Initial configuration dependence in a self-organizing robot

Abstract

In this paper, we describe the dependence of an initial state in a self-organizing robot on an optimal structure configuration, where a "fractum" is used as a basic unit. Each robot operates on a genetic algorithm (GA) by itself, and all of them will produce a desired configuration. However, problems such as a deadlock state can happen depending on the initial configuration. A deadlock state means a state in which no robots can move because each robot moves autonomously. It is proved from simulations that a difference in the initial configuration can affect both the deadlock rate and the number of movements of fracta needed to obtain an optimal structure configuration.