FastSLAM is a framework for simultaneous localization using a
Rao-Blackwellized particle filter. In FastSLAM, particle filter is used for the
mobile robot pose (position and orientation) estimation, and an Extended Kalman
Filter (EKF) is used for the feature location's estimation. However, FastSLAM
degenerates over time. This degeneracy is due to the fact that a particle set
estimating the pose of the robot loses its diversity. One of the main reasons
for loosing particle diversity in FastSLAM is sample impoverishment. It occurs
when likelihood lies in the tail of the proposal distribution. In this case,
most of particle weights are insignificant. Another problem of FastSLAM relates
to the design of an extended Kalman filter for landmark position's estimation.
The performance of the EKF and the quality of the estimation depends heavily on
correct a priori knowledge of the process and measurement noise covariance
matrices (Q and R) that are in most applications unknown. On the other hand, an
incorrect a priori knowledge of Q and R may seriously degrade the performance
of the Kalman filter. This paper presents a Neuro-Fuzzy Multi Swarm FastSLAM
Framework. In our proposed method, a Neuro-Fuzzy extended kalman filter for
landmark feature estimation, and a particle filter based on particle swarm
optimization are presented to overcome the impoverishment of FastSLAM.
Experimental results demonstrate the effectiveness of the proposed algorithm.