Swarm Intelligence 11th International

Conference ANTS 2018 Rome Italy

October 29 31 2018 Proceedings Marco
Dorigo
Visit to download the full and correct content document:
https://textbookfull.com/product/swarm-intelligence-11th-international-conference-ants
-2018-rome-italy-october-29-31-2018-proceedings-marco-dorigo/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Swarm Intelligence 12th International Conference ANTS

2020 Barcelona Spain October 26 28 2020 Proceedings
Marco Dorigo

https://textbookfull.com/product/swarm-intelligence-12th-
international-conference-ants-2020-barcelona-spain-
october-26-28-2020-proceedings-marco-dorigo/

Swarm Intelligence 10th International Conference ANTS

2016 Brussels Belgium September 7 9 2016 Proceedings
1st Edition Marco Dorigo

https://textbookfull.com/product/swarm-intelligence-10th-
international-conference-ants-2016-brussels-belgium-
september-7-9-2016-proceedings-1st-edition-marco-dorigo/

Cyberspace Safety and Security 10th International

Symposium CSS 2018 Amalfi Italy October 29 31 2018
Proceedings Arcangelo Castiglione

https://textbookfull.com/product/cyberspace-safety-and-
security-10th-international-symposium-css-2018-amalfi-italy-
october-29-31-2018-proceedings-arcangelo-castiglione/

Discovery Science 21st International Conference DS 2018

Limassol Cyprus October 29 31 2018 Proceedings Larisa
Soldatova

https://textbookfull.com/product/discovery-science-21st-
international-conference-ds-2018-limassol-cyprus-
october-29-31-2018-proceedings-larisa-soldatova/
Decision and Game Theory for Security 9th International
Conference GameSec 2018 Seattle WA USA October 29 31
2018 Proceedings Linda Bushnell

https://textbookfull.com/product/decision-and-game-theory-for-
security-9th-international-conference-gamesec-2018-seattle-wa-
usa-october-29-31-2018-proceedings-linda-bushnell/

Foundations of Intelligent Systems 24th International

Symposium ISMIS 2018 Limassol Cyprus October 29 31 2018
Proceedings Michelangelo Ceci

https://textbookfull.com/product/foundations-of-intelligent-
systems-24th-international-symposium-ismis-2018-limassol-cyprus-
october-29-31-2018-proceedings-michelangelo-ceci/

Static Analysis 25th International Symposium SAS 2018

Freiburg Germany August 29 31 2018 Proceedings Andreas
Podelski

https://textbookfull.com/product/static-analysis-25th-
international-symposium-sas-2018-freiburg-germany-
august-29-31-2018-proceedings-andreas-podelski/

Computational Logistics 9th International Conference

ICCL 2018 Vietri sul Mare Italy October 1 3 2018
Proceedings Raffaele Cerulli

https://textbookfull.com/product/computational-logistics-9th-
international-conference-iccl-2018-vietri-sul-mare-italy-
october-1-3-2018-proceedings-raffaele-cerulli/

Similarity Search and Applications 11th International

Conference SISAP 2018 Lima Peru October 7 9 2018
Proceedings Stéphane Marchand-Maillet

https://textbookfull.com/product/similarity-search-and-
applications-11th-international-conference-sisap-2018-lima-peru-
october-7-9-2018-proceedings-stephane-marchand-maillet/
 Marco Dorigo · Mauro Birattari
Christian Blum · Anders L. Christensen
Andreagiovanni Reina · Vito Trianni (Eds.)
LNCS 11172

Swarm Intelligence
11th International Conference, ANTS 2018
Rome, Italy, October 29–31, 2018
Proceedings

123
Lecture Notes in Computer Science 11172
Commenced Publication in 1973
Founding and Former Series Editors:
Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen

Editorial Board
David Hutchison
Lancaster University, Lancaster, UK
Takeo Kanade
Carnegie Mellon University, Pittsburgh, PA, USA
Josef Kittler
University of Surrey, Guildford, UK
Jon M. Kleinberg
Cornell University, Ithaca, NY, USA
Friedemann Mattern
ETH Zurich, Zurich, Switzerland
John C. Mitchell
Stanford University, Stanford, CA, USA
Moni Naor
Weizmann Institute of Science, Rehovot, Israel
C. Pandu Rangan
Indian Institute of Technology Madras, Chennai, India
Bernhard Steffen
TU Dortmund University, Dortmund, Germany
Demetri Terzopoulos
University of California, Los Angeles, CA, USA
Doug Tygar
University of California, Berkeley, CA, USA
Gerhard Weikum
Max Planck Institute for Informatics, Saarbrücken, Germany
More information about this series at http://www.springer.com/series/7407
Marco Dorigo Mauro Birattari
•

Christian Blum Anders L. Christensen

•

Andreagiovanni Reina Vito Trianni (Eds.)

•

Swarm Intelligence
11th International Conference, ANTS 2018
Rome, Italy, October 29–31, 2018
Proceedings

123
Editors
Marco Dorigo Anders L. Christensen
Université Libre de Bruxelles University of Southern Denmark
Brussels Odense
Belgium Denmark
Mauro Birattari Andreagiovanni Reina
Université Libre de Bruxelles University of Sheffield
Brussels Sheffield
Belgium UK
Christian Blum Vito Trianni
Artificial Intelligence Research Institute National Research Council
Bellaterra Rome
Spain Italy

ISSN 0302-9743 ISSN 1611-3349 (electronic)

Lecture Notes in Computer Science
ISBN 978-3-030-00532-0 ISBN 978-3-030-00533-7 (eBook)
https://doi.org/10.1007/978-3-030-00533-7

Library of Congress Control Number: 2018954072

LNCS Sublibrary: SL1 – Theoretical Computer Science and General Issues

© Springer Nature Switzerland AG 2018

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the
material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now
known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are
believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors
give a warranty, express or implied, with respect to the material contained herein or for any errors or
omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
 Preface

These proceedings contain the papers presented at ANTS 2018, the 11th International
Conference on Swarm Intelligence, held at the National Research Council (CNR) in
Rome, Italy, during October 29–31, 2018. The ANTS series started in 1998 with the
First International Workshop on Ant Colony Optimization (ANTS 1998). Since then,
ANTS, which is held bi-annually, has gradually become an international forum for
researchers in the wider field of swarm intelligence. In 2004, this development was
acknowledged by the inclusion of the term “Swarm Intelligence” (next to “Ant Colony
Optimization”) in the conference title. Since 2010, the ANTS conference is officially
devoted to the field of swarm intelligence as a whole, without any bias towards specific
research directions. This is reflected in the title of the conference: “International
Conference on Swarm Intelligence.”
This volume contains the best papers selected out of 69 submissions. Of these, 24
were accepted as full-length papers, while 12 were accepted as short papers. This
corresponds to an overall acceptance rate of 52%. Also included in this volume are 7
extended abstracts.
All the contributions were presented as posters. The full-length papers were also
presented orally in a plenary session. Extended versions of the best papers presented at
the conference will be published in a special issue of the Swarm Intelligence journal.
We would like to take this opportunity to thank the large number of people that were
involved in making this conference a success. We would like to express our gratitude to
the authors who contributed their work, to the members of the International Program
Committee, to the additional referees for their qualified and detailed reviews, and to the
staff at the Institute of Cognitive Sciences and Technologies (ISTC) of the CNR for
helping with organizational matters.
We hope the reader will find this volume useful both as a reference to current
research in swarm intelligence and as a starting point for future work.

July 2018 Marco Dorigo

Mauro Birattari
Christian Blum
Anders L. Christensen
Andreagiovanni Reina
Vito Trianni
 Organization

Organizing Committee
General Chair
Marco Dorigo Université Libre de Bruxelles, Belgium

Vice-General Chair
Mauro Birattari Université Libre de Bruxelles, Belgium

Local Organizing and Publicity Chair

Vito Trianni Italian National Research Council, Italy

Technical Program Chairs

Christian Blum Spanish National Research Council, Spain
Anders L. Christensen University of Southern Denmark, Denmark,
and Instituto Universitário de Lisboa (ISCTE-IUL),
Portugal

Publication Chair
Andreagiovanni Reina University of Sheffield, UK

Paper Submission Chair

Volker Strobel Université Libre de Bruxelles, Belgium

Program Committee
Michael Allwright Université Libre de Bruxelles, Belgium
Prasanna Balaprakash Argonne National Laboratory, USA
Jacob Beal BBN Technologies, USA
Giovanni Beltrame Polytechnique Montréal, Canada
Tim Blackwell Goldsmiths, University of London, UK
Mohammad Reza Bonyadi The University of Adelaide, Australia
Darko Bozhinoski Université Libre de Bruxelles, Belgium
Alexandre Campo Université Libre de Bruxelles, Belgium
Marco Chiarandini University of Southern Denmark, Denmark
Maurice Clerc Independent Consultant on Optimisation, France
Carlos Coello Coello CINVESTAV-IPN, Mexico
Oscar Cordon University of Granada, Spain
Nikolaus Correll University of Colorado Boulder, USA
Guido De Croon Delft University of Technology, The Netherlands
VIII Organization

Gianni Di Caro Carnegie Mellon University, USA

Luca Maria Gambardella Istituto Dalle Molle di Studi sull’Intelligenza
Artificiale, Switzerland
Melvin Gauci Harvard University, USA
Luca Di Gaspero University of Udine, Italy
Haibin Duan Beihang University, China
Andries Engelbrecht University of Pretoria, South Africa
Eliseo Ferrante University of Birmingham, Dubai, UAE
Gianpiero Francesca Toyota Motor Europe, Belgium
José García-Nieto University of Málaga, Spain
Simon Garnier New Jersey Institute of Technology, USA
Jorge Gomes University of Lisbon, Portugal
Morten Goodwin University of Agder, Norway
Roderich Gross University of Sheffield, UK
Frédéric Guinand University of Le Havre, France
Heiko Hamann University of Lübeck, Germany
Julia Handl University of Manchester, UK
J. Michael Herrmann University of Edinburgh, UK
Yara Khaluf Ghent University, Belgium
Xiaodong Li RMIT University, Australia
Simone Ludwig North Dakota State University, USA
Manuel López-Ibáñez University of Manchester, UK
Vittorio Maniezzo University of Bologna, Italy
Alcherio Martinoli Ecole Polytechnique Fédérale de Lausanne,
Switzerland
Massimo Mastrangeli Delft University of Technology, The Netherlands
Nithin Mathews Netcetera, Switzerland
Michalis Mavrovouniotis University of Cyprus, Cyprus
Yi Mei Victoria University of Wellington, New Zealand
Ronaldo Menezes Florida Institute of Technology, USA
Bernd Meyer Monash University, Australia
Martin Middendorf University of Leipzig, Germany
Alan Millard University of York, UK
Nicolas Monmarché University of Tours, France
Roberto Montemanni Istituto Dalle Molle di Studi sull’Intelligenza
Artificiale, Switzerland
Marco Montes de Oca Northeastern University, USA
Sanaz Mostaghim Otto von Guericke University Magdeburg, Germany
Konstantinos Parsopoulos University of Ioannina, Greece
Paola Pellegrini IFSTTAR, France
Carlo Pinciroli Worcester Polytechnic Institute, USA
Lenka Pitonakova University of Bristol, UK
Günther Raidl Vienna University of Technology, Austria
Katya Rodriguez-Vazquez National Autonomous University of Mexico, Mexico
Mike Rubenstein Northwestern University, USA
Erol Sahin Middle East Technical University, Turkey
 Organization IX

