A hypothesis of state covariance decorrelation effects to partial observability SLAM

Abstract

<span>This paper analyze the performance of partial observability in simultaneous localization and mapping(SLAM) problem. The study focuses mainly on the effect of having a decorrelation technique known as Covariance Inflation to the estimation. The matrix inversion will be the main element to be investigated through two conditions with respect to some defined environment namely as unstable partially observable SLAM and partially observable SLAM via matrix norm analysis. For assessment purposes, the Extended Kalman Filter estimation is referred as the estimator to understand how the conditions can influence the results. The simulation results depicted that, the matrix norm is able to determine the efficiency of estimation and is proportional to the uncertainties of the system.</span>

A review of multi-agent mobile robot systems applications

“…The main condition to successful navigation is the ability to localization and building maps of unknown environments used for a mobile robot, which is also called simultaneous localization and mapping (SLAM) [59]. The locomotion system is an essential part of mobile robot design that takes into account not only the medium where the robot moves but also other elements including maneuverability, controllability, topographical conditions, efficiency, and stability [60]- [62]. The navigation framework with both a local path planner and global path planner is shown in Figure 4 [63].…”

“…At large system scales, finding decentralized path planning and coordination solutions is key to efficient system performance [81]. According to the planning scope, path planning in the navigation system of mobile robots can be divided into global and local planning [62], [82] Another kind of multi-agent path finding is known as multi-agent observation planning when the multi-agent path-finding problem is reformulated to provide protection requirements in a way that the routes of the mobiles are arranged simultaneously with an observation plan [83].…”

Improving Covariance Matrix Diagonalization in SLAM of Mobile Robot

“…The computational and memory requirement of algorithm has been reduced through this implementation. A method of adding a pseudo Positive Semi definite (PsD) namely covariance inflation has been implemented to reduce computational cost [6][7][8]. In this method some minor changes on the covariance matrix has been done in order to decorrelate the system; which lead to low the computational cost.…”