Autonomous obstacle avoidance for fixed-wing unmanned aerial vehicles

Abstract This paper investigates a method for autonomous obstacle avoidance for fixed-wing unmanned aerial vehicles (UAVs), utilising potential fluid flow theory. The obstacle avoidance algorithm needs only compute the instantaneous local potential velocity vector, which is passed to the flight control laws as a direction command. The approach is reactive, and can readily accommodate real-time changes in obstacle information. UAV manoeuvring constraints on turning or pull-up radii, are accounted for by approximating obstacles by bounding rectangles, with wedges added to their front and back to shape the resulting fluid pathlines. It is shown that the resulting potential flow velocity field is completely determined by the obstacle field geometry, allowing one to determine a non-dimensional relationship between obstacle added wedge-length and the corresponding minimum pathline radius of curvature, which can then be readily scaled in on-board implementation. The efficacy of the proposed approach has been demonstrated numerically with an Aerosonde UAV model.