Roberto Santana University of the Basque Country, Spain

Thomas Schmickl University of Graz, Austria
Kevin Seppi Brigham Young University, USA
Christine Solnon LIRIS, CNRS, France
Thomas Stützle Université Libre de Bruxelles, Belgium
Dirk Sudholt University of Sheffield, UK
Yasumasa Tamura Tokyo Institute of Technology, Japan
Danesh Tarapore University of Southampton, UK
Guy Theraulaz Paul Sabatier University, France
Dhananjay Thiruvady Monash University, Australia
Jon Timmis University of York, UK
Elio Tuci Middlesex University, UK
Ali Emre Turgut Katholieke Universiteit Leuven, Belgium
Gabriele Valentini Arizona State University, USA
Michael Vrahatis University of Patras, Greece
Justin Werfel Harvard University, USA
Alan Winfield University of the West of England, UK
Masahito Yamamoto Hokkaido University, Japan

Additional Reviewers
Nicolas Cambier Université de Technologie de Compiègne, France
Yue Gu University of Sheffield, UK
Bahar Haghighat Ecole Polytechnique Fédérale de Lausanne,
Switzerland
Matthew Hall University of Sheffield, UK
Marcos Oliveira Leibniz Institute for the Social Sciences, Germany
Anil Ozdemir University of Sheffield, UK
Diego Pinheiro Florida Institute of Technology, USA
Judhi Prasetyo Middlesex University Dubai, UAE
Leonardo Stella University of Sheffield, UK
 Contents

Full Papers

A Study on Force-Based Collaboration in Flying Swarms . . . . . . . . . . . . . . 3

Chiara Gabellieri, Marco Tognon, Lucia Pallottino,
and Antonio Franchi

Automatic Design of Communication-Based Behaviors for Robot Swarms . . . 16

Ken Hasselmann, Frédéric Robert, and Mauro Birattari

Behavior Trees as a Control Architecture in the Automatic Modular

Design of Robot Swarms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Jonas Kuckling, Antoine Ligot, Darko Bozhinoski, and Mauro Birattari

Guidance of Swarms with Agents Having Bearing Only

and Limited Visibility Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
Rotem Manor and Alfred M. Bruckstein

Hybrid Control of Swarms for Resource Selection. . . . . . . . . . . . . . . . . . . . 57

Marco Trabattoni, Gabriele Valentini, and Marco Dorigo

Local Communication Protocols for Learning Complex Swarm Behaviors

with Deep Reinforcement Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Maximilian Hüttenrauch, Adrian Šošić, and Gerhard Neumann

Morphogenesis as a Collective Decision of Agents Competing

for Limited Resource: A Plants Approach . . . . . . . . . . . . . . . . . . . . . . . . . 84
Payam Zahadat, Daniel Nicolas Hofstadler, and Thomas Schmickl

Negative Updating Combined with Opinion Pooling in the

Best-of-n Problem in Swarm Robotics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Chanelle Lee, Jonathan Lawry, and Alan Winfield

On Mimicking the Effects of the Reality Gap with

Simulation-Only Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Antoine Ligot and Mauro Birattari

Optimization of Swarm Behavior Assisted by an Automatic Local

Proof for a Pattern Formation Task . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Mario Coppola and Guido C. H. E. de Croon
XII Contents

Quality-Sensitive Foraging by a Robot Swarm Through Virtual

Pheromone Trails . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Anna Font Llenas, Mohamed S. Talamali, Xu Xu, James A. R. Marshall,
and Andreagiovanni Reina

Search in a Maze-Like Environment with Ant Algorithms: Complexity,

Size and Energy Study. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
Zainab Husain, Dymitr Ruta, Fabrice Saffre, Yousof Al-Hammadi,
and Abdel F. Isakovic

Self-adaptive Quantum Particle Swarm Optimization

for Dynamic Environments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
Gary Pamparà and Andries P. Engelbrecht

Simulating Kilobots Within ARGoS: Models and Experimental Validation . . . 176

Carlo Pinciroli, Mohamed S. Talamali, Andreagiovanni Reina,
James A. R. Marshall, and Vito Trianni

Simulating Multi-robot Construction in ARGoS . . . . . . . . . . . . . . . . . . . . . 188

Michael Allwright, Navneet Bhalla, Carlo Pinciroli, and Marco Dorigo

Stability Analysis of the Multi-objective Multi-guided Particle

Swarm Optimizer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
Christopher W. Cleghorn, Christiaan Scheepers,
and Andries P. Engelbrecht

Swarm Attack: A Self-organized Model to Recover from Malicious

Communication Manipulation in a Swarm of Simple Simulated Agents . . . . . 213
Giuseppe Primiero, Elio Tuci, Jacopo Tagliabue, and Eliseo Ferrante

Task-Agnostic Evolution of Diverse Repertoires of Swarm Behaviours . . . . . 225

Jorge Gomes and Anders Lyhne Christensen

The Best-of-n Problem with Dynamic Site Qualities: Achieving

Adaptability with Stubborn Individuals . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
Judhi Prasetyo, Giulia De Masi, Pallavi Ranjan, and Eliseo Ferrante

The Impact of Interaction Models on the Coherence of Collective

Decision-Making: A Case Study with Simulated Locusts . . . . . . . . . . . . . . . 252
Yara Khaluf, Ilja Rausch, and Pieter Simoens

The Importance of Component-Wise Stochasticity in Particle

Swarm Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
Elre T. Oldewage, Andries P. Engelbrecht,
and Christopher W. Cleghorn
 Contents XIII

The Importance of Information Flow Regulation in Preferentially

Foraging Robot Swarms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Lenka Pitonakova, Richard Crowder, and Seth Bullock

The Role of Largest Connected Components in Collective Motion . . . . . . . . 290

Heiko Hamann

Why the Intelligent Water Drops Cannot Be Considered

as a Novel Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
Christian Leonardo Camacho-Villalón, Marco Dorigo,
and Thomas Stützle

Short Papers

A Cooperative Opposite-Inspired Learning Strategy

for Ant-Based Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
Nicolás Rojas-Morales, María-Cristina Riff, Carlos A. Coello Coello,
and Elizabeth Montero

A Solution for the Team Selection Problem Using ACO . . . . . . . . . . . . . . . 325

Lázaro Lugo, Marilyn Bello, Ann Nowe, and Rafael Bello

Boundary Constraint Handling Techniques for Particle Swarm

Optimization in High Dimensional Problem Spaces . . . . . . . . . . . . . . . . . . . 333
Elre T. Oldewage, Andries P. Engelbrecht,
and Christopher W. Cleghorn

Does the ACOR Algorithm Benefit from the Use of Crossover? . . . . . . . . . . 342
Ashraf M. Abdelbar and Khalid M. Salama

Embodied Evolution of Self-organised Aggregation

by Cultural Propagation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
Nicolas Cambier, Vincent Frémont, Vito Trianni, and Eliseo Ferrante

Experimental Evaluation of ACO for Continuous Domains

to Solve Function Optimization Problems. . . . . . . . . . . . . . . . . . . . . . . . . . 360
Ryouei Takahashi, Yukihiro Nakamura, and Toshihide Ibaraki

Gaussian-Valued Particle Swarm Optimization . . . . . . . . . . . . . . . . . . . . . . 368

Kyle Robert Harrison, Beatrice M. Ombuki-Berman,
and Andries P. Engelbrecht

Individual Activity Level and Mobility Patterns of Ants Within Nest Site . . . 378
Kazutaka Shoji

Learning Based Leadership in Swarm Navigation . . . . . . . . . . . . . . . . . . . . 385

Ovunc Tuzel, Gilberto Marcon dos Santos, Chloë Fleming,
and Julie A. Adams
XIV Contents

Maintaining Diversity in Robot Swarms with Distributed

Embodied Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
Iñaki Fernández Pérez, Amine Boumaza, and François Charpillet

On Steering Swarms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403

Ariel Barel, Rotem Manor, and Alfred M. Bruckstein

Vector Field Benchmark for Collective Search in Unknown

Dynamic Environments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 411
Palina Bartashevich, Welf Knors, and Sanaz Mostaghim

Extended Abstracts

A Honey Bees Mating Optimization Algorithm with Path Relinking

for the Vehicle Routing Problem with Stochastic Demands . . . . . . . . . . . . . 423
Yannis Marinakis and Magdalene Marinaki

Blockchain Technology for Robot Swarms: A Shared Knowledge

and Reputation Management System for Collective Estimation . . . . . . . . . . . 425
Volker Strobel and Marco Dorigo

Declarative Physicomimetics for Tangible Swarm

Application Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 427
Ayberk Özgür, Wafa Johal, Arzu Guneysu Ozgur,
Francesco Mondada, and Pierre Dillenbourg

Influence of Leaders and Predators on Steering a Large-Scale

Robot Swarm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429
John D. Lewis, Himanshi Jain, and Sujit P. Baliyarasimhuni

Movement-Based Localisation for PSO-Inspired Search Behaviour

of Robotic Swarms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 431
Sebastian Mai, Christoph Steup, and Sanaz Mostaghim

Of Bees and Botnets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433

Vijay Sarvepalli

Using Particle Swarms to Build Strategies for Market Timing:

A Comparative Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
Ismail Mohamed and Fernando E. B. Otero

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437

Full Papers
 A Study on Force-Based Collaboration
in Flying Swarms

Chiara Gabellieri2 , Marco Tognon1 , Lucia Pallottino2 ,

and Antonio Franchi1(B)
1
LAAS-CNRS, Université de Toulouse, CNRS, Toulouse, France
{marco.tognon,antonio.franchi}@laas.fr
2
Centro di Ricerca “E. Piaggio”, Dipartimento di Ingegneria dell’Informazione,
Università di Pisa, Pisa, Italy
lucia.pallottino@unipi.it

Abstract. This work investigates collaborative aerial transportation by

swarms of agents based only on implicit information, enabled by the
physical interaction among the agents and the environment. Such a coor-
dinating mechanism in collaborative transportation is a basic skill in
groups of social animals. We consider cable-suspended objects trans-
ported by a swarm of flying robots and we formulate several hypoth-
esis on the behavior of the overall system which are validated thor-
ough numerical study. In particular, we show that a nonzero internal
force reduces to one the number of asymptotically stable equilibria and
that the internal force intensity is directly connected to the conver-
gence rate. As such, the internal force represents the cornerstone of a
communication-less cooperative manipulation paradigm in swarms of fly-
ing robots. We also show how a swarm can achieve a stable transportation
despite the imprecise knowledge of the system parameters.

1 Introduction
Cooperative transportation without explicit communication, but based only on
on the indirect exchange of information through the physical interaction with
the environment is a very important feature for social animals. From a scientific
point of view, the problem has been addressed as Stigmery theory [8]; later, it
has been regarded as the main coordinating mechanism in groups of ants for
object transportation [9] (see Fig. 1 on the left) and has been indeed observed
and studied in [3,5,6,13].
Such skills observed in nature have inspired researchers to transfer them to
swarms of simple robotic agents. In fact, avoiding explicit communication would
reduce hardware and software complexity, and overcome possible communica-
tion failure issues. So far, the interest has been mainly focused on terrestrial
systems [3,9,17,22], where the possibility of decentralized transportation based
on physical interactions has been proved. Instead, in this work, we are interested
in the communication-less aerial transportation of objects by swarms. This is
particularly interesting not only from a scientific point of view, thanks to the
c Springer Nature Switzerland AG 2018
M. Dorigo et al. (Eds.): ANTS 2018, LNCS 11172, pp. 3–15, 2018.
https://doi.org/10.1007/978-3-030-00533-7_1
4 C. Gabellieri et al.

