Communication in Swarm Robotics

COMMUNICATION IN SWARM ROBOTICS
Anuradhika Pilli!
Eastern Michigan University!
900 Oakwood St, Ypsilanti, MI - 48197; Ph: +1 325-280-4643!
e-mail: apilli@emich.edu!
!
!
ABSTRACT
In this paper, the communication techniques in Swarm
Robotics using Ant Colony Optimization (ACO) are
discussed. This also includes the detailed explanation of the
topics: swarm robotics, Ant Colony Optimization and the
communication techniques in swarm robotics. Furthermore,
different studies and experiments done on swarm robotics by
few researchers are analyzed, compared and contrasted that
helped to determine the best technique for communication
between various swarm agents in swarm robotics. Also the
importance of this study is acknowledged by knowing few of
swarm robotics applications.
Keywords
Swarm Robotics, Stigmergy, Swarm agents, Ant Colony
Optimization (ACO), Eye-bots, Foot-bots, Foraging
behavior, Pheromone trail
1.INTRODUCTION
Since the past few decades communication between swarms
of insects has drawn interest of many. Not just
communication but also efficient means of navigation
through various broadcasting methods to obtain optimized
path has been a topic of great concern. Inspired by biological
swarms of insects, the concept of Swarm Robotics came into
existence.
The foraging framework plays a crucial role in Swarm
Robotics. Swarm robotics is a concept where group of
"robots" or "swarm agents" work together to accomplish a
complex task. Swarm robotics is a bio-inspired topic derived
from the well known algorithm, "Ant Colony Optimization
(ACO)." The concept of ACO is simple. Ants follow a
systematic and optimized method to find food and get it back
to their nests. Every ant upon finding food leaves a
pheromone trail (a volatile scented track) for other ants that
are wandering randomly in search of food, to find food. The
track becomes stronger when more ants pass on the same
trail, meaning the optimized path. This stigmergy [5] (a
mechanism of indirect coordination between agents or
actions) and foraging behavior [2] of ants introduced a
complete new dimension to Swarm Robotics.
For a successful foraging behavior [2], good communication
[5] between swarm agents is essential. The different methods
that can be employed in Swarm Robotics for communication
are telecommunication media like bluetooth, wireless LAN,
radio waves, light and other similar media transmissions
or communication via the environment(stigmergy). There has
always been an intense debate on which is the best
broadcasting method in Swarm Robotics, to find optimized
path.
In telecommunication, the information is exchanged between
every other swarm in various ways. Some of the popular
means of telecommunication are information exchange
via Bluetooth - this communication occurs when the agents
are in line of sight, communication using light sensors - the
agents detect light waves emitted by other robots and,
communication using radio waves - these waves enable
communication in any kind of environment. All these
telecommunication methods usually need an external robot or
computational device monitoring all the agents in order for
them to function effectively.
On the other hand, communication via environment is a
method where information transmission is done trying to
purely replicate the stigmergy [5] behavior of ants by
utilizing any kind of volatile chemical track as a source of
information. Using this method, the agents need not rely on
an external device for guidance which means they are self-
learning and also self-organizing which is the main basic idea
of Swarm Intelligence. Similar to ants that walk on ground
and communicate effortlessly on ground, stigmergy [5] in
swarm robotics is also limited to particular environment
types.
2.RELATED WORK STUDY
According to Fujisawa et al. (2014), telecommunication is
the general means of information transmission in Swarm
Robotics. Fujisawa et al. (2014) and Mayet et al. (2010)
claim that, the use of chemical compounds(stigmergy) in
foraging technique, that is, communication via environment
shows better results compared to the telecommunication
media. This paves a path to a controversial discussion as few
others like Ducatelle et al. (2011) strongly concur with the
idea of using telecommunication in Swarm Robotics.
2.1.Work of Fujisawa et al.
Fujisawa et al. (2014) and Mayet et al. (2010) support their
statement by demonstrating their experiments with robotic
swarms. In their experiment, Fujisawa et al. (2014) use
ethanol as a source of information which gets stronger when
many agents pass over it. This ethanol [3] also disappears
with time as it is volatile. Fujisawa et al. (2014) performed
three various tests on the robots.
Fujisawa et al. (2014) first performed a simulation test on
various parameters of scalibility like the presence or absence
of pheromone communication [3], number of robots and size
of the environment keeping all other parameters same in the
experiment. They found that performance is directly
proportional to number of robots and inversely proportional
to the environment size in both conditions, that is, presence
or absence of pheromone communication [3].
!1

