Semantic-Based Leader-Following Consensus of Multi-Agent Systems

This paper is concerned with the semantic-based consensus problem of leader-following multi-agent systems (MASs) subject to network-induced delays. A novel research framework for MASs is established under the semantic-driven communication mechanism, with a central focus on ‘semantic extraction – semantic transmission – semantic recovery’. Under this framework, the raw data is efficiently compressed, with concurrent modeling of semantic transmission errors arising during the compression process. Additionally, a semantic-based consensus protocol is proposed. In consideration of semantic transmission errors, a leader-following consensus criterion is derived with employing the Lyapunov-Krasovskii (L-K) methodology, ensuring that the closed-loop error system achieves exponential convergence towards a bounded set. An illustrative example is provided to verify the efficiency of the proposed method.