A general approach for consensus using optimistic planning

An important challenge in multiagent systems is consensus, in which the agents are required to synchronize certain controlled variables of interest, often using only an incomplete and time-varying communication graph. We propose a consensus approach based on optimistic planning (OP), a predictive control algorithm that finds near-optimal control actions for general dynamics and reward functions (costs). At every step, each agent uses OP to solve a local control problem with rewards that express the consensus objectives. Neighboring agents coordinate by exchanging their predicted behaviors in a predefined order. Due to its generality, OP consensus can adapt to any agent dynamics and, by changing the reward function, to a variety of consensus objectives. While theoretical analysis is still open, OP consensus is demonstrated in experiments for two problems. The first problem is velocity consensus (flocking) with a time-varying communication graph, where OP preserves connectivity better than a classical algorithm. The second problem is the leaderless and leader-based consensus of robotic arms, where OP easily deals with the nonlinear dynamics.