Broadcast Based Control of Multi-Agent Systems for Consensus

摘要

This research investigated the consensus problem where there is a centralized decision-maker which conveys action plans to the swarm members with the constraint that the action plan is broadcast to the members of the swarm, as against individual control actions by each agent to (1) Formulate the problem as one in which the central decision-maker has information about the states of the swarm members, (2) Determine a strategy for the decision-maker so that it can broadcast an action plan to all the swarm members at some frequency to bring about a consensus among the swarm members by having them implement the broadcast instruction, (3) Explore new solution for eliminating the known drawbacks(non-achievement of consensus and large computation time) using a linear programming based algorithm approach, (4) Refine this strategy to achieve quick convergence for large swarms, and (5) develop analytical convergence proofs and study the effect of varying number of swarm members and different randomness mechanisms on convergence.