Aggregation in Multi-agent Systems Under Connectivity Constraints in a Potential-Based System

While multi-agent systems or swarm systems interacting under potential-based systems in an ideal all-to-all communication scenario have been well understood, swarm behavior in constrained communication scenarios is not well explored. For a practical mission, in the absence of all-to-all communication and a centralized command center, agents or swarms need to act on limited neighborhood information. This paper has investigated swarm aggregation under directed and undirected graph-based networks of small-sized neighborhoods. It has been found that even with very low connectivity, ( \(40\%\) for small swarms and \(3\%\) for large swarms), certain connectivity conditions lead to milling behavior, a favorable configuration for aggregation. Stable configurations have been identified for random directed networks even with a graph neighbor of size 2 without a sophisticated connected network generation scheme. In addition, a new algorithm for connectivity-constrained undirected neighborhood generation has been proposed with sufficient simulation results. The new scheme ensures an almost guaranteed participation from each swarm member while satisfying connectivity constraints.