H_∞ filtering convergence and it's application to SLAM

KF-SLAM(Kalman filter-SLAM) have been used as a popular solution by researchers in many SLAM application. Nevertheless, it shortcomings of assumption for Gaussian noise limited its efficiency and demand researcher to consider better filter and algorithm to achieve a promising result of estimation. In this paper, we proposed one of its family, the H _∞ filter-based SLAM to determine its competency for SLAM problem. Unlike Kalman filter, H_∞ filter able to work in an unknown statistical noise behavior and thus more robust. It rely on a guess that the noise is in bounded energy and does not require a priori knowledge about the system. Therefore, we proposed the H_∞ filter as other available technique to infer the location for both robot and landmarks while simultaneously building the map. From the results of simulation, H _∞ filter produces better outcome than the Kalman filter especially in the linear case estimation. As a result, H_∞ filter may provides another available estimation methods with the capability to ensure and improve estimation for the robotic mapping problem especially in SLAM.