Fig. 1. Left: red ants in cooperative transportation. Right: the tension in spider webs
influences their natural frequency [2].

higher complexity of the problem, but also because a simple, robust and scalable
object aerial manipulation technique could meet the requirements demanded by
many real applications. Aerial transportation can benefit from a larger workspace
and the independence from uneven terrains. However, aerial robots, e.g., multi-
rotors, though agile and low cost, are typically characterized by a limited pay-
load. Hence, a cooperative approach is a very suitable solution. Some examples
of applications can be found in industrial contexts, in agriculture, and in search
and rescue missions to carry necessary equipment or first aid.
Considering the lack of results for aerial systems compared to grounded ones,
our purpose is to start filling this gap. In particular, we investigate the possibil-
ity of a communication-less approach for cooperative aerial manipulation with a
swarm of flying agents. Furthermore, we investigate if and how the load and the
cable physically connecting the robots may play the role of an implicit commu-
nication channel exploiting the forces exchanged. To the best of our knowledge,
this is the first work proposing a bio-inspired algorithm for communication-less
aerial manipulation by flying swarms, going beyond the two-robot-‘only’ sce-
nario, recently considered in [7,20,21].
The algorithm that we propose exploits a leader-follower paradigm where the
leader agent knows where to go and hence steers the object toward the desired
position. On the other hand, the follower agents follow the leader and help to
sustain the weight and manipulate the load, exploiting only the implicit infor-
mation contained in the force received from the load itself. It has been observed
that also groups of army ants Eciton burchelli [5] and Dorylus wilverthi [6] adopt
a distinct caste distribution in transportation groups, in the sense that groups
have a significant tendency to contain only one submajor, i.e., a particular type
of ant. These species of army ants have proved to be very efficient in transporting
objects together. Additionally, in the same works, it has been noticed that it is
usually a single ant, the submajor, that starts the motion of the object, and then
the rest of the group moves accordingly. Such a behavior is actually replicated
in the leader-follower paradigm proposed in this paper.
While animals usually deploy items by directly touching them (direct manip-
ulation), in our framework we have chosen an indirect manipulation technique
 A Study on Force-Based Collaboration in Flying Swarms 5

of the load through cables. This choice is motivated by different reasons and
it has been proved to be a very effective solution for cooperative aerial trans-
portation [10–12,19,23]. First of all, a cable attached to the agent center of mass
allows to minimize the coupling between the rotational dynamics of the agent
itself from the rest of the system dynamics. This is particularly useful for aerial
agents that are underactuated – the most common case – since they need to
change their attitude to be able to apply forces in any direction. Furthermore,
compared to other possible decoupling gripping mechanisms, as the ones in [14],
cables are simpler, low cost, and in general lighter. In this paper we demonstrate
through several numerical simulations that a swarm of N flying agents is capable
of collaborative manipulation skills based only on implicit communication. We
show that a twofold major role is played by the internal force applied to the
transported object. Internal forces are forces applied at the contact points on
the object that stretch or compress it without producing any movement, since
they counterbalance each other. The condition of zero internal force corresponds
to the case in which the agents transport the object while keeping the cables
vertical and applying only a force that compensates for the gravity. Firstly, we
have found that nonzero internal forces allow the swarm to univocally set the
attitude of the commonly transported object, and secondly, that larger internal
forces reduce the convergence time of the overall system to such unique equilib-
rium. This creates an interesting analogy with the role of tension in spider webs,
see Fig. 1-right. The breadth of analysis covers also the thorough investigation
of the leader forces depending on the swarm parameters and the analysis of the
benefits of a saturated nonlinear law for the leader force in order to tradeoff
compliance/safety and transportation accuracy.
The paper structure is the following: Sect. 2 illustrates the dynamic model
of the system. Then, we formulate the hypotheses regarding the properties of
the swarm, supported by the numerical results presented in Sect. 3. A thorough
discussion follows in Sect. 4. Final conclusions and future developments are pre-
sented in Sect. 5.

2 Model

The system is composed by a set of N flying agents attached to a cable suspended

load that must be deployed to a particular configuration. In our framework, each
agent interacts with the environment, hence we aim at a soft response similar
to the behavior of human or of animal during everyday interaction tasks. We
model the commonly transported object (the load) as a rigid body. The attitude
of the load is parametrized by Euler angles yaw, pitch and roll, indicated with
ψ, θ, and φ, respectively. Each agent is attached to the load by means of a cable
by means of which it can transfer forces. The cables are attached to the load at
the points Li , with i = 1, . . . , N , placed on the same plane, denoted by I. The
object center of mass (CoM) is indicated with G.
Each cable is modeled as a linear unidirectional spring, with a dissipative
term that damps its longitudinal oscillations. We assume that each flying agent
6 C. Gabellieri et al.

is endowed with a position controller. If the latter is sufficiently precise, we can

model the closed loop system as a simple double integrator. In this way we can
consider each agent as an actuated point mass capable of exchanging a force
with the external world.
By doing so, the proposed method can be applied to different aerial robots.
If we consider multidirectional-thrust platforms capable of controlling position
and orientation independently (popular in the field of aerial physical interac-
tion) [4,15,16,18], the double integrator is an exact model of the position-
controlled closed loop system. In the case of underactuated unidirectional-thrust
vehicles using a standard position controller, the double integrator is instead a
very good approximation. Furthermore, the time-scale separation between the
translational and rotational dynamics has been exploited in other works on aerial
manipulation like [12,14] where the robots are considered as point masses and
modeled as double integrators. Denoting by the vector pRi ∈ R3 the position
of the i-th agent, by Mi = mAi I3 its inertia matrix (with mAi ∈ R>0 ) and by
 
fi = fi,x fi,y fi,z the force that the i-th cable exerts on the object (so that
−fi is the force exerted by the cable on the agent), the dynamics of the i-th
agent is:

Mi p̈Ri = −fi + fCi , (1)

where fCi = Bi (ṗdRi − ṗRi ) + Ki (pdRi − pRi ) + πi (2)

is the ‘control’ force of each agent. To better understand how to control a multi-
rotor aerial robot in such a way, the reader can refer to [21]. Such control
force models three simple actions. First, the agent implements a spring like
action to move towards a desired position or follow a desired path (the apex d
indicates the ‘desired’ quantities). Secondly, the agent implements a dissipative
derivative term proportional to the velocity error. This action damps the oscil-
lations induced by the spring action. Finally, there is a force bias indicated with
 
πi = πi,x πi,y πi,z . This bias is essential to make the flying agents sustain
the weight of the load. We shall show that it plays an important role also for
shaping the system equilibria. Through πi it is possible to set reference internal
forces (forces that do not result in a motion of the object).
The static equilibrium equation of the object subject to the gravity vector
g ∈ R3 is given by [mL g  01×3 ] = W f, ¯ where W ∈ R6×3N is the grasp matrix,
that maps the forces at the contact points to a wrench applied at the object
center of mass and f¯ ∈ R3N ×1 collects the equilibrium cable forces. Resolving
the equation for f¯ we obtain:

f¯ = W † [mL g  01×3 ] + t, (3)

where † indicates a right (pseudo)inverse, and t ∈ null(W ) is the internal force,

which neither influences the object dynamics nor balances any external wrench.
Generally, in leader follower approaches it may occur that only one agent,
i.e., the leader, is aware of the desired trajectory (pdRi , ṗdRi ). On the other hand,
slave agents tend to stay where they are if no external action intervenes. We
 A Study on Force-Based Collaboration in Flying Swarms 7

model such agents by setting Ki = 0 Nm−1 and ṗdRi = 0 ms−1 in (2).

When the leader starts moving, the followers will perceive a modification of
the environment through a change in their cable force. This, in turn, will make
the followers move toward the leader agent trying to bring back the force to
the initial equilibrium value. It is worthy to note that, once each agent stops,
possible load oscillations are damped thanks to the dissipative action modeled
in (2).

3 Numerical Study

For a system of only two aerial agents and a beam/like load, in [21] we formally
proved, using a Lyapunov-based approach, the stability or instability of all the
possible equilibria and the passivity of the overall controlled system. However,
it is not trivial to extend those theoretical results to N > 2 and to a more
generally shaped object. In particular, it is not straightforward to solve the so
called equilibria inverse problem in [21], namely to find all the possible positions
of the agents and forces in the cables for each stable pose of the load. The authors
in [1] showed that such problem is very complex to solve even if just less than
six cables of assigned length are used. Analytical answers to the problem are
therefore difficult to reach. However, a numerical study of whether some of the
properties discovered in [21] for the two-agent system apply also to the swarm
case, where the number of follower agents may be arbitrary large and the object
not only a bar, is equally interesting. Thus, in this section we extrapolate some
conjectures on the expected behavior of the swarm system and validate these
hypotheses through a wide numerical study.

Table 1. The different simulations setups (scenarios).

Goal of the study Scenario Internal forces CoM Points Li Unknown parameters
Internal force role Sa t = 0 [N] G ∈ I On a circle None
Sb t = 0 [N] G∈
/ I On a circle None
Sc t = 0 [N] G∈
/ I On a circle None
Parametric uncertainty Sd t = 0 [N] G∈
/ I Random mL , N (only bounds)

Table 1 contains the description of the simulated conditions. The load has
mass mL = 5 kg and inertial matrix equal to JL = I3 kgm2 . The leader agent,
set as the agent 1, is fed with a sufficiently smooth reference trajectory consti-
tuted by a 5th order polynomial in 3D (rest to rest trajectory with zero acceler-
ation at start and end points), which lasts 10 s and covers 2 m. M1 = 0.5I3 kg,
Mi = 0.01I3(N −1) kg, B1 = 100I3 Nsm−1 Bi = 1.5I3(N −1) Nsm−1 , K1 =
1000I3 Nsm−1 , and Ki = 0I3(N −1) Nm−1 for i = 1. When N > 30, we changed
the apparent mass Mi , and the damping Bi of the followers, i.e., for i = 2, ..., N
reducing both by 90%. This allows the leader agent to drag the system without
8 C. Gabellieri et al.

