Enhanced SLAM for Autonomous Mobile Robots using Unscented Kalman Filter and Neural Network

Abstract

The novel method of mobile robot Simultaneous Localization And Mapping (SLAM), which is implemented by optimized Unscented Kalman Filter (UKF) Via a Radial Basis Function (RBF) for autonomous robot in unknown indoor environment is proposed. For atone the Unscented Kalman Filter based SLAM errors intrinsically caused by its linearization process, the Radial Basis Function Network is composed with Unscented Kalman Filter. A mobile robot localizes itself autonomously and makes a map simultaneously while it is tracking in an unknown environment. The offered approach has some benefits in handling a robotic system with nonlinear movements because of the learning feature of the Radial Basis Function. The simulation results show the powers and effectiveness of the proposed algorithm comparing with a Standard UKF. Keywords: Hybrid Filter, Mobile robot, RBF, SLAM, UKF