Towards the automatic design of automatic methods for the design of robot swarms - results and controllers

Citation Author(s):
Ken
Hasselmann
IRIDIA - Université libre de Bruxelles
Antoine
Ligot
IRIDIA - Université libre de Bruxelles
Mauro
Birattari
IRIDIA - Université libre de Bruxelles
Submitted by:
Ken Hasselmann
Last updated:
Mon, 07/08/2024 - 15:59
DOI:
10.21227/ngd1-mc72
License:
0
0 ratings - Please login to submit your rating.

Abstract 

The files contain the results and control software associated with our research paper.

 

We address the problem of designing a modular method for the automatic design of robot swarms, which involves defining the modules that will be then automatically selected and assembled into an appropriate architecture (e.g., a finite-state machine or a behavior tree). 

 

This data is associated with a paper in which we propose a method based on repertoires of neural networks automatically generated via a quality-diversity evolutionary algorithm.

Instructions: 

The instances of control software generated by Nata, Arlequin, Chocolate, and EvoStick, and the performances obtained both in simulation and on physical robots are available for download: Dataset.zip and Controllers.zip

Funding Agency: 
Fédération Wallonie-Bruxelles, H2020 European Research Council