The second test was a real setting in which Fujisawa et al.
used four sets of sensors: (1) touch sensors to detect food,
(2) phototransistors to detect nest, (3) alchol sensors to detect
pheromone trail and (4) an internal timer to check for
timeouts (timer starts when a robot loses its track and after 4
seconds if it does not find the track it changes its internal
state). The parameters such as size of the environment, food,
nest, and the duration of the experiment (20 mins) being the
same as in simulation experiment. The swarm size varied
from 1 - 10. The results were much similar to the simulation
experiment where the only difference is that the critical mass
behavior with pheromone communication [3] surpasses the
one without the use of pheromone communication [3].
The third test was the transport test. Setup was similar to the
real experimental setup. The start time with any number of
robots in the experiment was almost the same while the
execution time created a difference among them. The
performance also enhances with the use or pheromone
communication [3] rather than without it. Based on these
three tests, Fujisawa et al. (2014) conclude that
communication in swarms with pheromone is superior to the
communication without pheromone.
2.2.Work of Mayet et al.
Mayet et al. (2010) support the idea of communication via
environment by demonstrating their experiment in which
they use a single robot as a sample and also conduct a
simulation test. In their experiment, the robot consisted of red
and green LEDs, eight different infrared transiting and
receiving sensors, color CMOS camera and a phosphorescent
paint emitter [4]. The nest is red LED and the food source is
blue LED each of which is covered with transparent plastic.
A red light source had also been placed at a corner that acts
as “artificial sun”. This sun compass can be detected by the
robots in all directions.
Mayet et al. (2010) conducted a simulation test using this
setup. The red and green LED lights on robot show the state
of the robot. Red light indicates that the robot is in search of
food and the green indicates that it is going back to nest with
food. The camera detects the different colored LEDs based
on which the robot decides its direction of movement. The
test was conducted on different robot size with and without
using pheromone trail.
Initial state of robots is in red and upon finding the food
source it changes it state to green and search for the sun
compass to learn it’s location and depict the nest location
from there. Using pheromone trail - while coming back to the
nest with food, it releases the phosphorescent paint [4] which
disappears in time. This paint is identified by the UVLED
sensor built in the robot. The robot moves in the direction
with maximum intensity of paint. The results of this
simulation test conducted several times with varying robot
sizes showed that the task is performed efficiently with more
number of agents and using pheromone trail rather than
without using pheromone trail. In the real experiment,
Mayet et al. (2010) used a single robot with similar
functionalities as in the simulation test and the robot
navigates reliably between the food source and the nest.
2.3.Work of Ducatelle et al.
Ducatelle et al. (2011) present their point of view by
demonstrating their experiment in which they use
heterogenous robotic swarms. Every homogenous group has
its own work functionality. Initially, Ducatelle et al. (2011)
setup two different sub-swarms, mainly foot-bots [1]
and eye-bots [1] to investigate the local interactions and
mutual adaption among them. The foot-bots [1] are used for
navigation between the source and target in an indoor area.
The eye-bots [1] are flying robots that attach to
the ceiling. Eye-bots [1] first explore the indoor area and then
selecting a suitable place on the ceiling to sit on, from where
they could have a clear view of the whole arena so as to
direct the foot-bots [1] in navigation without any collision
with obstacles or other agents. The eye-bots [1] act as a
pheromone for the foot-bots [1]. The crucial task in their
study is how the eye-bots [1] need to be updated and how
could they give efficient direction to the foot-bots [1]. This
type of communication calls for synchronization and mutual
understanding between the different sub-swarms.
Ducatelle et al. (2011) further say that their system features
minimal information exchange, purely broadcast-based local
interactions based on short-range radio signals and simple
visual cues. As this approach is much similar to the
traditional telecommunication approach [5], Ducatelle et al.
(2011) say that the system is likely to be robust, adaptive
and scalable. Ducatelle et al. (2011) start off with the
description of the robots. Then they study the eye-bots [1].
Later they investigate the self-organized behavior of the
system for producing optimized results. And finally observe
how the eye-bots [1] adapt efficiently to the exact locations
on ceiling to give proper directions to the foot-bots [1]. The
experimental results show that the heterogenous swarms
work effectively and are quite robust as the synchronization
between them is flawless. As a result of this synchronization,
there are less collisions.
3.DISCUSSION AND CONCLUSION
The first two approaches are much similar. In both the cases
it had been proved that swarm communication is productive
with the use of pheromone trail rather than without using it.
Not only pheromone trail, but also the robot size was also
proportional to the productiveness. In the third approach,
communication was done using telecommunication, but with
the same idea of replicating the ACO algorithm. The eye-bots
[1] acted as a trail for the foot-bots [1]. With sterling
synchronization between foot-bots [1] and eye-bots [1], the
task can be fulfilled as outstandingly as in the first two
approaches.
Based on the three approaches, the use of chemical based
media or the pheromone communication [3] serves better.
This is for two reasons - Firstly, the agents are self-learning
which is the main basic concept of stigmergy [5] in Swarm
Intelligence, that is missing in the third approach that is
master and slave approach of eye-bots [1] and foot-bots [1]
mentioned above. The second reason is that, with the
increase in size of foot-bots [1] and the arena, the number of
eye-bots [1] should also be increased to work effectively and
the scope of each eye-bot may be limited. None-the-less both
the approaches serve fairly well in task accomplishment.
4.FUTURE STUDY
Basic drawback of pheromone based communication is that it
works only on ground of specific types and in
telecommunication it is the external control system. Not
!2

