A Gaussian Particle Filter based Factorised Solution to the Simultaneous Localization and Mapping problem

This paper presents a Gaussian Particle Filter based solution to the Simultaneous Localization and Mapping problem. Conventional SLAM algorithms estimate the map and the vehicle trajectory using either an Extended Kalman Filter (EKF), or a combination of EKF´s and particle filters, both of which have their inherent drawbacks which may result in the state estimate diverging from the true solution over time. In this paper, we will analyze these problems, and propose a solution in the form of the Gaussian Particle Filter based Factorised Solution to the SLAM (GPF-FastSLAM) algorithm. We will formulate the GPF-FastSLAM algorithm, and implement it in a simulated environment. The results obtained will be compared to the results from EKF-SLAM and FastSLAM algorithms. We will then further demonstrate the efficacy of the GPF-SLAM algorithm using data obtained in a high clutter filled marine environment, and compare the resulting estimate with EKF-SLAM and FastSLAM algorithms.