Emergent Collective Behavior in Multi-Agent Systems: An Evolutionary Perspective

Abstract

The study of collective behavior involves the analysis of interactions among a set of agents that yield collective outcomes at the level of the group. The behavior is said to be emergent when it cannot be understood simply as the sum of its constituent parts. Further, group-level outcomes can in turn influence individual interactions. The complexity of this interplay makes the study of emergence challenging and exciting. This dissertation is focused on the study of emergent collective behavior from the perspective of evolution. Evolution is a simple yet powerful algorithm, which when acting on interacting entities in a dynamic environment, yields an array of fascinating behavior as manifest in the natural world. Natural collectives display a wide variety of cooperative behavior and have evolved to efficiently manage the inherent tradeoff between robust behavior and adaptability to dynamic environments. These properties have motivated the design of bio-inspired algorithms for sensing and decision-making in robotic collectives. In this work, we study the evolutionary mechanisms for cooperation and tradeoff management in biological collectives, with a focus on four related topics: replicator-mutator dynamics, collective migration, collective pursuit and evasion, and decision-making dynamics in swarms.

The replicator-mutator dynamics define a canonical model from evolutionary theory and have recently been used to study the evolution of language and the behavioral dynamics of social networks. While the analysis of stable equilibria of these dynamics has been a focus in the literature, we prove that certain conditions suffice for the equations to exhibit stable limit cycles. These cycles correspond to oscillations of grammar dominance in language evolution and to oscillations in behavioral preferences in social networks. For the collective migration problem, it is well-established that a small group of leaders can guide a large swarm of followers. It is less clear how presumably self-interested individuals have evolved to take on such divergent roles. We design a network-based evolutionary model to understand the evolution of leadership in migration, with a focus on the role of network topology on the emergent dynamics. Pursuit and evasive behaviors are ubiquitous in biology and are key drivers for collective motion. We use computational simulations and analytical calculations to study a co-evolving pursuit and evasive system, and incorporate the evolved strategies in a cyclic pursuit-evasion collective motion model. The `stop-signaling' inhibitory mechanism has been recently shown to be critical to the decentralized decision-making dynamics in honeybee swarms. We investigate bifurcations in a model of swarm decision-making as a function of the stop-signal and the values of different alternatives, and present a comprehensive analysis of the dynamics of the model.