Conflicts in Collective Robotics

Abstract

The purpose of this chapter is twofold. First, we briefly discuss the concept of emergence in autonomy-based multi-agent systems and its relation to conflict. And second, in what constitutes the core of the chapter, we thoroughly examine an example of autonomy-based multi-agent system through a simulation of a particular application of object regrouping in the framework of collective robotics. In this example, conflict is present, and nothing is explicitly undertaken or planned ahead to cope with; furthermore, there is no form of direct communication among agents. Nevertheless, the system’s performance on this particular task scales good up to a moderate number of agents, even supralinearly in some situations. Moreover, the results suggest that this good performance scaling relates to the fact that there is in the system a type of conflict with a strong influence on performance, which remains nearly constant or even decreases as the number of agents is moderately increased; in a way, and for a moderate number of agents, conflict could be said to be implicitly “managed” in the system. Further work and reflection on this point might lead to a systematic study of implicit ways of reducing conflict.