How Swarms Differ: Challenges in Collective Behaviour Comparison

TL;DR

This paper investigates how different feature sets and similarity measures affect the comparison of collective swarm behaviours, highlighting the importance of robustness and proposing a self-organised map approach for better differentiation.

Contribution

It evaluates the robustness of existing feature sets and similarity measures in swarm behaviour comparison and introduces a self-organised map method to identify indistinguishable behaviour regions.

Findings

Certain feature set and similarity measure combinations are more effective in distinguishing behaviors.

The interplay between features and measures significantly impacts behaviour classification.

A self-organised map approach helps locate regions of similar or indistinguishable behaviours.

Abstract

Collective behaviours often need to be expressed through numerical features, e.g., for classification or imitation learning. This problem is often addressed by proposing an ad-hoc feature set for a particular swarm behaviour context, usually without further consideration of the solution's resilience outside of the conceived context. Yet, the development of automatic methods to design swarm behaviours is dependent on the ability to measure quantitatively the similarity of swarm behaviours. Hence, we investigate the impact of feature sets for collective behaviours. We select swarm feature sets and similarity measures from prior swarm robotics works, which mainly considered a narrow behavioural context and assess their robustness. We demonstrate that the interplay of feature set and similarity measure makes some combinations more suitable to distinguish groups of similar behaviours. We also propose a self-organised map-based approach to identify regions of the feature space where behaviours cannot be easily distinguished.