Multi-sensor measurement fusion via adaptive state estimator

Abstract

An adaptive state estimator is developed to fuse the measurements extracted from multiple sensors for tracking the same maneuvering target. The proposed approach consists of Multi-Band Standard Kalman Filter (MBSKF) and a learning processor. Based on Bayesian estimation scheme, the likelihood function of learning processor can be approximated by the Gaussian basis function whose smoothing factor is related to the estimated bandwidth by taking an average of innovation error covariance matrices of MBSKF. Based upon learning processor and MBSKF, adaptive state estimation is extended to handle the switching plant in the multi-sensor environment. The simulation results are presented which demonstrate the effectiveness of the proposed approach. © 2011 Springer-Verlag Berlin Heidelberg.