Covariance inflation efficiency in H∞ filter based SLAM

This paper analyzes the performance of H_∞ Filter(HF) based SLAM(Simultaneous Localization and Mapping) by applying the Covariance Inflation method. We show that via Covariance Inflation, the updated state error covariance matrix is restrictively depends on the number of decorrelated landmarks such as if smaller number of landmarks are being decorrelated from other landmarks, then the decorrelated landmarks has smaller covariance than the correlated landmarks. Even more, HF with Covariance Inflation shows better confidence regarding its estimation in comparison with EKF with Covariance Inflation if the initial state covariance and measurement noise are big. The results are evaluated through several simulation analysis and consistently supports our claims.