Self-Organization of Heterogeneous Self-Propelled Particle Swarms

Hiroki Sayama, Bioengineering Binghamton University, State University of New York,

Self-propelled particle swarm models are computational models of many particles that are capable of autonomous acceleration and local kinetic interaction. Their dynamics have been extensively studied in physics, theoretical biology, and computational science communities because of their useful implications for the understanding of collective behavior of various autonomous agents (e.g., bacteria, fish, birds, pedestrians) as well as their potential of application to practical problem solving. Earlier studies mostly focused on homogeneous swarms, assuming that the same set of kinetic rules uniformly apply to all the particles. Here we extend our scope to heterogeneous swarms in which more than one type of particles can co-exist. Through extensive computer simulations we studied what kind of patterns/motions could emerge out of the mixtures of multiple types of particles, and found that heterogeneous self-propelled particle swarms usually undergo spontaneous mutual segregation, often leading to the formation of multilayer structures. Driven by their own endogenous forces, the aggregates of particles may additionally show more dynamic macroscopic behaviors, including oscillation, rotation, and linear or even chaotic motion. Interactive evolutionary exploration further revealed the possibility of more complex, even biological-looking structures and behaviors when several different types are mixed. These results suggest a novel direction of understanding and engineering collective behavior of physical agents such as distributed robotic systems.