Simultaneous localization and mapping based on multilevel-EKF

Author

Wang, Hongjian ; Wang, Jing ; Qu, Liping ; Liu, Zhenye

Author_Institution

Dept. of Autom., Harbin Eng. Univ., Harbin, China

fYear

2011

fDate

7-10 Aug. 2011

Firstpage

2254

Lastpage

2258

Abstract

Simultaneous localization and mapping (SLAM) problem is an attractive topic in the mobile vehicle research. It is a navigation algorithm essentially. Extended Kalman Filter (EKF) is the most popular implementation to solve the SLAM problem for its simpleness and effectiveness. But the linearization errors of EKF are inevitable because of many factors like inaccuracy of the system model. Moreover, the estimated precision will be depressed because a moving vehicle will inherently accumulate errors in its position estimate as a result of the noise introduced in the dead-reckoning. In order to reduce the linearization errors and improve the estimated precision, a new SLAM algorithm based on multilevel-EKF is developed. Simulation demonstrated that the accuracy of multilevel-EKF-SLAM is superior to the standard EKF-SLAM. It not only weakened the influence of linearization greatly but also improved the estimated accuracy.

Keywords

Kalman filters; SLAM (robots); linearisation techniques; mobile robots; motion control; SLAM algorithm; extended Kalman filter; linearization error; multilevel-EKF; navigation algorithm; simultaneous localization and mapping; Computational modeling; Jacobian matrices; Noise; Simultaneous localization and mapping; Uncertainty; Vehicles; Dead-reckoning; EKF; Multilevel-EKF; Navigation; SLAM;

fLanguage

English

Publisher

ieee

Conference_Titel

Mechatronics and Automation (ICMA), 2011 International Conference on

Conference_Location

Beijing

ISSN

2152-7431

Print_ISBN

978-1-4244-8113-2

Type

conf

DOI

10.1109/ICMA.2011.5986290

Filename

5986290