Modelagem e Controle de um Quadricóptero em Ambiente de Simulação

Abstract

In this work, a control system for attitude and position of a quadcopter is developed. To achieve this, a dynamic model for the vehicle is first elaborated following the Newton-Euler formalism. Next, to increase the simulation's representativeness, models for the sensors typically installed on the vehicle, such as accelerometer, gyroscope, magnetometer, and GPS, are included. Attitude state estimates are obtained by fusing data from these sensors, except for the GPS, which is specifically used to estimate position states after filtering. The control system is designed considering the linearized attitude model and the simplified position model. Both resulting models are discretized and employed in the design. Specifically, the controllers are based on state feedback, with gain matrices adjusted using the Linear Quadratic Regulator (LQR).

Simulation results show that the proposed control system can drive the vehicle to desired positions and follow varied trajectories. However, the results reveal a certain delay in following references, especially in stabilizing the yaw state ψ. It is believed that this issue can be resolved by adjusting the weighting matrices.