Multi-Agent Reinforcement Learning Control for Consensus Problems of Uncertain Nonlinear Multi-Agent Systems
Hua-yu Zhu, Weijie Mao · 2021 China Automation Congress (CAC) · 2021
In this paper, we consider the consensus problem of distributed multi-agent systems with nonlinear dynamics and disturbances. The underlying system is described based upon the stochastic communication network topology. To solve this problem, we propose a multi-agent reinforcement learning-based method that converts the consensus problem with multiple nonlinear agents into one learning target and automatically learns effective strategies. The actor-critic method is adopted for updating learning policies. Moreover, we design a distributed communication method to ensure that each agent in the multi-agent system can obtain consensus information. Two examples are presented to show the effectiveness and potential of the proposed design technique.