A Study on Autonomous Functional Differentiation of a Group of Agents Inspired by a Cell Differentiation Mechanism of Early Mammalian Embryos

Abstract

The paper describes a function determination method of an agent in a multi-agent system based on a newly proposed cell differentiation model. Challenges in previous studies in multi-agent systems include dealing with agent failures and being highly dependent on wireless communication. Therefore, we tried to solve these challenges by using cell differentiation in early mammalian embryos, because cells in early mammalian embryos recognize their position and differentiate according to their position in highly robust ways. Namely, we aim to apply the mechanism to the autonomous functional determination of a group of agents that should decide their function according to their relative location in the group. For this aim, the paper proposed an Nf2-Amot complex model of the living cell. The Nf2-Amot model was expected to gain more robust differentiation of living cells. The model together with the conventional model proposed by Caluwé applied to the agents expected to distinguish whether they were located inside or outside of the crowd without any wireless communications. The result showed that the proposed method is more robust than that of using the conventional model.