A decentralized approach for multi-UAV multitarget tracking and surveillance

In this paper, we consider the problem of cooperative multitarget tracking and surveillance with multiple UAVs in a decentralized network architecture. In sensor fusion, we developed a decentralized particle filtering approach to maintain target location estimates. Sample sets are converted to Gaussian Mixtures using expectation maximization, and communicated between agents using are rquest/response strategy. Upon receiving Gaussian Mixtures from other agents, each UAV fuses them into its own estimates using generalized covariance intersection. In decision making, we decoupled the problem into two abstraction levels. At the higher level, leaders are selected for each subnetwork. Each leader, then, assigns operating regions to other agents within its subnetwork. At the lower level, individual agents plan their own motions to actively track targets within their assigned regions. Our approach is evaluated in a number of simulation runs. The results demonstrate the effectiveness of our decentralized approach.