Search by UAVs with Flight Time Constraints using Game Theoretical Models

Abstract

This paper addresses the problem of obtaining search routes for UAVs with ∞ight time constraints performing search operations in an unknown region. The UAVs start the search mission from a base station and have to return to a base station for refuelling. This requirement acts as a constraint on the available paths. Optimal search route decisions for a flnite horizon are obtained using a game theoretical framework depending on the level of cooperation between UAVs. The performance of these game theoretical strategies { noncooperative Nash and cooperative { are evaluated for difierent length look ahead policies. Simulation results show that the non-cooperative strategies perform as well as the cooperative strategy even though they require no communication between UAVs. Unmanned aerials vehicles (UAVs) are widely being considered for search and surveillance operations in remote, inaccessible, and hostile regions. The UAVs are equipped with sensors that collect necessary information as they ∞y over a region. The UAVs need to ∞y many sorties to collect complete and accurate information about the unknown region. In this sense, these UAVs are difierent from search munitions which usually are of single use type and are destroyed at the end of a single mission. The UAVs have limited fuel capacity and have to return to the base for refuelling before they ∞y another sortie. A desirable feature for these UAVs is to have the ability of on-board autonomous decision-making for real-time search route planning. An efiective search route plan is one that takes into account the presence of other UAVs and the amount of uncertainty that various areas in the search region have. This problem becomes complex when more than one UAV is deployed to perform the search operation. The search of an unknown region can be more efiective if the UAVs cooperate with each other. The search decisions have to be over a flnite horizon search space, although this does not ensure optimality of the full sortie. Moreover, the route decisions have to be made at every step over a flnite horizon, taking into account the requirement that the UAV has to return to a base station. The problem of search route planning becomes still more di‐cult when there are multiple base stations and a UAV has the option of selecting one of them for refuelling. The other aspect of decision-making involves the optimality of a search step (movement of a UAV from one cell to another at a given time) based on the uncertainty associated with the surrounding area. If we assume a scenario where the UAVs are tracked by some other airborne system that updates them on the location of the other UAVs, then the decision-making is reduced to taking a most favourable joint decision by all the UAVs provided they communicate among themselves to arrive at this joint decision. However, quite often the UAVs may have very limited communication capability. Under these circumstances, a UAV has to depend on decision-making algorithms that consider only the location of the other UAVs and the uncertainty associated with the search region to arrive at an optimal decision. In this paper, we propose a game theoretical approach to route decision-making that takes into account various levels of communication capabilities possessed by the UAVs while taking the ∞ight time (or refuelling) constraint into account.