Kin Recognition, Similarity, and Group Behavior

Abstract

This paper presents an approach to describing\ngroup behavior using simple local interactions\namong individuals. We propose that for a given\ndomain a set of basic interactions can be defined\nwhich describes a large variety of group behaviors.\nThe methodology we present allows for simplified\nqualitative analysis of group behavior through the\nuse of shared goals, kin recognition, and minimal\ncommunication. W e also demonstrate h o w these\nbasic interactions can be simply combined into\nmore complex compound group behaviors.\nTo validate our approach we implemented an array of basic group behaviors in the domain of spatial interactions among homogeneous agents. We\ndescribe some of the experimental results from two\ndistinct domains: a software environment, and a\ncollection of 20 mobile robots. We also describe\na compound behavior involving a combination of\nthe basic interactions. Finally, we compare the\nperformance of homogeneous groups to those of\ndominance hierarchies on the same set of basic behaviors.