Title :
Square root unscented information filter for multiple model estimation
Author :
Guoliang Liu; Guohui Tian
Author_Institution :
Sch. of Control Sci. &
Abstract :
This paper proposes a new multiple model estimation method which employs the square root version of unscented information filter (SRUIF) in the framework of interacting multiple model (IMM). Comparing with the general UIF, the square root version has better numerical characteristics, such as improved numerical accuracy, double order precision and preservation of the symmetry for the covariance matrix. However, the SRUIF has to been further improved in order to achieve better performance in case the system has multiple models, e.g., maneuvering object tracking. Therefore, we here propose to use the IMM with the SRUIF (IMM-SRUIF) for solving multiple model estimation problem. A classical simulation of the maneuvering object tracking problem is demonstrated, which shows the advantages of the proposed IMM-SRUIF method.
Keywords :
"Estimation","Covariance matrices","Computational modeling","Numerical models","Kalman filters","Object tracking","Sensors"
Conference_Titel :
TENCON 2015 - 2015 IEEE Region 10 Conference
Print_ISBN :
978-1-4799-8639-2
Electronic_ISBN :
2159-3450
DOI :
10.1109/TENCON.2015.7372914