Supporting material for the paper:

Evolution of Neuro-Controllers for Robots'
Alignment using Local Communication

by Álvaro Gutiérrez Martín, Elio Tuci and Alexandre Campo
July 2008

Submitted to International Journal of Advanced Robotic Systems

Table of Contents
  1. Polarisation data of Real Experiments
  2. Spatial arrangements of the robots in each post-evaluation test
  3. Real Experiments Movies

Polarisation data of Real Experiments:

We tested the best genotype obtained by artificial evolution in groups of 3, 4, and 6 physical robots across 30 repeated experiments. We use a specific measure of polarisation to calculate the degree of alignment of all the robots. Polarisation P(G) of a group of robots G is defined using the angular nearest neighbour:

where θann(i) is the relative orientation of the angular nearest neighbour of the robot i. If all robots are aligned, then P(G) = 0. Conversely, if headings are evenly distributed, P(G) = 2π. If headings are random, i.e. drawn from a uniform distribution, then P(G) = π in average.

All the data collected on real experiments are in Data Section.

Spatial arrangements of the robots in each post-evaluation test

In this section we show the spatial arrangements of the e-pucks for the 3, 4 and 6 and robot experiments.


Figure 1: Pictures showing the spatial configurations the robots are forming in the (a) three robots test; (b) four robots tests; (c) six robots tests; In these pictures the robots are depicted facing the same direction. However, the reader should bear in mind that, at the beginning of a post-evaluation trial, the robots are facing randomly chosen orientation.

Real Experiments Movies

All movies recorded can be found at Movies Section.