Distributed semantic mapping for heterogeneous robotic teams

Abstract

In this paper we summarize our current work on distributed semantic mapping within heterogeneous robot teams in large scale unstructured environments. We extract semantic information from sensor readings and use it to perform robust registration of sub-maps from different agents. We further use it to reduce network traffic by excluding detected areas of high uncertainty. For fast development and verification of our approaches, we employ a multi-robot real-time simulation.