Simulating emergent behavior of autonomous swarm systems using agent-based modeling

This paper demonstrates a computational framework comparing the emergent behavior of two types of swarm drone systems using Agent-Based Modeling (ABM). The two swarm models are a Leader–Follower (L-F) swarm model, a modified version of Wilensky’s “Ant Lines” model, and a Flocking model based on a simplified Reynolds “Boids” model. The objective of the simulated operation is to deliver a user-defined number of drones of each type to a target area of interest. The resulting visualized product shows how complex behaviors emerge as both models navigate an environment populated with randomly placed obstacles. The research shows that the Flocking swarm is most efficient in over 40,000 simulated cases. However, with more obstacles added to the simulation environment, the L-F model improves its success rate in these cases and is best overall for faster task completion. The results show the potential of using ABM as part of a user’s toolkit for resource allocation and scenario-based decision-making.