Distributed Design of a Vision-based Mapping and Localization System for a Mobile Robot

Abstract

In this paper we present a distributed architecture for the S-PTAM stereo SLAM system. This architecture is developed on the ROS framework, separating the localization and mapping threads in two independent ROS nodes. The distributed SLAM system is ideal for mobile robots with low computing power since it allows the onboard location module and the mapping module -which has a higher computing cost- to be executed in a base station, taking maximum advantage of the processing available onboard. For the validation of the system, experiments were carried out in a public dataset. The results obtained show the feasibility of a distributed architecture and its efficient implementation.