Progressive information filtering fusion for multi-sensor nonlinear systems

Abstract

This paper studies the distributed fusion estimation problem for nonlinear systems. A common method for nonlinear filtering is to linearize the nonlinear state and measurement functions, and the linearization error is a major factor that affects the filtering accuracy. To reduce the impact of linearization error on the distributed fusion, a progressive information filter (PIF) is designed as local estimators by improving the working points of Taylor series expansion, thus the distributed fusion results of linear systems can be extended to nonlinear systems. Moreover, by introducing a termination condition for the progressive measurement update, the designed PIF can avoid over-estimation effectively. Finally, simulations of a target tracking example are presented to demonstrate the advantage and effectiveness of the proposed results.