Unscented Guidance for Waypoint Navigation of a Fixed-Wing UAV

Aerodynamic uncertainty can lead to large variations in the dynamics of fixed-wing unmanned aerial vehicles. Thus, waypoint navigation is normally done using a feedback controller. In some environments navigation can be done in the open-loop provided that the control is desensitized to the uncertainty. In this paper, unscented optimal control is used to design such open-loop controls for guiding a UAV between waypoints. The unscented optimal control problem considers the effects of parametric uncertainty and is formulated as a deterministic optimal control problem so that it can be solved computationally. The results show an improvement of 30% in the mean terminal error and a 35% reduction in the standard deviation for a canonical turn maneuver. A second unscented problem that includes an additional control over the thrust provides an improvement of 50% in the mean error and a 60% reduction in the standard deviation. The unscented guidance provides a new framework for waypoint navigation and could prove to be a promising approach for non ideal environments.