Abstract— In this paper, an interactive Kalman filter -IKF-1is proposed to demonstrate the interaction between targets as to2how the behavior of a desired target is affected by the behavior of3its neighbors. The IKF utilizes two types of interactions available4in multi-agent systems, namely, cooperative and competitive.5The IKF is similar to the distributed Kalman filter -DKF- in6terms of architecture, method of representation of equations, and7use of neighborhood weight matrix while IKF appears to be a8general form of DKF. In this method, a network of IKF nodes is9constructed such that each node is associated with every target.10There are edges between nodes for which the corresponding11targets have effect on each other. Time-varying weights are12used to control the interaction information exchanged among13IKF nodes. The method of calculating interaction weights in14the weight matrix plays a key role on the estimation results.15The calculation of optimal IKF gain and evaluations on MOTP,16MOTA, and MSE metrics illustrate the effectiveness of the17proposed filter in vehicle tracking

Keywords