Towards Self-Organising Structure Formations: A Decentralized Approach

摘要

We investigate collective decision-making and parallel information processing in regular networks of simple mobile agents. We introduce a novel class of locally connected mobile automata networks where each node changes its internal state vector depending on information received from its neighbours. A configuration of the internal states and local sensor information is translated to a propulsive motion of the network's elements. We show how local rules of automaton internal states transitions can govern the evolution of the network toward task-specific topologies. In this paper, we present simple mechanisms which allow a collective of mobile agents to dynamically organise into simple spatial structures.