Consensus of multi-agent systems: A relative-input-output approach

This paper studies the multi-agent consensus problem of complex networks consisting of general homogeneous continuous linear time-invariant systems with controls under a time-invariant directional communication topology. A new approach based on relative-input-output is proposed to solve the problem. In contrast to a popular method where each agent is equipped with an observer, and hence involving heavy observer dynamics, the proposed approach is static and utilizes relative-input and relative-output eliminating the observers aforementioned, therefore is practically much simpler and easier to implement significantly reducing computational and communicational burdens and costs. Extension of the framework is then made to tackle a dual problem: consensus of distributed filtering without controls. Two numerical examples are provided to validate the proposed approach. © IFAC.