Conference PaperPDF Available

Collision-Induced "Priority Rule" Governs Efficiency of Pheromone-Communicating Swarm Robots

Authors:

Abstract and Figures

The recruiting system in foraging ant colonies is a typical example of swarm intelligence. The system is underpinned by the use of volatile pheromones which form a trail connecting from nest to food. We have incorporated this property into the behavior of the swarm of real robots. Because the trail is narrow, avoiding overcrowding on the trail, as well as in the environment, is a critical issue in maintaining efficiency of the swarm behavior. In this paper, we studied how "priority rule, a behavioral rule under which a robot is given priority over the other robot in collision, affect the group-foraging performance of pheromone-mediated swarm robots. Using real robot experiments, we found that the alteration in the priority rules can have substantial effects on the group-foraging performance. Our results highlight the importance of im-plementing "fine-tuningalgorithms to improve the performance of com-plex swarm systems.
Content may be subject to copyright.
Collision-Induced “Priority Rule” Governs
Efficiency of Pheromone-Communicating Swarm
Robots
Ryusuke Fujisawa
1
, Shigeto Dobata
2
, Yuuta Sasaki
1
, Riku Takisawa
1
, and
Fumitoshi Matsuno
3
1
Hachinohe Institute of Technology, Aomori, Japan
swarm.ant@gmail.com
2
University of the Ryukyus, Okinawa, Japan
dobatan@gmail.com
3
Kyoto University, Kyoto, Japan
matsuno@me.kyoto-u.ac.jp
Abstract. The recruiting system in foraging ant colonies is a typical
example of swarm intelligence. The system is underpinned by the use of
volatile pheromones which form a trail connecting from nest to food. We
have incorp orated this property into the behavior of the swarm of real
rob ots. Because the trail is narrow, avoiding overcrowding on the trail,
as well as in the environment, is a critical issue in maintaining efficiency
of the swarm behavior. In this paper, we studied how “priority rule,
a behavioral rule under which a robot is given priority over the other
rob ot in collision, affect the group-foraging performance of pheromone-
mediated swarm robots. Using real robot experiments, we found that
the alteration in the priority rules can have substantial effects on the
group-foraging performance. Our results highlight the importance of im-
plementing “fine-tuningalgorithms to improve the performance of com-
plex swarm systems.
1 Introduction
1.1 Swarm and Pheromone Communication in Ants
We define the term “swarm as a distributed autonomous system, in which
each individual acts autonomously only according to local information in the
given environment without any global information [1]. A global-level behaviour
of the swarm emerges through frequent local interactions among individuals. This
emergence has two remarkable properties: robustness through which a swarm can
adapt to changes in its internal states, and flexibility through which it can adapt
to changes in its external states (e.g., the environment) [2].
Individuals of social insects communicate with one another to form a swarm,
known as a colony. Among others, ants and termites are known to form especially
complex societies, and their formation is often facilitated by using pheromones
[3].
2
A pheromone is a chemical or set of chemicals produced by a living organism
that transmits information to other members of the same species [3]. In this
paper, we focus on foraging behaviour of ants using a pheromone. When an ant
finds food and brings it back to the nest, it secretes a pheromone that forms a
trail. Other ants trace the pheromone trail and can reach the food. An ant stops
to lay down the pheromone trail when it cannot find the food. Consequently,
the pheromone trail volatilizes (or diffuses) into the environment, making the
previous information meaningless to the ants. The algorithm of this indirect
recruiting system is a simple but advanced communication method. Indeed, “Ant
Colony Optimization” [4] was inspired by the above-explained mechanism.
1.2 Related Studies
In swarm robotics, several studies using real or virtual pheromones have been
reported previously. Sugawara et al. [5] and Garnier et al. [6] achieved the forag-
ing behaviour of ants using a swarm of robots and a virtual pheromone (with a
projector and screen). These studies represented a well-conceived measurement
system. Pheromone diffusion is an important factor in real pheromone studies,
and implementing it is a very difficult problem. To adjust the duration of the
pheromone signal, the pheromone should be, for example, mixed with other sub-
stance(s) to change concentration of the pheromone. In addition, there are few
high-performance chemical sensors available; these difficulties have led the re-
searchers to use a virtual pheromone. Shimoyama et al. [7] achieved pheromone
tracking behaviour using a “chimeric” system implementing real insect anten-
nae and pheromone, but they did not consider swarm behaviour, and the use
of biomaterials is usually difficult for swarm robotics. Purnamadjaja et al. [8]
studied swarm robots that communicate using two real chemical substances. The
latter regulates a gas sensor in a refined way. However, only one robot secretes
the pheromone, so this system allows for only one-sided communication. In the
previous study, we focused on the effect of concentration of the pheromone on
the performance of group-foraging [9].
1.3 Priority on the Pheromone Trail
A critical issue in managing the swarm system is how to control for overcrowding.
The negative impact of overcrowding on swarm performance becomes crucial
especially in a larger swarm size (Kriger et. al [10]), even though direct and
thus potentially damaging collision itself could be avoided in some way (e.g.
[11]). In real ant colonies, the existence of priority rules like the one in the
present study has been demonstrated ( [12–14]). Also in ants, the priority rules
are induced when two ants collide; such a situation most likely occurs on the
pheromone trail. Moreover, some studies showed experimentally that the rule
