Neuro-fuzzy learning of locust's marching in a Swarm

Abstract

This study deals with the identification of the behavior of an individual in a group of marching locusts, as observed under laboratory conditions. In particular, the study focuses on the intermittent motion (walking initiation and pausing) of the locusts using Adaptive Neuro-Fuzzy Inference System (ANFIS). Several possible fuzzy rules were examined in a trial-and-error approach, before establishing a reliable set of rules. Analysis of this set led to a consequent reduced fuzzy controller. The results of this study serve as a first step towards achieving the long-term goal of understanding how the behavior of an individual locust translates to the collective swarm movement. As part of achieving this goal, we plan on building a locust-like robot and investigating its behavior within a living swarm of locusts. On a more general level, this study demonstrates, for the first time, that ANFIS can be used to support the understanding of biological systems by translating experimental data into meaningful control laws.