Modeling navigation by weaver ants in an unfamiliar, featureless environment.

Abstract

The motion of individual weaver ants from Oecophylla smaragdina was tracked within a small arena to study the algorithm that these ants use for navigation. The arena, a floor tile, was either clean or partly covered by a mild chemical repellent. Statistical properties of the observed motion of the ant can be described by a model that is analogous to the Langevin theory of the motion of Brownian particles. With each time step, the velocity of the ant changes by a random vector with a robust probability distribution. When the average ant encounters the chemical repellent it responds, like a particle seeing a potential energy barrier, by initially slowing before recovering towards its equilibrium state of motion. The model accounts for most qualitative properties of motion with a small number of parameters.