Robust map building for an autonomous robot using low-cost sensors

Abstract

The paper describes an approach for building a map of an indoor environment with a mobile robot, using a combination of odometry and sonar range sensors. Aiming for real-time large scale mapping on low-cost platforms with limited sensory and computational equipment, we discard high-complexity techniques like probabilistic SLAM. Instead, the algorithms presented here involve odometry correction and automatic pose recalibration, enabling us to build a coherent map that can be used for navigation and self-localization. Experimental results from different environments prove the efficiency of our approach.