capable of solving trajectory tracking problems. Two approaches were developed: first using fuzzy systems as universal approximators for designing direct adaptive fuzzy controllers, in which a new proposal using a robust and exact differentiator is used to estimate an unknown adaptation law parameter and genetic algorithm to optimize the parameters found with manual tuning; another using tensor product transformation, a numerical technique based on higher order singular value decomposition (HOSVD), to find a fuzzy model that approximates the quadcopter model, and, in possession of it, develop a PDC (parallel distributed compensation) controller with exponential stability.