Improvement of network fragility for multi-robot robustness

Abstract

This paper proposes a novel concept of robustness for a multi-robot network. Although connectivity is a helpful indicator for robustness against failures of the robots, preserving a high connectivity limits the configuration space of the system. To relax this problem, we propose to preserve a large connected component instead of the high connectivity of the entire network, and present articulation node importance and graph fragility as indicators of the robustness. An estimation method for the node importance and a control law using the importance value are also introduced to improve the network robustness.