Communication Mapping for Robot Teams

Abstract

Communication is a fundamental building block in almost every robotics application, acting as the thread that holds together teams and connects humans to robots deployed in the field. However, accurately predicting how well robots will be able to communicate during deployment and incorporating such information into planning algorithms is particularly challenging. In this article, we present the Link Quality Communication Map (LQCM) – a method for mapping the potential for two robots to communicate. Our communication map builds a discretized representation of the environment and uses Expected Transmission Count (ETX), a link quality metric commonly used for data routing in wireless networks, to represent the ability of two robots to communicate in the environment. We also present a method for predicting ETX for pairs of robots. This article lays out the details of building an LQCM, highlights various properties inherent to these maps, discusses how these maps can be used in a variety of robotics applications, and reviews the results and lessons learned from our own deployments of LQCMs in the field. To validate our theoretical results, we generated communication maps for a variety of environments and used them for various robotics applications including multi-robot environmental monitoring and determining regions with guaranteed communication quality to a base station. Our results show an average decrease of 18.2% in data transmission times when compared to the current community standard for representing communication in robot planning and a decrease of over 90% in extreme cases. We also ran extensive experiments on ETX datasets from our field experiments to evaluate the accuracy of our ETX prediction method and evaluated how well the method handles malicious communication jamming.