Stochastic models of a cooperative autonomous unmanned aerial vehicle problem

Directing Unmanned Aerial Vehicles (UAVs) to behave in an ‘intelligent’ manner constitutes a very interesting and challenging problem. Such vehicles have steadily increased in importance in military and other applications, where they have several advantages over manned aircraft. This paper presents a flexible model that allows multiple UAVs to cooperatively search for targets in a given environment or area, using a method to efficiently store dynamic target location probability distributions and a dynamic programming implementation. Several key approximations are also given that produce a feasible and effective solution in the presence of constraints on communication and computational power, as demonstrated via comprehensive simulation studies.