A Probabilistic Model for Understanding and Comparing Collective Aggregation Mechanisms

Abstract

This paper presents a flexible probabilistic model for the description of aggregation processes in autonomous collective robotics. Two different experiments are considered: one carried out by the authors, the other by Beckers et al. [1] with teams of reactive autonomous robots which differ from a morphological as well as from a control point of view. Rather than simulating robots moving within an environment, the probabilistic model represents the clustering activity as a sequence of probabilistic events during which cluster sizes can be modified depending on simple geometrical considerations and robot control parameters. It is shown that, for both considered robotic platforms, the evolution of the cluster sizes is perfectly described, both qualitatively and quantitatively, by the probabilistic model. By comparing the results at the model level, a better understanding is gained of the influence of the interaction geometry and of the robot control parameters on the collective aggregation dynamics.