A decentralized game theoretic approach for team formation and task assignment by autonomous unmanned aerial vehicles

A number of autonomous unmanned aerial vehicles (UAV) are tasked to search an unknown/uncertain environment, and neutralize targets whenever they come across such threats. Due to resource constraints of the UAVs, neutralizing some threats require simultaneous effort of a certain number of UAVs in minimum. When a UAV detects multiple such targets which require team efforts, it needs to communicate with its neighbour UAVs asking for their cooperation to form colaitions to neutralize the targets. Since the involved entities are dynamic in nature, a UAV may have to terminate information exchange with its neighbours within a few iterative steps so that the decision making process can be finished before any topological change takes place. Moreover, due to sensor range and communication bandwidth limitations, a UAV cannot know about other tasks available to its neighbours in an instantaneous manner. Through communication, UAVs can learn correspondence between their own utilities and utilities of the neighbour UAVs to different joint action profiles, but not the decisions of other UAVs. We formulate the problem that a UAV faces when it detects multiple such targets and needs to decide as to which target it should try to neutralize, given the uncertainty over the decisions of its neighbour UAVs, in a game theoretic framework. We propose a correlated equilibrium concept based decentralized game theoretic solution that requires local information of the UAVs.