A probabilistic early collision-warning scheme for UAVs in 3D space

Abstract

Unmanned aerial vehicles, bustling autonomously over the sky above our heads, demand a high level of intellectual ability to avoid possible collision. An early warning radar is proposed for such a purpose. The radar uses a far-reaching signal to detect the existence of another UAV maneuvering in vicinity, and then uses a second near-reaching signal to validate the detection. A safe distance is defined as a threshold below which collision will happen and thus a warning should be triggered. Instead of being measured, the distance between the UAVs is assumed to be random under the rationale that the UAVs are highly maneuverable. With the safe distance given, an arbitrarily high hit-rate of warning and an arbitrarily small false-alarm rate can be achieved by adjusting the strengths of probing signals.