Fully Distributed Adaptive Optimal Coordination of Heterogeneous Multi-Agent Systems

This paper establishes a general framework for solving the fully distributed adaptive optimal coordination (DAOC) problem of heterogeneous linear multi-agent networks over undirected or directed communication networks. The fully DAOC problem aims to design distributed optimization algorithms for a set of heterogeneous linear agents, such that the output of each agent can be driven to the optimal state so as to optimize the whole system's performance without the need of global information in the network. First, two fully DAOC protocols are proposed for undirected communication networks using the output regulation technique and edge/node-based adaptive parameter methods, respectively. Then, in combination with a left eigenvalue estimator, the proposed node-based DAOC protocol is extended to the setting where the communication network is weight-unbalanced and directed. Finally, the effectiveness of the theoretical results is demonstrated by practical applications of optimal coordination in multi-RLC circuit systems.