Advanced search
Start date
Betweenand


Automatic tuning of a controller's parameters for a quadrotor with unknown model during hover

Full text
Author(s):
Conrado Silva Miranda
Total Authors: 1
Document type: Master's Dissertation
Press: Campinas, SP.
Institution: Universidade Estadual de Campinas (UNICAMP). Faculdade de Engenharia Mecânica
Defense date:
Examining board members:
Janito Vaqueiro Ferreira; Ely Carneiro de Paiva; Marcelo Becker
Advisor: Janito Vaqueiro Ferreira
Abstract

This work's objective is to develop a controller for a quadrotor with unknown parameters. For this system's control, usually advanced controllers that require knowledge of the system's parameters or simple controllers that require manual gain tunning are used. The controller developed has low complexity and is able to adjust its parameters automatically to minimize a cost function during flight, without requiring knowledge of any system's parameter. However, this controller requires the system's states, which aren't available directly. Hence an analysis of the use of filtering methods to estimate these states is conducted, comparing many possible stochastic models. Nonetheless, the filtering requires correctly calibrated sensors, which led to the creation of new algorithms for calibrating the sensors used. The algorithms developed in these three areas represent steps in the direction of creating quadrotors that operate in diverse environments in a robust way (AU)

FAPESP's process: 12/01511-6 - Embedded systems for research with quadrotors: design and development
Grantee:Conrado Silva Miranda
Support Opportunities: Scholarships in Brazil - Master