1 1

0.5
0.5

0
0

-0.5
-0.5

-1
5 -1
6 5 6
4 4
0 2 2
0 0 0
-2 -2
-5 -4 -5 -4
-6 -6

(a) t = 0 (b) t = 0

1.5
1

0.5 1

0
0.5
-0.5

-1 0
6 5
4
2
0 0 -0.5
-2
5
-4 -5 0
-6 2 4 6
-5 -4 -2 0
-6

(c) cables randomly attached on the (d) Simulated system during trans-
surface of the object portation

Fig. 2. Simulated system for the case N = 6 ((a), (b), (c)) and N = 60 (d)). The
object is depicted as a grey surface, and the light spot on it coincide with its center of
mass G. In case G ∈ / I, the light spot is instead the projection of G on I. The cables
are the black lines, while the circles represent the robots (the darker one is the leader).

applying too large forces, similarly to what happens in biological systems as

described in Sect. 4.
Scenarios Sa , Sb and Sc refer to the cases in which the parameters of the
system are perfectly known, and so is N . Based on what we demonstrated in [21]
for the two agents system, we decided to investigate the role of the internal force
on the object for the equilibria of the system. We considered the case in which the
points Li lie on a circle centered around the projection of G on I (see Figs. 2(a),
(b), and (d)). We ran different simulations where N is a random number between
2 and 100, and the initial pose of the object has a random value of yaw between
0 rad and π/4 rad.
In the scenarios Sa and Sb the force bias in (2) is set to sustain the weight
of the load without internal forces, namely πi = [0 0 mL g/N ] ∀i = 1, . . . , N ,
where g is the intensity of g. This implies vertical cables. Once the leader agent
stops, the system converges to an equilibrium in which the final attitude of the
transported object is not univocally determined but depends on the trajectory
resulting from the initial condition and the leader desired trajectory. In partic-
ular, the results in the first line of plots of Fig. 3 shows a completely arbitrary
 A Study on Force-Based Collaboration in Flying Swarms 9

5 0.5

40
0 0

30 -5 -0.5

20 -10 -1

-15 -1.5
10

-20 -2
0 5 10 15 0 10 20 30 40 50 0 10 20 30 40 50

45 5

40 0

35 0 -0.2

30 -0.4

25 -5 -0.6

20 -0.8

15 -10 -1

10 -1.2

5 -15 -1.4
0 10 20 30 40 50 0 50 100 150 200 250 300 350 0 50 100 150

0.3
40

35 0 0.2

30 -1
0.1
25
-2
0
20
-3
15 -0.1
-4
10
-0.2
5 -5

0 -0.3
2 4 6 8 10 12 14 16 18 0 10 20 30 40 50 60 0 5 10 15 20

Fig. 3. Object attitude during transportation. First row: scenario Sa : fint = 0 N and
G ∈ I. Second row: scenario Sb : fint = 0 N and G below I. Third row: scenario Sc :
fint = 0.8N and G placed below I. The dotted line is the desired value. The first,
second, and third columns refer to the trajectories of the yaw (ψ), pitch (θ), and roll
(φ) angles of the object, respectively.

final orientation of the load for the scenario Sa , and the second line of Fig. 3 an
arbitrary final yaw for the scenario Sb .
In the scenario Sc we set the force bias πi in (2) so that the system reaches
an equilibrium with the cables forces applying non zero horizontal forces. More
in detail, indicating with ¯li the vector that connects the projection of G on I to
the position of point Li at the final attitude, we set [πi,x πi,y ] = fint ¯li /||¯li ||,
where fint is thus the intensity of each agent’s planar force bias. In other words,
the desired horizontal forces in the cables are oriented radially and outward the
object in the final configuration, similarly to what is depicted in Fig. 2(b). Since
the object reaches an equilibrium, and the external wrench on the object at
the equilibrium is only the vertical force due to the gravity, as in (3), such non
vertical components of the cable forces, which do not cause any motion of the
object and do not compensate any external wrench, generate an internal force
t = 0. In this way, the object reaches always the same attitude: zero pitch,
zero roll, and the same yaw. The third line of Fig. 3 shows the results of such
simulations. Notice that the same results have been obtained even when G ∈ I.
10 C. Gabellieri et al.

(a) Mi = 0.001I3(N−1) kg, (b) Mi = 0.01I3(N−1) kg,

Bi = 0.15I3(N−1) N s m−1 Bi = 0.15I3(N−1) N s m−1

(c) Mi = 0.001I3(N−1) kg, (d) Mi = 0.01I3(N−1) kg,

Bi = 1.5I3(N−1) N s m−1 Bi = 1.5I3(N−1) N s m−1


2 2
Fig. 4. Intensities of the vertical force fi,z and of the horizontal force fi,x + fi,y that
the agents apply to the object during the transportation for four different parameters
of the followers in scenario Sc with N = 70. The dotted line refers to the leader agent,
while the solid lines refer to the followers. In the subcaptions i = 2, . . . , N .

Figure 4 shows the evolution of the forces that the agents apply to the load in
four different cases belonging to scenario Sc .
We also decided to see whether the value of fint influences the speed of
convergence with which the swarm stabilizes the object in scenario Sc . Figure 5
shows the evolution of the load attitude when transported by a group of five
agents for different values of fint .
We present now the results concerning scenario Sd , where neither the exact
real mass of the load, mL , nor the exact total number of agents, N , are known by
the agents. Only some upper and lower bounds are given, indicated with mmax ,
mmin , Nmax and Nmin respectively. Therefore, we introduced a particular choice
of πi .

⎡ ⎤ ⎧
0 ⎪
⎨−KZ if |fi,z | > fmax
πi = ⎣ 0 ⎦ where Ui =
z
KZ if |fi,z | < fmin . (4)
⎪
⎩
fi,z + Ui
z
0 otherwise
 A Study on Force-Based Collaboration in Flying Swarms 11

0.4 25 2

0
0.2 20
-2
0
15 fint = 2 N -4
fint = 2 N fint = 2 N
-0.2 fint = 4 N
fint = 4 N 10 -6 fint = 4 N
-0.4 fint = 6 N
fint = 6 N -8 fint = 6 N
fint = 0.5 N
fint = 0.5 N 5
-0.6
-10 fint = 0.5 N

-0.8 0 -12
0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100

Fig. 5. Evolution of the object attitude (yaw, pitch and roll, respectively) during trans-
portation for different values of the intensity of the internal force in scenario Sc with
N = 5.
25 10 20

20 0
10

15 -10
0
10 -20
-10
5 -30

-20
0 -40

-5 -50 -30
0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25

Fig. 6. Attitude of the load (yaw, pitch and roll, respectively) during the transportation
task in scenario Sd with random cables attachment points and uncertain load mass and
agents number.

Namely, we are defining a dead-zone in the sensed vertical force such that if
the robots perceive a vertical force in the cable that is inside a certain range
defined by fmax and fmin , they ignore it; otherwise, they apply an upward or
downward force trying to restore a vertical force inside the predefined range.
KZ is a constant that determines the responsiveness of the robots in trying to
maintain the vertical force inside [−fmax , −fmin ]. We choose fmax = mNmax min
g

and fmin = Nmax . In this way, we guarantee that the overall force exerted
mmin g

by the agents will be enough to sustain the object weight. In particular, we

are not choosing a precise reference force distribution. The actual final force
in each cable is induced by the choice of the bias (which does not depend on
the exact values of N and mL ) and is not the same for all the agents. Such an
implementation allows the robots to successfully cope with a variation of the
parameters of the swarm and of the object. Simulations results for this scenario,
see Fig. 6, show that the system stabilizes after the transportation. However,
due to the uncertain conditions, the final pose of the object cannot be known a
priori. The upper and lower bounds have been set to mmax = 7 kg, mmin = 3 kg,
Nmax = 50 and Nmin = 7.
We conclude this section proposing and testing an alternative version of (2)
for the leader agent. Thanks to the model (1) and (2), the leader agent does not
blindly follow the desired trajectory but is aware of the outer world. In the choice
of the parameters of (2) one has to face a clear trade-off between compliance
12 C. Gabellieri et al.

10 10

5 5

0 0
0 20 40 60 80 100 0 20 40 60 80 100
||f|| ||f||
20 30
master
20 master
10
10

0 0
0 20 40 60 80 100 0 20 40 60 80 100

(a) (b)

Fig. 7. The three components of pR1 and pdR1 (dotted lines of the same color) and the
intensity of the cable forces. By modeling fC1 as in (5) it is possible, with the same
parameters, to have the leader follow its desired trajectory under normal conditions
(a), and to limit the forces exerted in case the swarm does not follow it (b).

(i.e., energy consumption and safety) and tracking error (i.e., performance). A
behavior that is too compliant can compromise the reach of the final position,
a behavior that is too stiff can require a large amount of force from the leader
side and thus a lot of energy consumption and an increased risk of ruptures.
Due to space limitations we omit the performed numerical results showing such
intuitive trade-off. A possible solution to better deal with such trade-off is to
introduce a nonlinear saturation model like, e.g., the following one:

fC1 = B1 tan−1 (ṗdR1 − ṗR1 ) + Ki tan−1 (pdR1 − pR1 ) + πi . (5)

The relative results are shown in Fig. 7.

4 Discussion

We first simulated symmetric cables attachment points, lying on a circle around

the object center of mass (or its projection on I). Applying a non-zero internal
force that stretches the object the agents are capable of controlling univocally
the final orientation of the object without the need of communication. This
result actually reflects what we had discovered and formally proven for the two-
agent system and a beam-like load in [21]. It is possible to assume that also
biological agents, transporting an object, may be able to pull the edge of the
object towards themselves, applying a force orthogonal to the edge, and thus
generating a resulting internal force altogether. For example, applying a constant
horizontal force bias that equilibrates the object at the initial configuration, the
swarm may be capable of transporting the load and repositioning it with the
same attitude without the need of explicit communication. This suggests that
the local sensing allows a sort of collective memory, as if the swarm of agents
remembered the initial orientation of the object and were capable of recreating it.
 A Study on Force-Based Collaboration in Flying Swarms 13

We were also interested in understanding the role played by the internal

forces in the convergence rate. In fact, observing some biological structures, it
seems reasonable to assume that a higher internal force in the object might lead
to a faster convergence of the system to the equilibrium. For example, in [2] the
authors observe the dynamic response of different spider webs. It came out that
an initial tension in the net changes the natural frequency in the sense that the
net is faster in recovering its equilibrium after being perturbed with respect to
the case without initial tension. Similarly, we found that increasing the intensity
of the horizontal bias force in fC , and hence in turn increasing the internal force
in the object, leads to faster a convergence rate, as shown in Fig. 5.
The apparent mass of the leader agent has been set to a larger value than the
one of the followers. Actually, it has been observed in groups of different army
ants (Eciton burchelli, Dorylus wilverthi) that the front ant in prey retrieval
groups is larger than the others, and it is characterized by a larger dry weight.
Additionally, in our simulations, during the dynamic part, the leader agent has to
apply greater forces to the object than the other ones in the group. An example
is shown in Fig. 4. However, this seems to be true also in super efficient prey
retrieval groups of army ants [5], where a single submajor initiates the motion of
the item by itself, and only after the rest of the group helps in the transportation.
The damping parameter of the followers Bi and their masses Mi have been
tuned depending on the number of robots. This has been done to reduce the
force that the leader agent needs to apply in order to let each follower start
moving. This force increases with the increasing of the damping parameters of
the followers that are related to energy dissipation of the overall system. The
master force in case of a small N with a large damping is comparable to the force
that the master applies with a large N and small damping, a behavior which is
reported in Fig. 4(a). To equilibrate as much as possible the force exerted by the
leader for different values of N , one solution is to modify the damping parameter
of the followers based on N . Another way to modify the effect of the follower
inertia on the leader it is to modify their mass. Compare Figs. 4(d) and (b) to
see the effect of the follower inertia on the master applied force, and Figs. 4(d)
and (b) to see the effect of the damping parameter. Finally, comparing Figs. 4(d)
and (a) the benefits of decreasing both the followers mass and damping emerges.
It is not immediately clear which is the possible biological meaning of tuning the
mass of the agents though. However, it is interesting to highlight that Eciton
burchelli ants, very efficient in transportation tasks, tend to maintain a constant
ratio between the total weight of porters and the weight of the carried load [5]. f
Concerning the leader spring parameter, it is not trivial to find the right balance
in order to have a leader agent capable of both following the desired trajectory
not being too much perturbed by small forces (for instance the ones caused by
the followers initial resistance to the motion) and being sensitive to large forces
on the cable indicating that something is not going as expected. However, that
is exactly what an intelligent biological system would do. We were able to mimic
this behavior by using a nonlinear control action on the leader agent, see (5). In
this way, the leader follows the desired trajectory accurately for small deviations,
14 C. Gabellieri et al.

but it changes its behavior consistently if the followers stay still and not follow
as expected (see Fig. 7).
Finally, the results in Fig. 6, which refer to scenario Sd , show the potential
of the algorithm also in conditions were the swarm is not completely aware of
the parameters of the system. Even with a limited and imperfect knowledge the
agents are capable of commonly carrying an object in a stable way.

