Abstract
This paper deals with the Simultaneous Localization and Mapping (SLAM) problem via H∞ filter with compensations for intermittent observations. In the SLAM problem, a mobile robot sometimes loses its proper observation because of obstacles between the robot and landmarks or the limit of its sensor range. In order to estimate the landmarks and its own position accurately based on such intermittent observation data, the proposed method detects them by comparing the observation with its estimation and compensates with a diagonal matrix whose components are switched between 0 and 1. This paper also shows the convergence of the state error covariance matrices. In simulation and experimental results, the correctness of derived theorems for the convergence are confirmed. We can show robot's state and environment information can be estimated with the proposed compensation method for intermittent observations.
