A novel method for extraction of front boundary agents in multi-agent systems

To investigate the question of decision-making individuals in distributed multi-agent systems, this study introduces the concept of front boundary agents based on observations of various collective movements in biological swarms and the notion of boundary sets in topology. A formal definition of the model of front boundary agents is provided. Considering the inherent uncertainty of real-world boundaries, we designed two algorithms to identify the set of front boundary agents: the Upper Convex Hull (UCH) algorithm and the Fast Front Boundary Detection (FFBD) algorithm. To compare the performance of these two algorithms, we conducted simulation experiments in regular and random clusters and randomly selected eight experimental clusters to design a questionnaire survey. The survey responses were subsequently compiled and ranked. Finally, three evaluation metrics-goodness of fit, directed Hausdorff distance, and mean derivation-were employed to assess the consistency between the algorithmic outputs and the survey results. The evaluation demonstrates that the FFBD algorithm satisfies the "majority rule" principle and significantly outperforms the UCH algorithm in terms of both accuracy and stability.