Trajectory tracking control of quadrotor UAV with LPV approach MPC controller

Abstract

For the last couple of decades, quadrotor control is one of the trending research topics since it has wide areas in use. On the other hand, quadrotors are relatively complex systems since they have six degrees of freedom and only four inputs. Three inputs are pitch, roll, and yaw angle, and one input is the thrust force of the motors. From that point, two control loops have been used for most research about quadrotor control. Also in this research trajectory tracking control of the quadrotor is obtained by two control loops. The position control has been obtained by feedback control, which is responsible for reaching the desired x,y, and z coordinates and speed of the quadrotor from each direction relative to the earth frame. The MPC (Model Predictive Control) achieves attitude control with LPV (Linear Parameter Varying) approach. LPV method is obtained by collecting all of the nonlinear parameters on the state-space equations of the quadrotor to the A matrix. Since MPC control works on the discrete-time from each time sample, we can assume all the variables on the matrix A as constants, and the control algorithm can be handled as a linear system. But the quadrotor model remains nonlinear. For testing this approach on the quadrotor MATLAB model is designed and tested on different trajectories. From each trajectory, the quadrotor catches the trajectory and follows it smoothly. From this research perspective, the LPV-MPC method has good results on trajectory tracking but it needs to be tested on the real-time system to observe CPU usage on the quadrotor.