Multi-Sensor Fusion for A Brain-Inspired SLAM System

Abstract

A brain-inspired simultaneous localization and mapping (SLAM) system, called RatSLAM, that requires information of odometry and visual scenes traveled before, is used to construct a cognitive map for a mobile robot. While existing RatSLAM systems, that use raw odometry, easily suffer from the problem of low accuracy of maps in complex environments. In this paper, we employ a multi-sensor fusion method to provide better odometry for RatSLAM system. Experiment results demonstrate that the proposed system, based on multi-sensor fusion, show significant improvements on the cognitive mapping results. Thus, the proposed system is able to construct more precise cognitive maps.