Vulnerability of UAV Sense and Avoid to Exploitations: Non-Cooperative Trajectory Modifications

摘要

This work shows how sense and avoid algorithms proposed for autonomous Unmanned Aerial Vehicles (UAV) can potentially be exploited to modify their trajectory as desired. In particular, it is shown that a UAV equipped with ADS-B based collision avoidance can be forced to follow any trajectory by at most two malicious (real or hoaxed) UAVs. The well known Velocity Obstacle method has been used as basis for the collision avoidance algorithm, but similar result could be obtained with different methods. This vulnerability shows the need for trustworthy GNC algorithm design for control of the autonomous systems that will crowd the future skies.