5 Conclusions
This work is a simulative study on communication-less cooperative object trans-
portation by swarms of aerial agents. The main focus concerns the important
role of the internal force to make asymptotically stable certain equilibria of the
system and to enhance the manipulation capabilities of the swarm. An imperfect
knowledge of the system parameters has been also treated. Several parallelisms
with biological examples are discussed as well. Validation of the proposed algo-
rithm on real platforms and theoretical proofs is left as an important future work.
Of course, in order to realize a practical implementation, additional aspects, such
as a collision avoidance technique, are required, especially for very large number
of agents. Furthermore, the relaxation of some hypothesis, e.g., on the cables
attachment points or on the rigidity of the load, may represent an interesting
enlargement of the application domain.

Acknowledgments. This research was partially supported by the ANR, Project

ANR-17-CE33-0007 MuRoPhen.

References
1. Abbasnejad, G., Carricato, M.: Direct geometrico-static problem of undercon-
strained cable-driven parallel robots with n cables. IEEE Trans. Robot. 31(2),
468–478 (2015)
2. Alam, M.S., Wahab, M.A., Jenkins, C.H.: Mechanics in naturally compliant struc-
tures. Mech. Mater. 39(2), 145–160 (2007)
3. Berman, S., Lindsey, Q., Sakar, M.S., Kumar, V., Pratt, S.C.: Experimental study
and modeling of group retrieval in ants as an approach to collective transport in
swarm robotic systems. Proc. IEEE 99(9), 1470–1481 (2011)
4. Brescianini, D., D’Andrea, R.: Design, modeling and control of an omni-directional
aerial vehicle. In: 2016 IEEE International Conference on Robotics and Automa-
tion, Stockholm, Sweden, pp. 3261–3266, May 2016
5. Franks, N.R.: Teams in social insects: group retrieval of prey by army ants (eciton
burchelli, hymenoptera: Formicidae). Behav. Ecol. Sociobiol. 18(6), 425–429 (1986)
6. Franks, N.R., Sendova-Franks, A.B., Anderson, C.: Division of labour within teams
of new world and old world army ants. Anim. Behav. 62(4), 635–642 (2001)
7. Gassner, M., Cieslewski, T., Scaramuzza, D.: Dynamic collaboration without com-
munication: vision-based cable-suspended load transport with two quadrotors. In:
2017 IEEE International Conference on Robotics and Automation, Singapore, pp.
5196–5202, May 2017
 A Study on Force-Based Collaboration in Flying Swarms 15

8. Grassé, P.P.: La reconstruction du nid et les coordinations interindividuelles

chezbellicositermes natalensis etcubitermes sp. la théorie de la stigmergie: Essai
d’interprétation du comportement des termites constructeurs. Insectes Sociaux
6(1), 41–80 (1959)
9. Kube, C.R., Bonabeau, E.: Cooperative transport by ants and robots. Robot.
Auton. Syst. 30(1–2), 85–101 (2000)
10. Manubens Ferriol, M., Devaurs, D., Ros, G.L., Cortés, J.: A motion planning app-
roach to 6-D manipulation with aerial towed-cable systems. In: Proceedings of the
2013 International Micro Air Vehicle Conference and Flight Competition, Toulouse,
France, pp. 1–7 (2013)
11. Masone, C., Bülthoff, H.H., Stegagno, P.: Cooperative transportation of a payload
using quadrotors: a reconfigurable cable-driven parallel robot. In: 2016 IEEE/RSJ
International Conference on Intelligent Robots and Systems, pp. 1623–1630, Octo-
ber 2016
12. Michael, N., Fink, J., Kumar, V.: Cooperative manipulation and transportation
with aerial robots. Auton. Robot. 30(1), 73–86 (2011)
13. Moffett, M.W.: Cooperative food transport by an asiatic ant. Natl. Geogr. Res.
4(3), 386–394 (1988)
14. Nguyen, H.N., Park, S., Lee, D.J.: Aerial tool operation system using quadrotors
as rotating thrust generators. In: 2015 IEEE/RSJ International Conference on
Intelligent Robots and Systems, Hamburg, Germany, pp. 1285–1291, October 2015
15. Park, S., Her, J., Kim, J., Lee, D.: Design, modeling and control of omni-directional
aerial robot. In: 2016 IEEE/RSJ International Conference on Intelligent Robots
and Systems, Daejeon, South Korea, pp. 1570–1575 (2016)
16. Rajappa, S., Ryll, M., Bülthoff, H.H., Franchi, A.: Modeling, control and design
optimization for a fully-actuated hexarotor aerial vehicle with tilted propellers. In:
2015 IEEE International Conference on Robotics and Automation, Seattle, WA,
pp. 4006–4013, May 2015
17. Rubenstein, M., Cabrera, A., Werfel, J., Habibi, G., McLurkin, J., Nagpal, R.: Col-
lective transport of complex objects by simple robots: theory and experiments. In:
2013 International Conference on Autonomous Agents and Multi-agent Systems,
pp. 47–54 (2013)
18. Ryll, M., et al.: 6D physical interaction with a fully actuated aerial robot. In:
2017 IEEE International Conference on Robotics and Automation, Singapore, pp.
5190–5195, May 2017
19. Sreenath, K., Kumar, V.: Dynamics, control and planning for cooperative manip-
ulation of payloads suspended by cables from multiple quadrotor robots. In:
Robotics: Science and Systems, Berlin, Germany, June 2013
20. Tagliabue, A., Kamel, M., Verling, S., Siegwart, R., Nieto, J.: Collaborative trans-
portation using MAVs via passive force control. In: 2017 IEEE International Con-
ference on Robotics and Automation, Singapore, pp. 5766–5773 (2016)
21. Tognon, M., Gabellieri, C., Pallottino, L., Franchi, A.: Aerial co-manipulation with
cables: the role of internal force for equilibria, stability, and passivity. IEEE Robot.
Autom. Lett. Spec. Issue Aer. Manip. 3(3), 2577–2583 (2018). https://doi.org/10.
1109/LRA.2018.2803811
22. Wang, Z., Schwager, M.: Force-amplifying n-robot transport system (force-ANTS)
for cooperative planar manipulation without communication. Int. J. Robot. Res.
35(13), 1564–1586 (2016)
23. Wu, G., Sreenath, K.: Geometric control of multiple quadrotors transporting a
rigid-body load. In: 53rd IEEE Conference on Decision and Control, Los Angeles,
CA, pp. 6141–6148, December 2014
 Automatic Design
of Communication-Based Behaviors
for Robot Swarms

Ken Hasselmann1 , Frédéric Robert2 , and Mauro Birattari1(B)

1
IRIDIA, Université Libre de Bruxelles, Brussels, Belgium
{khasselm,mbiro}@ulb.ac.be
2
BEAMS, Université Libre de Bruxelles, Brussels, Belgium
frrobert@ulb.ac.be

Abstract. We introduce Gianduja, an automatic design method that

generates communication-based behaviors for robot swarms. Gianduja
extends Chocolate, a previously published design method. It does so
by providing the robots with the capability to communicate using one
message. The semantics of the message is not a priori fixed. It is the auto-
matic design process that implicitly defines it, on a per-mission basis, by
prescribing the conditions under which the message is sent by a robot
and how the receiving peers react to it. We empirically study Gianduja
on three missions and we compare it with the aforementioned Chocolate
and with EvoCom, a rather standard evolutionary robotics method that
generates communication-based behaviors. We evaluate the behaviors
produced by the three automatic design methods on a swarm of 20 e-puck
robots. The results show that Gianduja uses communication meaning-
fully and effectively in all the three missions considered. The aggregate
results indicate that, on the three missions considered, Gianduja per-
forms significantly better than the two other methods under analysis.

1 Introduction
In swarm robotics, communication plays a central role and can significantly
enhance collective performance [3]. Designing effective communication mech-
anisms is challenging and design choices can have an important impact on
the effectiveness, complexity, and cost of a swarm [2]. Notwithstanding the
advancements achieved in the last decade [4,7,24,29,34,43,51], the design of
robot swarms is still at dawn and no generally applicable methodology has been
proposed so far [8,11,21]. Automatic design methods are a promising way of
approaching the issue [6,15]. In automatic methods, the design problem is cast
into an optimization problem: a space of solutions is searched via an optimiza-
tion algorithm, with the goal of maximizing a performance measure. Most of the
The proposed method was implemented and tested by KH. The experiments were
designed by the three authors. This paper was drafted by KH, refined by MB, and
revised by the three authors. The research was conceived and directed by MB.
c Springer Nature Switzerland AG 2018
M. Dorigo et al. (Eds.): ANTS 2018, LNCS 11172, pp. 16–29, 2018.
https://doi.org/10.1007/978-3-030-00533-7_2
 Automatic Design of Communication-Based Behaviors for Robot Swarms 17

research on the automatic design of robot swarms has been inspired by neuro-
evolution [37,47]. In this approach, robots are controlled by a neural network,
whose parameters are obtained via artificial evolution [12,27,31,38,44,45,47,50].
Other methods have been proposed that are based on different control archi-
tectures and/or different optimization algorithms [16,18,22,30]. Among them,
Chocolate [16] produces probabilistic finite state machines by using the irace
optimization algorithm [35] to assemble preexisting low-level behaviors and con-
ditions, and to fine-tune their parameters. The low-level behaviors, which define
the actions that individual robots can perform, are: exploration, stop, phototaxis,
anti-phototaxis, attraction to neighbors, repulsion from neighbors. The condi-
tions, which define events that cause a transition between low-level behaviors,
are: black-floor, white-floor, gray-floor, neighbor-count, inverted-neighbor-count,
fixed-probability.
In this paper, we study the automatic design of collective behaviors that rely
on communication. In particular, we are interested in exploring the case in which
messages exchanged by the robots do not have an a priori defined semantics. We
wish to develop an automatic design process that, on a per-mission basis, defines
(i) the conditions under which a robot broadcasts a message and (ii) the effects
that this message has on the behavior of the receiving peers.
We introduce Gianduja, a new instance of AutoMoDe [18]. Gianduja extends
Chocolate by adding the capability of locally broadcasting a single message and
reacting to it. We test Gianduja on three missions that we shall call aggrega-
tion, stop, and decision. We present results of experiments performed with a
swarm of 20 e-puck robots [36].
Within the evolutionary robotics approach, it has already been shown that an
automatic design process can (implicitly) give a semantics to an a priori mean-
ingless message. Nonetheless, this has been demonstrated only on teams of two
robots [1,49]. The novel elements that we propose in this paper are that: (1) we
study the emergence of a message semantics in swarm robotics and we demon-
strate it with a swarm of 20 robots; (2) we show that a message semantics can
emerge also when robots are controlled by a finite state machine; and (3) we
consider three different missions in which the emerging semantics is different.

