We consider a type of routing problems common in defence and security, in which we control a fleet
of unmanned aerial vehicles (UAVs) that have to reach one or more target locations without being
detected by an adversary. Detection can be carried out by a variety of sensors (radio receivers, cameras,
personnel, etc) placed by the adversary around the target sites. We model the act of detecting a UAV
from first principles by noting that sensors work by monitoring frequencies in the electromagnetic
spectrum for signals or noise emitted. By this, we are able to provide a flexible and versatile nonlinear
optimisation framework in which the problem is modeled as a novel trajectory optimisation problem
with paths of the UAVs as continuous arcs in an Euclidean space. The flexibility of our approach is
exhibited by the fact that we can easily consider various relevant objectives, among them minimising
the overall probability of detection and maximising the location error that the adversary experiences
when trying to locate our UAVs. Our model is also versatile enough to consider the act of jamming, in
which one or more of our UAVs intentionally send out signals to interfere with the operations of the
adversary’s sensors. Numerical results show the flexibility of our framework, and that we can solve
realistic instances of this problem type.
Article
View Trajectory Optimization of Unmanned Aerial Vehicles in the Electromagnetic Environment