Connectivity-Preserving Flocking of Multiagent Systems via Selecting Critical Neighbors

Preserving global network connectivity for multiagent flocking is quite challenging, particularly in obstacle places where agents have limited perception capabilities. The state-of-the-art method to preserve integrity requires estimating the algebraic connectivity, the process of which may be computationally prohibitive. In this paper, we propose a distributed connectivity-preserving strategy to perform agile and efficient flocking maneuvers in obstacle-rich places based on the idea of maintaining connectivity with critical neighbors (i.e., critical links). Here, the critical neighbors of an agent, disconnections with which cause the global network to be disconnected, are determined by our proposed rules relying on the local geometric topology and hop-count value of neighbors. In the meantime, the mobility constraint is imposed on control commands to guarantee the critical-link connectivity and collision avoidance. Moreover, we employ the mean-shift theory to redesign the flock-cohesion rule so that all agents can reaggregate after crossing obstacles. Theoretical analysis reveals that our proposed algorithm can preserve the global network connectivity if the initial network is connected and the initial hop-count condition is satisfied. Numerical simulations show more than 50% improvement in algorithm efficiency (i.e., time consumption) compared to the algebraic-connectivity-estimation method.