Multi-agent adaptive consensus of networked systems on directed graphs

This paper considers adaptive consensus problem on directed graphs of multi-agent systems with unknown control directions. The proposed consensus protocol assumes that each agent receives information about delayed values of state and control signals of respective neighbors. In case of unknown sign of the high-frequency gain of agent dynamics, the proposed algorithm employs Nussbaum-like switching function. The algorithm allows for a designer to select low gain or high gain consensus protocols. Assuming that the underlying graph is strongly connected, it is proved that all agent states converge toward the same value, and the inter-agent coupling parameters are convergent functions. Copyright © 2015John Wiley & Sons, Ltd.