Initiating Lane and Band Formation in Heterogeneous Pedestrian Dynamics

Abstract

Self-avoiding agents such as pedestrians or road vehicles can exhibit different types of collective and coordinated dynamics. Prominent examples are stop-and-go waves and lane formation, or nonuniform patterns and ordered structures at bottlenecks and intersections. Non-linear effects, phase transitions, and metastability in the collective dynamics of interacting agents raise interesting theoretical questions. Besides scientific interests, understanding and controlling collective performances from individual interaction rules is fundamental to authorities. In this contribution, we show using a two-species agent-based model that heterogeneity can generically initiate segregation and spontaneous formation of lanes and bands. Two universal heterogeneity mechanisms are identified. In the first one, we attribute statically two different values to the parameters of the two types of agents. We aim here to model static heterogeneous individual characteristics. In the second model, we attribute dynamically two different values for the parameters according to the type of the closest agent in front. In contrast to the first model for which the heterogeneity lies statically in agent characteristics, we aim to model dynamic heterogeneity in the interactions. Simulation results show that self-organized lane and band formations spontaneously occur when the heterogeneity factors are sufficiently high. More precisely, we observe the emergence of longitudinal lanes when the heterogeneity lies in the agents when transversal bands arise if we assume heterogeneity in the interactions. The different organizations of the flow highly influence the system's performance. Lane patterns significantly improve the flow, while band formation acting as gridlocks result in lower performance.