ESPCI web site
Collective motion & environmental path entropy
We discuss a “bottom up” model for collective motion. This involves moving agents that re-orientate so as to maximise a measure of the entropy of their environment in the future. We discuss why such principles might confer fitness on rather general grounds. We then discuss the dynamics that arise from them that lead to cohesive, co-aligned (ordered) swarms, in spite of the fact that the algorithm does not explicitly encode for co-alignment or cohesion. This is in contrast to “top down” models, in which these features are explicitly engineered into the models. Bottom-up models may therefore be better able to explain why qualitative phenomenon such as cohesion and co-alignment arise. We discuss several features of the dynamical scheme, including a novel order-disorder transition. Finally, we develop heuristics that mimic this motion. These may have applications in the context of self-propelled robotic systems.