Decentralized Unified Position-Attitude Control of Nonlinear UAVs
In this paper, we propose a fully decentralized, nonlinear feedback control law to maneuver multiple nonlinear quadrotor UAVs with interagent collision avoidance and natural deadlock resolution. Most existing work on this problem adopts a cascaded position-attitude control scheme and utilizes linearized dynamics in controller synthesis. In this work, each UAV controls its position and attitude simultaneously in one unified step, with no dynamics linearization involved at any stage. The proposed scheme is based on a generalization of Gauss's principle of least constraint that allows constrained systems of any order and any type and that identifies, differentiates, stabilizes, partitions, and incorporates the active constraints at each time instant. The control actions result from asymptotically stabilizing the active constraints by user-specified natural frequencies and damping ratios according to a generalized constraint stabilization. Two numerical examples are used to demonstrate the effectiveness of the present method, whose performance on collision avoidance and deadlock resolution is sufficiently close to that of a centralized method.
