Amoeba Like Multi-Cell Robot Control System - Obstacle Avoidance using Distributed Pattern Generator

摘要

In this paper, we presented a distributed multi-cell robot control system that enables autonomous and flexible movement. Flexible deformation and actions are features of living things like amoebae. The remarkable characteristic of multi-cellular amoeba is that it has neither specialized sensation organs nor motile organs, however, as a whole, it can present unified movement toward food coordinatively. The facts have been clarified to show that the self-organization phenomenon of the entrainment of non-linear oscillators is deeply related to such amoeboid collective behavior. With an emphasis on these, we proposed an amoeba-like multi-cell robot control model, and applied it to obstacle avoidance as an example of a spatial searching environment. The proposed model consisted of a new self-organization model named as the Distributed Pattern Generator, in which, each cell unit has an oscillator and the unit's state including sensory input is represented as oscillation field, which enabled homogeneous cell units to integrate information of the whole though entrainment with local interactions. The results of computer experiments showed the effectiveness control system of coordination among cell units.