Optimal Intruder Collision Avoidance for UAVs via Waypoint Tracking

Abstract

This paper presents a guidance strategy for UAV navigation to a destination while safely avoiding static and moving obstacles along the way, building on previous work that reduces the problem of collision avoidance to one of waypoint tracking. The strategy developed in this paper introduces a “safety ellipsoid” approach, a more general variation of the “collision cone” introduced in pre- vious work, to assess whether obstacles are critical enough to require a maneuver, and generate temporary waypoints if so. A well-known proportional navigation optimal guidance law is utilized to reach the waypoints with minimum control effort and thereby avoid the obstacles. The strategy is validated via physically accurate, image-in-the-loop simulation, using Microsoft AirSim, of a head-on encounter between a UAV and an intruder aircraft whose position is estimated through a visual tracking system. The simulations demonstrate that this approach may be particularly useful in collision scenarios in which aircraft must meet international “remain well-clear” (RWC) standards to avoid other vehicles in the airspace.