UAVs formation guidance using nonlinear model predictive control

Abstract

In this thesis, Nonlinear Model Predictive Control (NMPC) for formation guidance and control of a UAV is addressed. There are some problems that degrade performance of the NMPC for formation. Therefore, the modification of proposed NMPC is addressed to overcome the problems. An NMPC algorithm predicts the behavior of the system including possibility of collision over some horizon firstly. Next, NMPC generates optimal control input for the horizon in consideration of collision avoidance using gradient-descent method. This procedure is repeated in every step. Since the whole system performance is considered in the cost of NMPC, trajectory optimization for path planning can be avoided. In addition, the proposed method handles easily input saturation and state constraints. Computation time of this approach is faster and computation load is smaller than some other optimal control due to these reasons. Thus, proposed method in this thesis can be applied to UAVs formation in real time.