Abstract
Distributed coordination control of complex multi-agent networks has received significant attention over the past decades, due to its widely recognized advantages and potentials in many applications such as large scale sensor networks, robotic networks, power grid, distributed computing clouds, social networks and biological networks. Within the control theory community, the main research task in an engineering multi-agent network system is to design distributed control algorithms, which only use local neighborhood information, to achieve some specified global objectives.
Distributed consensus and synchronization are two important problems in complex engineering multi-agent networks. Consensus is concerned with reaching a network-wide agreement on some quantities of interest while each agent can only access local information. Synchronization defines the correlated-in-time behavior between different agents achieved by local interaction strategies. Synchronization control is also employed to achieve distributed coordinated tracking objectives in multi-agent networks.
A multi-agent network can be quite complex. The amount of agents could be huge, each agent might have complicated dynamics which might possibly be nonlinear and unknown, and the interaction topology among the agents could change over time. Regardless of the control algorithms employed, coordinated control of multi-agent networks relies heavily on interactions among agents, and for this reason the interaction topology plays an essential role in the development of the distributed coordination control algorithms as well as the analysis of their performance. Two major approaches, the common Lyapunov function approach and the multiple Lyapunov function approach, have been
demonstrated to be effective in developing and analyzing control protocols for multi-agent networks with dynamic interaction topologies.
In this thesis, we investigate distributed coordination control problems of complex multi-agent networks with dynamic interaction topologies, which include three main parts summarized as follows.
i) In the first part of this thesis, we consider the distributed synchronization control problem for a multi-agent network with nth order unknown nonlinear agent dynamics. We first establish the results for both an undirected time varying interaction topology and a directed time varying interaction topology. A standard assumption on the interaction topology connectivity in the literature on this type of problems is made, namely, the interaction topology is connected. Distributed synchronization control algorithms are developed and the desired synchronization performance of the multi-agent network system is proven.
ii) In the second part of this thesis, we address the distributed consensus control problem of a multi-agent network with high order linear agent dynamics under a directed jointly connected interaction topology. This enriches the existing literature on multi-agent distributed consensus control under a jointly connected directed interaction topology, which was limited to very simple single integrator agent dynamics. The distributed control protocols are developed and the consensus is proven.
iii) In the final part of this thesis, we investigate the distributed coordinated tracking problem for a multi-agent network under a state dependent jointly connected dynamic interaction topology. In many applications such as in situations where information exchange among agents is conducted via equipped sensors, the interaction topology depends on agents' states such as the distance among agents, and cannot be simply assumed as a function of time. The distributed control protocols are developed and the coordinated tracking is proven. In addition, a new topology connectivity enhancing mechanism is proposed to help ensure coordinated tracking in real world implementations. In comparison with the existing topology connectivity preserving algorithms, our proposed topology connectivity enhancing mechanism is effectivein maintaining both the initially existent topology edges and the newly formed topology edges, which enables us to relax the more restrictive requirement that the interaction topology is initially connected.
