Integration of Vision based SLAM and Nonlinear Filter for Simple Mobile Robot Navigation

Abstract

Simultaneous Localization and Mapping (SLAM) is a widely used technique which can build up a map at unknown environment and figure out its position without prior map information. By applying SLAM method, this paper presents simple integrated navigation system for planar mobile robot equipped with vision sensor and encoders. Because of the nonlinearity of the measurement model, it is used a particle filter which has powerful performance under nonlinear and multi-modal conditions. Finally, the DR (dead reckoning) performance is demonstrated by analyzing the variance of estimated position error when the number of particles and feature points are changed.