Основные понятия
This paper presents a class of continuous- and discrete-time homogeneous multi-agent systems for which scalable non-collaborative (fully distributed) linear protocols can be designed.
Аннотация
The paper focuses on identifying a class of continuous- and discrete-time multi-agent systems (MAS) for which a scalable non-collaborative (i.e., scale-free fully distributed) linear protocol design is developed. The authors have identified conditions on agent models that enable the design of scalable linear protocols. These conditions are necessary if the agents are single input and single output.
The key highlights and insights are:
For continuous-time MAS, the authors provide necessary conditions for the solvability of the scale-free state synchronization problem without localized collaborative information exchange. These conditions include the agent model being stabilizable, detectable, neutrally stable, minimum phase, and having a relative degree of 1.
Under these necessary conditions, the authors present a two-step protocol design for continuous-time MAS. The first step designs a pre-compensator to make the agent model left-invertible, and the second step designs a non-collaborative dynamical protocol for the left-invertible agents to achieve state synchronization.
For discrete-time MAS, the authors show that the necessary conditions are stabilizability, detectability, and neutral stability of the agent model. They then provide a scale-free non-collaborative linear protocol design using a stable observer with the CSS architecture.
The proposed protocols are shown to achieve state synchronization for any fixed communication graph and any number of agents, without requiring any knowledge about the communication network.