Improved FastSLAM based on the particle fission for mobile robots

Abstract

In this paper, we propose an improved method of FastSLAM for mobile robots. This improved method gives a solution to the sample impoverishment of ordinary particle filter by preprocessing before the resample. Experimental results show that the improved algorithm improves accuracy and real-time of robot localization and mapping compared to the EKF-SLAM as well as the regular FastSLAM and it is of better robustness. It provides a new approach to robot SLAM in out-door unknown environment.