Investigating State Covariance Properties During Finite Escape Time in H∞ Filter SLAM

Abstract

This paper deals with the investigation of finite escape time problem in H∞ Filter based localization and mapping. Finite escape time in H∞ Filter has restricted the technique to be applied as the mobile robot cannot determine its location effectively due to inconsistent information. Therefore, an analysis to improved the current H∞ Filter is proposed to investigate the state covariance behavior during mobile robot estimation. Three main factors are being considered in this research namely the initial state covariance, the γ values and the type of noises. This paper also proposed a modified H∞ Filter to reduce the finite escape time problem in the estimation. The analysis and simulation results determine that the modified H∞ Filter has better performance compared to the normal H∞ Filter as well as to Kalman Filter for different γ, initial state covariance and works well in non-gaussian noise environment.