2 Related Work
Communication—be it direct or indirect, explicit or implicit—is an integral part
of most robot swarms demonstrated so far. As a result, the literature on com-
munication in swarm robotics is extremely large and covering it goes beyond the
scope of this paper. In particular, we will not cover studies in which communi-
cation has been a priori defined by the designer—e.g., [2,3,9,14,28]. Instead, we
will focus on studies in which communication has been automatically designed.
The vast majority of studies in which communication emerged from an auto-
matic design process belong within evolutionary robotics [37,47,48]. Quinn et al.
[41,42] were the first to study the emergence of communication between agents.
In their studies, robots move in an arbitrary direction while staying close to each
Another random document with
no related content on Scribd:
machine. Fortunately I had presence of mind enough to keep my
watch going, as well as the captain’s chronometer, for otherwise I
would have had no knowledge of the passage of time. Once or twice
the scarlet women visited the ship, but seemed nervous and wary,
and made no effort to approach or molest me, merely gazed about
as if searching for something—perhaps for me—and then retiring.
Several times, too, I ventured on deck, and peered over the ship’s
side, but saw none of the giantesses, although with the glasses I
could see crowds of the beings about the city in the distance.
“Also, I noticed among them, several individuals who were much
smaller than the rest, and who appeared to be men, although I could
not be sure. I also discovered, and almost lost my life in the
discovery, that the atmosphere of this place is unfit for human beings
to breathe, and is thick with sulphurous fumes. Close to the ground
these fumes are so dense that a person would succumb in a few
moments, but at the height of the Chiriqui’s decks, nearly seventy
feet above the rocky bed on which she rests, the air is breathable,
although it causes one to choke and cough after a few minutes. And
I am sure that the houses of these giant beings have been built on
the summits of the basalt columns in order to avoid the suffocating
fumes of the lower levels. Later, too, I learned that the membrane-
like frills upon these creatures are a sort of gills, or as I might say,
natural gas-masks, which by some means enable the beings to
breathe the sulphur-laden air. But even with these, they avoid the
lower areas where the fumes are the worst, and only visit them when
necessity arises, which accounts for my being left in peace, with
none of the horrible women near the ship, for days at a time. I
discovered the presence of the sulphur gas on the first day when,
attempting to eat, I removed my gas-mask. Suffocating as I found
the fumes, I was compelled to endure them, and gradually I became
slightly accustomed to them, so that now I have little trouble in
breathing during the short time it takes me to eat my meals. At all
other times I must wear the apparatus, and I thank God that this is
so, for I know now that it is the gas-mask which so far has preserved
my life.
“On the tenth day after my arrival I noticed a number of the
giantesses gathering about the huge, spherical airship which still
rested on its cradle near the Chiriqui, but which, I have forgotten to
state, ceased to emit its green or red lights after it had landed. Lying
there it resembled nothing so much as a gigantic can-buoy or a
floating mine, if one can imagine a buoy two hundred yards in
diameter.
“On the day I mentioned, all interests seemed to be centered on
the thing, and cautiously peering from the shelter of the deck-house,
I watched the proceedings. Presently several of the women entered
the sphere through an opening in its middle band; the aperture
closed behind them, and immediately there was a low, humming
sound as of machinery. As the sounds issued from the sphere, the
cables to which were attached the smaller spheres (which glowed
red when carrying the Chiriqui through the air) were drawn in until
the two smaller spheres were resting in recesses at the axes of the
large sphere, and where they appeared merely as hemi-spherical
projections. Then, slowly at first, but with ever increasing speed, the
slender rods about the large sphere began to move back and forth,
or rather in an oscillating manner, until they were vibrating with such
rapidity that they appeared merely rays of light. Slowly, majestically,
the immense globe rose from its cradle, and gathering headway,
leaped upward to an immense height. Then, tilting at an angle, it
passed over the city and headed for an immense pinnacle of rock,
which, fully seven miles from where I stood, reminded me of a
gigantic chimney or funnel.
“Although it was barely visible to the naked eye, I could see it
distinctly through the glasses, and I watched it with the most intense
and concentrated interest. For a few moments it remained, poised a
hundred feet or so above the pinnacle. Then, from the towering,
tapering rock, a terrific jet of steam roared forth, and striking the
great spherical machine above it, hurled it upward and beyond my
vision. Give close heed to these words, whoever may, by God’s
grace, be listening to what I say, for upon them may hinge the fate of
the human race. Only by this means, by being shot upward by this
titanic jet of steam, can the airship leave this subterranean land and
emerge through the crater by which it entered bearing the Chiriqui.
Within this place it can sail at will; once above the crater opening it
can travel anywhere, although it cannot land; but by some unknown
force or magnetic attraction or freak of gravitation the machine
cannot ascend through the crater, although, when over it, it will drop
like a plummet through the opening. And herein—for the sake of
humanity, listen to this and remember my words—lies a means of
destroying the machine, for by surrounding the crater with powerful
guns the sphere can be shelled as it emerges and utterly destroyed.
To attempt to do so as it returns to the crater would be suicidal, for
once in the outer air, it emanates vast quantities of most poisonous
gas, and all living things within a radius of several miles would be
struck down unconscious, as were my companions on the Chiriqui.
Even if gas-masks were worn, it would be most difficult to destroy
the machine as it descended, for it travels with incredible speed in its
descent and, moreover, the terrible creatures who man the thing
would see that enemies lurked near and would find some means of
destroying them, or by the mysterious magnet force they control,
would draw even the heaviest cannon to the machine as an ordinary
magnet draws needles or iron filings. So if the thing is to be
destroyed, it must be done as the machine emerges from the crater.
Would to God that I could tell where the crater is, but beyond feeling
sure it is at the summit of an Andean peak, I have no means of
locating it.
“But I was telling of what occurred on that tenth day when the
spherical airship was projected from my sight by the blast of steam.
As the machine vanished, the women who had watched its
departure, returned to their city, and I swept the landscape with my
glasses, wondering at the bleak, terrible scenery and bizarre colors.

“As I focussed the binoculars upon a level plateau, perhaps a mile

from where the Chiriqui rested, I gasped in surprise. Clearly defined,
lay the remnants of what had once been a steamship! Had I given
the matter thought, I might have known that the Chiriqui was not the
first vessel to have fallen a victim to these awful beings; but the sight
of another ship’s skeleton came to me as a terrific shock. As nearly
as I could judge, the vessel had been dismantled, for only the great
steel frame remained, with the mighty boilers and other portions of
the ship scattered about, and gruesomely like some mammoth
creature lying disemboweled upon the earth.
 “I was consumed with a mad desire to visit that pathetic wreck, but
I knew not to what dangers I would be exposed, once I left the
security of my ship. Not a being was in sight, however, and carefully I
studied the land, visually measuring the relative distances between
myself and the wreck, and between the city and the route I must
traverse. Having already observed that the giantesses moved slowly
and cumbrously on foot, I at last decided that even if they attempted
to intercept me I could regain the Chiriqui before I was overtaken, so
I threw caution to the winds and prepared to undertake my
hazardous journey. Slinging the loaded rifle on my back, with the
revolver at my belt, and still further arming myself with a keen-edged
fireaxe, I hunted up the pilot’s ladder, lowered it over the lowest side
of the ship,—which was also the side farthest from the city,—and
clambering down the Chiriqui’s lofty sides, leaped down upon the
ground. To my amazement, I landed in a dense jungle of dry, tough
vegetation which rose to my shoulders. From the deck, looking
directly downwards, I had thought this dull-green growth a short, wiry
grass, and, of course, in its relative proportion to the gigantic women,
it was no higher than ordinary grass to a normal human being. It was
a wonderful example of the theory of relativity, but my mind was not
interested in scientific matters at the time, and I merely gave thanks
that the miniature jungle,—which I saw was composed of giant
lichens—would afford me cover through which I might sneak in
safety, and with little chance of detection.
“Without much difficulty I made my way to the other vessel, and
found her even more dissected than I had supposed. Why the
denizens of the place had torn her to bits I did not then know, but
certain portions of her machinery and fittings had been left intact,
and, as I examined these, I made another and most astounding
discovery. Deeply engraved upon a brass plate was the ship’s name
‘U. S. S. Cyclops!’ For a space I stood staring, scarcely able to
believe my eyes. Here then was the solution to that mystery of the
sea, the disappearance of the collier, as laden with manganese, she
vanished without word or trace when off the Barbados during the
World War. No doubt, I thought, many a mystery of the sea had been
caused by the damnable work of these beings with their infernal
machine. But why, for what reason, did they capture ships? Why did
they carry off the unconscious persons upon the vessels? And why
did they tear the vessels apart? It was all a mystery which, in all its
horrible, gruesome, ghoulish details I was soon to solve.
“There was nothing more to be learned from the remains of the
Cyclops, and in safety I returned to the Chiriqui to find, to my
surprise and terror, that a gang of the monstrous females had
boarded the ship in my absence and were stripping her of
everything. But as they caught sight of me, all threw down whatever
they had and fled precipitately, leaving me once more in undisputed
possession of the ship. I was relieved at this, for it was obvious that I
had no need to fear the creatures. By now, too, I had formulated a
theory to account for this strange dread of a being who was a puny,
miserable thing compared to them. Unquestionably my gas-mask
rendered me a most grotesque and unknown creature in their eyes.
My remaining alive and active while all others upon the ship had
succumbed to the noxious gas had probably caused them to think
that I was a supernatural being. The fact that I could go about and
breathe the sulphur-laden air would cause them to regard me with
even greater wonder and superstition, and, as I found later, the fact
that I was never seen to eat, confirmed their belief that I was some
mysterious being against whom their gases and their deviltries were
of no avail.
“I had not much time to devote to such matters, however. Soon
after regaining the Chiriqui I heard excited cries from the land, and
looking over the ship’s rails, I found an immense crowd had gathered
near the empty cradle of the airship, and that all were gazing
upward. Following their example, I stared into the greenish void and
instantly understood. Descending rapidly towards the plain, came the
great sphere, and, suspended below it, was the hull of another
captive ship. And as I focussed my glasses upon this, I rubbed my
eyes and gaped. The dull gray color, the lines, the raking funnels, the
barbettes and gun muzzles left no room for doubt. Incredible as it
seemed, the captive vessel was a warship! What hope then had my
fellow men upon earth? What chance was there if these giant
creatures could send forth their flaming machine, and by it, capture
the fastest, most powerful war-vessels—all within the space of a few
hours?
 “Rapidly the machine and its burden approached, and presently
descended gently dropping the war vessel close to the Chiriqui. My
worst fears were confirmed. The vessel was an American destroyer,
the McCracken, and I knew that scores of my countrymen must lie
unconscious upon her, and in a few moments would be carried off to
some unknown horrible fate. What that fate was I had already
surmised. That first demonstration of the ferocious cannibalism of
the giantesses upon the Chiriqui’s deck had been enough to make
my blood run cold.
“But I had not yet guessed even a fraction of the true horror of it.
Scarcely had the McCracken been dropped upon the earth, when
the women swarmed upon her, and once more I saw the creatures
gathering the inert forms of men and carrying them to the city. And
rapidly, too, they commenced dismantling and tearing the destroyer
into bits. How they had accomplished this with the Cyclops had
puzzled me, but now I witnessed the process close at hand. From
the vicinity of the waterfall, lines or pipes were led to the vessel’s
side; presently there was the roaring sound of steam; dense clouds
of vapor arose from the cataract; the water ceased to flow, and from
the extremities of the lines or tubes twenty-foot jets of blinding flame
shot out. As easily as though made of wax, the steel sides, the
massive beams, the armored barbettes of the warship melted and
were cut by these jets, and as the pieces fell apart, the spherical
airship took a position above the vessel, and by its magnetic power,
lifted tons of the fragments, then sailing off, deposited them in some
spot beyond the city. It was then, as I saw the ship rapidly dissolving
before my eyes, that the inspiration came to me which may make it
possible for me to communicate with the outside world and may, if
God wills, serve to warn my fellow men of the fate which will
overtake them if these terrible creatures are allowed to follow out
their plans. As the jets of flame cut through the McCracken’s
superstructure, and the radio antennae fell in a tangled mass across
the deck, I forgot all else and rushed to the wireless room of the
Chiriqui. Here was my chance. If the ship’s radio transmitter was still
in working order; if the auxiliary battery was still charged, I might
send out messages which, small as the chances were, might reach
the ears of some of the countless thousands of persons who listened
each night at their receiving sets. I trembled with fear that I would
find the transmitter injured or dismantled. I shook with dread that the
battery might be dead. I felt faint with apprehension that the
message, if sent, might never penetrate the sulphur-laden
atmosphere or might never reach the outer world. And I realized,
with a sickening sinking of my heart, that even if heard my
communication might be regarded as a hoax, and no attention would
be given it. But I would do my best. The radio set had not been
molested. Everything was in working order, and I set myself the task
of transmitting my story each night at the same hour, repeating it
over and over again, until the storage batteries are exhausted, for to
get up steam and start the dynamos is beyond my powers. Had I
knowledge of Morse I would send my story by that code, but I have
not, and so—I must cease. For the love of your race and of your
dear ones listen, I beseech you, until I can resume.”

