Reduced-order distributed fusion with application to object tracking
Vladimir Shin, Vahid Hamdipoor
In this paper, we propose a novel reduced-order track-to-track fusion filter (ROF) for estimating not all state variables, but only those variables that indicate useful information of a target system for control. The ROF algorithm is designed for multisensory continuous-time stochastic systems. Its communication loads and computational complexity are not so complicated due to usage of the reduced-order local Kalman filters. Performance of the ROF and its estimation accuracy using the covariance intersection fusion are demonstrated on a 2D motion model with several GPSs. Comparative analysis of the ROF with the global optimal centralized Kalman filter is presented. Simulation results demonstrate practical effectiveness of the proposed ROF.