Model Predictive Lateral Control of a UAV Formation Using a Virtual Structure Approach

The ability to fly formations of multiple Unmanned Aerial Vehicles (UAVs) is increasingly being considered as a way of enabling a wealth of new applications, from multi observer unambiguous digital elevation mapping to interferometric imaging and receiver localization, to name a few. The main challenges in formation flying are associated with the need for fine coordinated guidance, with fine navigation and control of the absolute and relative positions and orientations of each vehicle. This paper addresses the coordinated lateral control problem of achieving an optimal trade-off between formation keeping accuracy and energy consumption. The control approach uses a semi-autonomous virtual structure approach for formation control, where the absolute station-keeping commands are scheduled from a ground control station and the UAVs autonomously control their two dimensional relative lateral positions and orientations to maintain a virtual structure. The virtual structure is chosen to be a triangular formation of three UAVs, without loss of generality. Two formation control strategies are compared. The first approach uses a nonlinear Lyapunov stabilizing controller. The second one uses feedback linearization laws, to which linear model predictive controllers (MPC) are superposed to optimize the formation keeping autopilots, in the presence of controller constraints. The feedback linearlised MPC controller is easy to implement and outperforms the nonlinear controller in terms of settling time, under the same controller constraints.