A Multi-Agent Model Based on Rooted Leadership Structure and General Dynamics and Self-Adaptive Mechanism

Abstract

This paper proposes a first-order multi-agent model that overcomes the limitations of previous models by incorporating rooted leadership structure, general nonlinear dynamics, and self-adaptive feedback mechanisms. The model consists of multiple agents positioned in a multidimensional space and a virtual leader with environmental perception. Each agent follows fixed general nonlinear dynamics, whereas the group’s leadership structure follows an asymmetric fixed-coupling topology. The model is stable if certain mild assumptions about the agents’ dynamics and group coupling topology are satisfied, as proven using the Lyapunov function method. In order to verify the effectiveness of this model in group outdoor motion, simulation results demonstrate that all agents ultimately dynamically form a cohesive swarm and follow the virtual leader’s movement toward the destination, even in cases where the movements of all agents have a certain degree of randomness or in cases that the model scaled up to a much higher number of agents.

Keywords

• Multi-agent
• nonlinear
• rooted leadership
• self-adaptive
• stability