Résumé
Murmuration, i.e. starlings gathering and swirling with extraordinary spatial coherence, is one of the most impressive kind of bird flocking. It is now well accepted that this collective behavior emerges from individual ones and that no global control is involved. In other words, every starling has an equivalent status in the flock and there is no leader deciding how the murmuration evolves. Considering this phenomenon , Reynolds' individual-based rules have been investigated and implemented a number of times to create compelling computer-animated models of the aerial movement of swarm-like flocks of starlings. Reynolds' model is considered as a classic Agent Based Model (ABM) and integrated as a flagship example in many ABM platforms. Still, it turns out that implementing Reynolds' model is not sufficient per se in the sense that all murmuration simulations use tricks to achieve a convincing animation of this phenomenon. Especially, virtual leaders or points of interest are used to orientate the starlings, which somehow contradicts the no-global-control perspective, and thus suggests that murmu-ration dynamics is not yet fully grasped. This paper first highlight this aspect of existing murmuration simulations and then show that it is possible to obtain murmuration-like dynamics by only rethinking how Reynolds' are usually implemented. Especially, the proposed model does not require the existence of a virtual leader nor embed any global aspect. The objective of this article is to show that it is possible to obtain complex coordinated flight dynamics using a very simple ABM and without adding external stimulus nor additional features, that is by only implementing Reynolds's rules thanks to the IRM4S modeling perspective (an Influence Reaction Model for Simulation). So, in this article, we will first focus on the existing implementations of flocking model to list the advantages and limits and then propose our solution based on the IRM4S approach.