Reaching Consensus Among Mobile Agents: A Distributed Protocol for the Detection of Social Situations

TL;DR

This paper introduces a distributed protocol enabling mobile agents to collaboratively detect and understand social situations by sharing perceptions and reaching consensus based on social cues, validated on real and synthetic data.

Contribution

The paper presents a novel distributed protocol for agents to collaboratively identify and analyze social situations using subjective assessments and social cues.

Findings

Protocol successfully detects social situations in real-world datasets.

Agents reach consensus on social situation parameters effectively.

Method outperforms baseline approaches in accuracy and robustness.

Abstract

Physical social encounters are governed by a set of socio-psychological behavioral rules with a high degree of uniform validity. Past research has shown how these rules or the resulting properties of the encounters (e.g. the geometry of interaction) can be used for algorithmic detection of social interaction. In this paper, we present a distributed protocol to gain a common understanding of the existing social situations among agents. Our approach allows a group of agents to combine their subjective assessment of an ongoing social situation. Based on perceived social cues obtained from raw data signals, they reach a consensus about the existence, parameters, and participants of a social situation. We evaluate our protocol using two real-world datasets with social interaction information and additional synthetic data generated by our social-aware mobility model.