Hybrid filter based simultaneous localization and mapping for a mobile robot

Abstract

We propose a hybrid filter based SLAM (Simultaneous Localization and Mapping) for a mobile robot to compensate for the EKF (Extended Kalman Filter) based SLAM error inherently caused by the linearization process. A mobile robot autonomously explores the environment by interpreting the scene, building an appropriate map, and localizing itself relative to this map. A probabilistic approach has dominated the solution to the SLAM problem. This solution is a fundamental requirement for robot navigation. The EKF algorithm with a RBF (Radial Basis Function) has some advantages in handling a robotic system having nonlinear dynamics because of the learning property of neural networks. We modified an already developed Matlab simulation source for the hybrid filter-SLAM for simulation and comparison. The simulation results showed the effectiveness of the proposed algorithms as compared with an EKF-based SLAM. © 2009 Springer Berlin Heidelberg.