Clone Swarms: Learning to Predict and Control Multi-Robot Systems by Imitation

TL;DR

This paper introduces SwarmNet, a neural network model that predicts and imitates multi-robot swarm behavior with high accuracy, and extends it to handle noisy, uncertain environments in robotics.

Contribution

The paper presents SwarmNet, a novel neural network architecture for centralized prediction and imitation of swarm behaviors, outperforming previous models and adaptable to real-world uncertainties.

Findings

SwarmNet achieves high prediction accuracy on artificial swarm data.

Modifying existing models gradually approaches SwarmNet's performance.

An extension of SwarmNet handles nondeterministic and noisy environments.

Abstract

In this paper, we propose SwarmNet -- a neural network architecture that can learn to predict and imitate the behavior of an observed swarm of agents in a centralized manner. Tested on artificially generated swarm motion data, the network achieves high levels of prediction accuracy and imitation authenticity. We compare our model to previous approaches for modelling interaction systems and show how modifying components of other models gradually approaches the performance of ours. Finally, we also discuss an extension of SwarmNet that can deal with nondeterministic, noisy, and uncertain environments, as often found in robotics applications.