Cirta: An Emergentist Methodology to Design and Evaluate Collective Behaviours in Robots' Colonies

摘要

An emergentist approach is necessary as long as we want to synthesise collective systems producing at the macro-level more than the sum of the micro-level individual capabilities or exhibiting new behaviours. In this paper, we introduce the methodology Cirta to design and evaluate emergent collective behaviours in robots' colonies and which consists (1) in specifying the conditions under which a desired macro-level behaviour is realised by micro-level interactions, (2) in instanciating the generic conditions in concrete interactions and (3) in evaluating the probabilities of occurrences. The different phases of Cirta are illustrated using a collective foraging behaviour. This paper is about the third stage where we use a stochastic model by Markov's chains to characterise the dynamics of each minirobot of the ecological niche and the dynamics of tracks and we prove that the global behaviour corresponds to chains' formation of minirobots connecting the nest to a food source.