RFastSLAM : A FastSLAM Algorithm based Rank Kalman Filter

Abstract

Solving Jacobi matrices of nonlinear functions and particle number degeneracy are two key challenges in fast simultaneous localization and mapping (FastSLAM). This paper proposes a new robust FastSLAM algorithm based on the rank Kalman filter called Rank FastSLAM(RFastSLAM). In the framework of Rao-Blackwellized particle filter (RBPF), the proposed distribution function is approximated by rank sampling points, and the estimation results are closer to the true values. The number of particles are smaller than that of FastSLAM by the rank statistics principle. From the simulation results, it can be seen from the simulation results that the RFastSLAM effectively slows down the particle degradation phenomenon and improves the estimation accuracy of the mobile robot.