Nonlinear Decentralized Model Predictive Control for Unmanned Vehicles Moving in Formation
Unmanned vehicles operating in formation may perform more complex tasks than vehicles working indi- vidually. In order to control a formation of unmanned vehicles, however, the following main issues must be faced: vehicle motion is usually described by nonlinear models, feasible control actions for each vehicle are constrained, collision between the members of the formation must be avoided while, at the same time, the computational efforts must be kept low due to limitations on the onboard hardware. To solve these problems, a nonlinear decentralized model predictive control algorithm is presented in this paper. The adopted model is based on the nonlinear kinematic equations describing the motion of a body with six degrees of freedom, where each vehicle shares information with its leader only by means of a wireless local area network. Saturation and collision-free constraints are included within the formulation of the optimization problem, while de- centralization allows to distribute the computational efforts amongst all the vehicles of the formation. In order to show the effectiveness of the proposed approach, it has been applied to a formation of quadrotor vehicles. Simulation results prove that the approach presented in this paper is a valid way to solve the problem of controlling a formation of unmanned vehicles, granting at the same time the possibility to deal with constraints and nonlinearity while limiting the computational efforts through decentralization.