limiting it to the ground, researchers can take the inspiration
of jet planes and utilize the smoke concept that fades in a
while as the pheromone trail in pheromone communication
[3]. Even on water the oil leaking vessels can be a source of
inspiration. The agents can release some de-toxic chemicals
on or in the water that may serve as the pheromone trail. A
further advanced study in both approaches and a combination
of those could be of major enhancement in Swarm
Intelligence concept which would serve complex tasks much
easily than being discussed now.
5.WHY THIS STUDY IS IMPORTANT?
There are many applications of swarm robotics. One major
application is that it can be used in military forces where
much blood shed can be reduced. It is used in crowd
simulation, crowd control, path optimization and in
complexity optimization. All these applications are not
possible if there is no proper communication between the
swarms. Hence an effective communication technique is
needed to work wonders in accomplishing complex tasks.
6.REFERENCES
1. Ducatelle, F., Di Caro, G. A., Pinciroli, C.,
& Gambardella, L. M. (2011). Self-organized
cooperation between robotic swarms. In Swarm Intell
(5): (pp. 73–96). Austria: Springer Science + Business
Media.
2. Edelen, M. R. (2003). Thesis on Swarm intelligence and
stigmergy: Robotic implementation of foraging
behavior. http://drum.lib.umd.edu/bitstream/1903/107/1/
dissertation.pdf
3. Fujisawa, R., Dobata, S., Sugawara, K., & Matsuno, F.
(2014). Designing pheromone communication in swarm
robotics: Group foraging behavior mediated by chemical
substance. In Swarm Intell (8): (pp. 227–246).
Newyork: Springer Science + Business Media.
4. Mayet, R., Roberz, J., Schmickl, T., & Crailsheim, K.
(2010). Antbots: A feasible visual emulation of
pheromone trails for Swarm Robots. In Dorigo, M. et al.
(Eds.), Ants (pp. 84-94). Berlin Heidelberg: Springer-
Verlag.
5. Swarm communication in Jasmine Swarm robotic
platform. http://www.swarmrobot.org/
Communication.html
!
!
!
!
!
!3

Communication in Swarm Robotics

Recommended

More Related Content

What's hot (20)

Similar to Communication in Swarm Robotics (20)

Communication in Swarm Robotics