Agile fixed-wing UAV motion planning with knife-edge maneuvers

A modern class of small fixed-wing unmanned aerial vehicles are physically capable of performing exceptional aerobatic maneuvers. This paper presents a methodology for including one of the more functional of these maneuvers, the knife-edge, in motion planning. This is achieved by separating the top-level motion planner from the dynamics and control of the knife-edge maneuver. By coupling feedback laws to feedforward control policies, a control system is developed for conventional trajectory tracking, as well as transitioning into and holding constant altitude and velocity knife-edge flight. A demonstration of how the knife-edge maneuver interacts with a motion planner is provided. The motion planner is based on a modified version of the rapidly-exploring random trees (RRT) algorithm. The algorithm is designed to generate a smooth, collision-free straight-line path that is used to construct a time-dependent reference trajectory for feedback tracking. For demonstration, environments are constructed with passages near as narrow as the aircraft's wingspan, which can more safely be passed through in knife-edge flight. Simulations are conducted to display the control system's ability to track the motion plan through the environments, including during narrow passages where knife- edge is employed.