On the Stability of Swarm Consensus Under Noisy Control

Representation of a swarm of independent robotic agents under graph-theoretic constructs allows for more formal analysis of convergence properties. We consider the local and global convergence behavior of an N-member swarm of agents in a modified consensus problem wherein the connectivity of agents is governed by probabilistic functions. The addition of a random walk control ensures ε-stability of the swarm consensus. Simulation results are given to show the rate of convergence to consensus and planned experiments are described.