Here the message broke off abruptly, and Frank and I sat staring at
each other, fearing to speak lest we might interrupt or miss the words
which might come, and listening with straining ears at the head-sets.
For an hour we sat there and then, once more the voice spoke.
“The doom that I feared is approaching. I have been here for three
months and this will, I know, be my final message. Oh that I could
only be sure that someone has heard my words, that my fate has not
been in vain but has served to warn my fellow beings. But I must
hurry on. I have learned everything of importance. I have watched,
studied and have even learned to understand much of the language
of these beings. I found that there were men. They are puny beings
compared to the women, though ten-foot giants compared to normal
men, and they are cowed, abject, mere slaves of the females. Only
enough male children are permitted to survive to propagate the race.
All others are killed.
“As they reach manhood only those males of super-intelligence,
strength and virility are permitted to live. The others are destroyed
and—yes, horrible as it sounds, their bodies, like those of the
murdered infants and of the aged, sick or infirm, are devoured. And
as fast as the males attain middle age their lives are forfeited. Long
ago these beings subsisted upon the few wild creatures which
roamed their land; but long ago all these were exhausted and human
flesh became the only meat. There is no vegetable food, and for a
time the sacrificed surplus males, and the aged, provided food for
the race. But gradually the male births decreased, female children
preponderated, and with the increased population resulting, the
males were too few to nourish the others. Then, through what
damnable accident or design I do not know, the creatures went forth
in their airship and discovered the teeming millions of human beings
on earth.
“But the bulk of humanity was and still is safe from them, at least
until new means of attacking mankind are devised, for the globular
airship cannot approach the land. The very power it uses to lift the
greatest steamships and carry them off, draws the machine to the
earth and holds it fast. But above water, which acts as an insulator
apparently, the apparatus can operate at will. And they have a two-
fold purpose in capturing ships. All the available metal in this land
was exhausted in constructing two of the spherical machines. One of
these never returned from its first trip, and only the one remains. To
construct more, these giant women plan to use the metal salvaged
from captured ships, until a vast fleet of the infernal things is ready to
go forth and wipe the seas clean of ships and human beings. And
the bodies of the men and women, struck down by the gas, are to
serve as food for these demons in human form.
“This is the most horrible, blood-curdling thing of all. Rendered
unconscious by the gas, the victims remain in a state of suspended
animation indefinitely, exactly as do grubs, spiders and insects when
stung by certain species of wasps and placed in their nests to
provide food for their young. Stacked in great storage vaults these
breathing, living, but paralyzed human beings are kept, and as
needed, are taken out.
“Already they have a supply on hand sufficient to last them for
over a year. Some of the Cyclops company are still preserved; there
are over three hundred from the Chiriqui, hundreds from other ships,
and the entire crew of the McCracken.
“All these things I learned little by little, and mainly through a
friend, for marvelous as it may seem, I have a friend—if friend he
can be called, a miserable, trembling, terrified male, who, doomed to
death, sought to escape his fate and sought refuge with me,
dreading my presence less than his doom, and hoping that such a
feared and almost reverenced being as myself might protect him. For
two months he has been my companion, but he cannot eat anything
but meat and the supply of meat upon the ship is getting low, and
sooner or later he must succumb. And the women, maddened at his
escape from their clutches, though not yet daring to approach too
closely to me, are getting bolder. Some time, at some unguarded
moment, they will find the poor fellow alone and will fall upon him.
And in his terror, in an effort to buy his life, he will, I know, reveal to
them that I am but an ordinary mortal, a man who eats and drinks
and who survived the gas by mechanical and not supernatural
means. But I will not be taken alive by these fearful female
cannibals. When the time comes, as I know it will, I will blow my
brains out, and though they may devour my body they will not rend
me alive. No more ships have been brought in here since the
McCracken was captured. But this I know is due to the fact that all
the energies of these creatures are being devoted to building
additional air machines. This work goes on in a vast cavern beyond
the city where tremendous forces, furnaces with heat beyond human
conception and machines of which we know nothing, are controlled
by the internal steam, the radiant energy and the magnetic powers of
the earth’s core.
“And now, again let me implore any and all who may hear my
words to give close attention to what I say, for here again is a means
by which humanity may combat and destroy these ghastly, gigantic
cannibals. The spherical air-machines are helpless from above.
Their magnetic or electrical forces extend only downwards. The
gasses they throw out are heavier than air and descend but cannot
ascend, and by means of swift planes, huge bombs and machine
guns, the things can be easily destroyed. And they cannot travel
without throwing off the dazzling green light. Only when motionless
are they dark. And so they will offer easy marks and can be readily
detected. So, I beseech you who may hear, that the governments
are notified and warned and that a fleet or many fleets of airplanes
properly equipped patrol the seas, and at first sight of one of the
green meteors rise above it and utterly destroy it without mercy.
 “Wait! I hear a terrified scream.... I am back again at the
transmitter. It was the fellow who has been with me. Poor devil! He
has met his fate, but after all it was the custom of his people, and,
moreover, he would have starved to death in a few days. For that
matter I, too, face starvation. The ship’s stock is running low; all the
food upon the McCracken was destroyed in cutting up that vessel,
and unless another ship is captured I will have no food after two
weeks more. What a strange thought! How terrible an idea! That the
awful fate of hundreds of my fellows would be my salvation! But I will
never live to die from hunger. I can hear the terrible screams of my
late companion on the deck outside. God! It is the end! The fellow
must have told the enraged females. His body has been torn to
shreds. With bloody hands and reeking lips they are rushing towards
the upper deck where I sit. They are here! This is my last word! God
grant that I have been heard! I am about to⸺”
Crashing in our ears came the report of a pistol.
The End

1 The message as it came in, was halting, and interrupted, with many unintelligible
words and repetitions, as if the sender were laboring under an intense strain or was an
amateur. For the sake of clarity and continuity, the communication has been edited and
filled in, but not altered in any detail.
2 The metropolitan papers reported the meteor on the eighteenth and stated it was
observed by those on the Chiriqui on the evening of the seventeenth, but it must be
remembered that the Chiriqui was in the western Pacific and hence had gained a day
in time.

Transcriber’s Note: This story appeared in the July 1927 issue of

Amazing Stories Magazine.
*** END OF THE PROJECT GUTENBERG EBOOK A VOICE FROM
THE INNER WORLD ***

Updated editions will replace the previous one—the old editions will
be renamed.

Creating the works from print editions not protected by U.S.

copyright law means that no one owns a United States copyright in
these works, so the Foundation (and you!) can copy and distribute it
in the United States without permission and without paying copyright
royalties. Special rules, set forth in the General Terms of Use part of
this license, apply to copying and distributing Project Gutenberg™
electronic works to protect the PROJECT GUTENBERG™ concept
and trademark. Project Gutenberg is a registered trademark, and
may not be used if you charge for an eBook, except by following the
terms of the trademark license, including paying royalties for use of
the Project Gutenberg trademark. If you do not charge anything for
copies of this eBook, complying with the trademark license is very
easy. You may use this eBook for nearly any purpose such as
creation of derivative works, reports, performances and research.
Project Gutenberg eBooks may be modified and printed and given
away—you may do practically ANYTHING in the United States with
eBooks not protected by U.S. copyright law. Redistribution is subject
to the trademark license, especially commercial redistribution.

START: FULL LICENSE

THE FULL PROJECT GUTENBERG LICENSE
 PLEASE READ THIS BEFORE YOU DISTRIBUTE OR USE THIS WORK

To protect the Project Gutenberg™ mission of promoting the free

distribution of electronic works, by using or distributing this work (or
any other work associated in any way with the phrase “Project
Gutenberg”), you agree to comply with all the terms of the Full
Project Gutenberg™ License available with this file or online at
www.gutenberg.org/license.

Section 1. General Terms of Use and

Redistributing Project Gutenberg™
electronic works
1.A. By reading or using any part of this Project Gutenberg™
electronic work, you indicate that you have read, understand, agree
to and accept all the terms of this license and intellectual property
(trademark/copyright) agreement. If you do not agree to abide by all
the terms of this agreement, you must cease using and return or
destroy all copies of Project Gutenberg™ electronic works in your
possession. If you paid a fee for obtaining a copy of or access to a
Project Gutenberg™ electronic work and you do not agree to be
bound by the terms of this agreement, you may obtain a refund from
the person or entity to whom you paid the fee as set forth in
paragraph 1.E.8.

1.B. “Project Gutenberg” is a registered trademark. It may only be

used on or associated in any way with an electronic work by people
who agree to be bound by the terms of this agreement. There are a
few things that you can do with most Project Gutenberg™ electronic
works even without complying with the full terms of this agreement.
See paragraph 1.C below. There are a lot of things you can do with
Project Gutenberg™ electronic works if you follow the terms of this
agreement and help preserve free future access to Project
Gutenberg™ electronic works. See paragraph 1.E below.
1.C. The Project Gutenberg Literary Archive Foundation (“the
Foundation” or PGLAF), owns a compilation copyright in the
collection of Project Gutenberg™ electronic works. Nearly all the
individual works in the collection are in the public domain in the
United States. If an individual work is unprotected by copyright law in
the United States and you are located in the United States, we do
not claim a right to prevent you from copying, distributing,
performing, displaying or creating derivative works based on the
work as long as all references to Project Gutenberg are removed. Of
course, we hope that you will support the Project Gutenberg™
mission of promoting free access to electronic works by freely
sharing Project Gutenberg™ works in compliance with the terms of
this agreement for keeping the Project Gutenberg™ name
associated with the work. You can easily comply with the terms of
this agreement by keeping this work in the same format with its
attached full Project Gutenberg™ License when you share it without
charge with others.

1.D. The copyright laws of the place where you are located also
govern what you can do with this work. Copyright laws in most
countries are in a constant state of change. If you are outside the
United States, check the laws of your country in addition to the terms
of this agreement before downloading, copying, displaying,
performing, distributing or creating derivative works based on this
work or any other Project Gutenberg™ work. The Foundation makes
no representations concerning the copyright status of any work in
any country other than the United States.

1.E. Unless you have removed all references to Project Gutenberg:

1.E.1. The following sentence, with active links to, or other

immediate access to, the full Project Gutenberg™ License must
appear prominently whenever any copy of a Project Gutenberg™
work (any work on which the phrase “Project Gutenberg” appears, or
with which the phrase “Project Gutenberg” is associated) is
accessed, displayed, performed, viewed, copied or distributed:
 This eBook is for the use of anyone anywhere in the United
States and most other parts of the world at no cost and with
almost no restrictions whatsoever. You may copy it, give it
away or re-use it under the terms of the Project Gutenberg
License included with this eBook or online at
www.gutenberg.org. If you are not located in the United
States, you will have to check the laws of the country where
you are located before using this eBook.

1.E.2. If an individual Project Gutenberg™ electronic work is derived

from texts not protected by U.S. copyright law (does not contain a
notice indicating that it is posted with permission of the copyright
holder), the work can be copied and distributed to anyone in the
United States without paying any fees or charges. If you are
redistributing or providing access to a work with the phrase “Project
Gutenberg” associated with or appearing on the work, you must
comply either with the requirements of paragraphs 1.E.1 through
1.E.7 or obtain permission for the use of the work and the Project
Gutenberg™ trademark as set forth in paragraphs 1.E.8 or 1.E.9.

1.E.3. If an individual Project Gutenberg™ electronic work is posted

with the permission of the copyright holder, your use and distribution
must comply with both paragraphs 1.E.1 through 1.E.7 and any
additional terms imposed by the copyright holder. Additional terms
will be linked to the Project Gutenberg™ License for all works posted
with the permission of the copyright holder found at the beginning of
this work.

1.E.4. Do not unlink or detach or remove the full Project

Gutenberg™ License terms from this work, or any files containing a
part of this work or any other work associated with Project
Gutenberg™.

1.E.5. Do not copy, display, perform, distribute or redistribute this

electronic work, or any part of this electronic work, without
prominently displaying the sentence set forth in paragraph 1.E.1 with
active links or immediate access to the full terms of the Project
Gutenberg™ License.

1.E.6. You may convert to and distribute this work in any binary,
compressed, marked up, nonproprietary or proprietary form,
including any word processing or hypertext form. However, if you
provide access to or distribute copies of a Project Gutenberg™ work
in a format other than “Plain Vanilla ASCII” or other format used in
the official version posted on the official Project Gutenberg™ website
(www.gutenberg.org), you must, at no additional cost, fee or expense
to the user, provide a copy, a means of exporting a copy, or a means
of obtaining a copy upon request, of the work in its original “Plain
Vanilla ASCII” or other form. Any alternate format must include the
full Project Gutenberg™ License as specified in paragraph 1.E.1.

1.E.7. Do not charge a fee for access to, viewing, displaying,

performing, copying or distributing any Project Gutenberg™ works
unless you comply with paragraph 1.E.8 or 1.E.9.

1.E.8. You may charge a reasonable fee for copies of or providing

access to or distributing Project Gutenberg™ electronic works
provided that:

• You pay a royalty fee of 20% of the gross profits you derive from
the use of Project Gutenberg™ works calculated using the
method you already use to calculate your applicable taxes. The
fee is owed to the owner of the Project Gutenberg™ trademark,
but he has agreed to donate royalties under this paragraph to
the Project Gutenberg Literary Archive Foundation. Royalty
payments must be paid within 60 days following each date on
which you prepare (or are legally required to prepare) your
periodic tax returns. Royalty payments should be clearly marked
as such and sent to the Project Gutenberg Literary Archive
Foundation at the address specified in Section 4, “Information
about donations to the Project Gutenberg Literary Archive
Foundation.”
 • You provide a full refund of any money paid by a user who
notifies you in writing (or by e-mail) within 30 days of receipt that
s/he does not agree to the terms of the full Project Gutenberg™
License. You must require such a user to return or destroy all
copies of the works possessed in a physical medium and
discontinue all use of and all access to other copies of Project
Gutenberg™ works.

• You provide, in accordance with paragraph 1.F.3, a full refund of

any money paid for a work or a replacement copy, if a defect in
the electronic work is discovered and reported to you within 90
days of receipt of the work.

• You comply with all other terms of this agreement for free
distribution of Project Gutenberg™ works.

1.E.9. If you wish to charge a fee or distribute a Project Gutenberg™

electronic work or group of works on different terms than are set
forth in this agreement, you must obtain permission in writing from
the Project Gutenberg Literary Archive Foundation, the manager of
the Project Gutenberg™ trademark. Contact the Foundation as set
forth in Section 3 below.

1.F.

1.F.1. Project Gutenberg volunteers and employees expend

considerable effort to identify, do copyright research on, transcribe
and proofread works not protected by U.S. copyright law in creating
the Project Gutenberg™ collection. Despite these efforts, Project
Gutenberg™ electronic works, and the medium on which they may
be stored, may contain “Defects,” such as, but not limited to,
incomplete, inaccurate or corrupt data, transcription errors, a
copyright or other intellectual property infringement, a defective or
damaged disk or other medium, a computer virus, or computer
codes that damage or cannot be read by your equipment.

1.F.2. LIMITED WARRANTY, DISCLAIMER OF DAMAGES - Except

for the “Right of Replacement or Refund” described in paragraph
1.F.3, the Project Gutenberg Literary Archive Foundation, the owner
of the Project Gutenberg™ trademark, and any other party
distributing a Project Gutenberg™ electronic work under this
agreement, disclaim all liability to you for damages, costs and
expenses, including legal fees. YOU AGREE THAT YOU HAVE NO
REMEDIES FOR NEGLIGENCE, STRICT LIABILITY, BREACH OF
WARRANTY OR BREACH OF CONTRACT EXCEPT THOSE
PROVIDED IN PARAGRAPH 1.F.3. YOU AGREE THAT THE
FOUNDATION, THE TRADEMARK OWNER, AND ANY
DISTRIBUTOR UNDER THIS AGREEMENT WILL NOT BE LIABLE
TO YOU FOR ACTUAL, DIRECT, INDIRECT, CONSEQUENTIAL,
PUNITIVE OR INCIDENTAL DAMAGES EVEN IF YOU GIVE
NOTICE OF THE POSSIBILITY OF SUCH DAMAGE.

1.F.3. LIMITED RIGHT OF REPLACEMENT OR REFUND - If you

discover a defect in this electronic work within 90 days of receiving it,
you can receive a refund of the money (if any) you paid for it by
sending a written explanation to the person you received the work
from. If you received the work on a physical medium, you must
return the medium with your written explanation. The person or entity
that provided you with the defective work may elect to provide a
replacement copy in lieu of a refund. If you received the work
electronically, the person or entity providing it to you may choose to
give you a second opportunity to receive the work electronically in
lieu of a refund. If the second copy is also defective, you may
demand a refund in writing without further opportunities to fix the
problem.

1.F.4. Except for the limited right of replacement or refund set forth in
paragraph 1.F.3, this work is provided to you ‘AS-IS’, WITH NO
OTHER WARRANTIES OF ANY KIND, EXPRESS OR IMPLIED,
INCLUDING BUT NOT LIMITED TO WARRANTIES OF
MERCHANTABILITY OR FITNESS FOR ANY PURPOSE.

1.F.5. Some states do not allow disclaimers of certain implied

warranties or the exclusion or limitation of certain types of damages.
If any disclaimer or limitation set forth in this agreement violates the
law of the state applicable to this agreement, the agreement shall be
interpreted to make the maximum disclaimer or limitation permitted
by the applicable state law. The invalidity or unenforceability of any
provision of this agreement shall not void the remaining provisions.

1.F.6. INDEMNITY - You agree to indemnify and hold the

Foundation, the trademark owner, any agent or employee of the
Foundation, anyone providing copies of Project Gutenberg™
electronic works in accordance with this agreement, and any
volunteers associated with the production, promotion and distribution
of Project Gutenberg™ electronic works, harmless from all liability,
costs and expenses, including legal fees, that arise directly or
indirectly from any of the following which you do or cause to occur:
(a) distribution of this or any Project Gutenberg™ work, (b)
alteration, modification, or additions or deletions to any Project
Gutenberg™ work, and (c) any Defect you cause.

Section 2. Information about the Mission of

Project Gutenberg™
Project Gutenberg™ is synonymous with the free distribution of
electronic works in formats readable by the widest variety of
computers including obsolete, old, middle-aged and new computers.
It exists because of the efforts of hundreds of volunteers and
donations from people in all walks of life.

Volunteers and financial support to provide volunteers with the

assistance they need are critical to reaching Project Gutenberg™’s
goals and ensuring that the Project Gutenberg™ collection will
remain freely available for generations to come. In 2001, the Project
Gutenberg Literary Archive Foundation was created to provide a
secure and permanent future for Project Gutenberg™ and future
generations. To learn more about the Project Gutenberg Literary
Archive Foundation and how your efforts and donations can help,
see Sections 3 and 4 and the Foundation information page at
www.gutenberg.org.
 Section 3. Information about the Project
Gutenberg Literary Archive Foundation
The Project Gutenberg Literary Archive Foundation is a non-profit
501(c)(3) educational corporation organized under the laws of the
state of Mississippi and granted tax exempt status by the Internal
Revenue Service. The Foundation’s EIN or federal tax identification
number is 64-6221541. Contributions to the Project Gutenberg
Literary Archive Foundation are tax deductible to the full extent
permitted by U.S. federal laws and your state’s laws.

The Foundation’s business office is located at 809 North 1500 West,

Salt Lake City, UT 84116, (801) 596-1887. Email contact links and up
to date contact information can be found at the Foundation’s website
and official page at www.gutenberg.org/contact

Section 4. Information about Donations to

the Project Gutenberg Literary Archive
Foundation
Project Gutenberg™ depends upon and cannot survive without
widespread public support and donations to carry out its mission of
increasing the number of public domain and licensed works that can
be freely distributed in machine-readable form accessible by the
widest array of equipment including outdated equipment. Many small
donations ($1 to $5,000) are particularly important to maintaining tax
exempt status with the IRS.

The Foundation is committed to complying with the laws regulating

charities and charitable donations in all 50 states of the United
States. Compliance requirements are not uniform and it takes a
considerable effort, much paperwork and many fees to meet and
keep up with these requirements. We do not solicit donations in
locations where we have not received written confirmation of