contributes to efficient transportation ( [13, 14]). These findings suggest that
a “fine-tuning” of the trail-foraging system have played a crucial role in the
adaptive evolution of the ant colony systems. Implementing similar condition-
dependent fine-tuning may sometimes improve the p erformance of robot swarm,
3
as well as the swarm algorithm itself. Although collision-avoiding algorithms are
routinely implemented in swarm robotics, to our knowledge, previous studies
have not assessed directly how they could contribute to the group performance.
In this experimental study, we implemented collision-induced “priority rules” in
our trail pheromone-based robot swarm and assessed the effects of the alteration
in the rules on the group-foraging performance.
2 Swarm Behaviour Algorithm
2.1 Basic Algorithm
Fig. 1. State transition rule for swarm
behaviour
We assumed a finite experimental field
with only agents, a food and a nest.
The task of the swarm is to find
a food in the field, just like forag-
ing ant colonies. To accomplish this
task, an agent can attract other agents
indirectly using a pheromone secre-
tion. We used the swarm behaviour
algorithm for group foraging using
pheromone communication [9], which
is described as a deterministic finite
automaton. The algorithm shown in
Fig. 1 can read as follows: We de-
fined the following three states S
i
(i =
1, 2, 3), six perceptual cues (stimuli)
P
j
(j = 1, · · · , 6) and three effector cues
(actions) E
k
(k = 1, 2, 3). The agent whose state is S
i
selects the action E
k
(i = k). If the agent in state S
i
detects the perceptual cue P
j
, the state of the
agent transits to S
k
(Fig. 1). The details of S
i
, P
j
, and E
k
are as follows. S
i
:
S
1
, Search (Agent does not have any information of the food); S
2
, Attraction
(Agent has the location information of the fo od); S
3
, Tracking (Agent has only
the direction information of the location of the food). P
j
: P
1
, Contact with food;
P
2
, Arrival at nest; P
3
, Presence of pheromone; P
4
, Timeout; P
5
, Collision with
object; P
6
, Completion of collision processing. E
k
: E
1
, Random walk; E
2
, Secrete
pheromone along the nest direction; E
3
, Follow the pheromone path toward the
food. We assumed that all agents can detect the direction of the nest, which we
feel is a reasonable assumption.
2.2 Collision Processing and “Priority Rules”
We propose the following interaction rules. When colliding, an robot is designed
to take one of the following two reactions, “Stay” and “Back”, according to its
current internal states(S
i
) that are perceived by the robot itself. During the re-
action “Stay”, the robot stops moving for a given time (1[sec]) and then regains
4
Table 1. Description of the “priority rules”
Patterns BSB BBS BBB SBB SBS SSB BSS SSS
S
1
Back Back Back Stop Stop Stop Back Stop
S
2
Stop Back Back Back Back Stop Stop Stop
S
3
Back Stop Back Back Stop Back Stop Stop
Actions after collision are described. “Backmeans disengagement from collision point,
keeping the head in the same direction as before. “Stay means temporary stop. See
the main text for details.
the same internal state as it took when colliding. During the reaction “Back”,
the robot moves directly backward (10[cm]) from its position, and then regains
the same internal state. This behavior was originally implemented to avoid po-
tential collision-induced congestion [1]. The rob ot can take different reactions
depending on each of the three internal states (S
1
-S
3
), so that there arise eight
possible combinations of reaction rules (Table 1). The term “priority rules” is
most suitable when the combinations are NBS and NSB (N = S,B hereafter),
i.e., when the reaction rules are invoked on the pheromone trail. Nevertheless,
we hereafter use this term to indicate the total set of these combinations.
3 Hardware and Experimental Design
Fig. 2. Overview of a developed robot (AR-
GOS02)
We have developed robots which
can tune the duration of pheromone
activity [15]. The robot is shown
in Fig. 2. The robot is cylin-
der shape, diameter 150 [mm],
height 225 [mm], weight 1.45
[kg] and speed 0.1 [m/s]. Power
source is six series-connected
Ni-MH batteries (1.2 [V] 4500
[mAh]). The robot is con-
structed by four layers, and each
layer is supported by spacers.
At first (bottom) layer, there
are two DC motors, two alcohol
sensors, two micro pump and
discharge spout for alcohol. There are color sensors, full-color LEDs and push
switches for detecting collision at second layer. There are tank for alcohol and
six batteries at third layer. At fourth layer, there are system circuit board,
LED, and LCD indicator, Wireless USB, power circuit and nest sensor. Batter-
ies are connected on power circuit at fourth layer. The system has eight micro-
computers(CY8C29466), two masters and six slaves.
5
Fig. 3. Experimental field
for pheromone communicat-
ing robots
Figure 3 shows the experimental field. A nest
and a food are set on the opposite side. The
field size is 3600 [mm] × 3600 [mm], and is sur-
rounded by walls. Nest size is ϕ600 [mm], fo od
size is ϕ300 [mm], and food detection area is
ϕ600 [mm]. Initial positions of the robots were
set randomly on the field. The duration of ex-
periment was 20 [min]. We ran experiments ten
times for each of the eight priority rules (Table
1), and the number of foraging was recorded as
a measure of swarm performance. The effects of
swarm size and reaction rules on the swarm per-
formance were analyzed statistically using mul-
tiple Poisson regression.
In order to assess the effects of alteration of reaction rules at each state, we
compared 12 pairs of priority rules which differ only at one of the three states.
The effects of swarm size - 2 (denoted as x
1
), change of reaction rules from S to
B (denoted as x
2
; S=0, B=1), and the interaction of x
1
and x
2
, on the swarm
performance (denoted as y) were analyzed statistically using multiple regression
with the model: y = β
0
+ β
1
x
1
+ β
2
x
2
+ β
1×2
(x
1
× x
2
) + ϵ. The intercept β
0
and the slopes (β
1
, β
2
, β
1×2
) of the variables were tested against β = 0 using
t -tests.
4 Results
Figure 4, 5 shows some typical results of the experiments with ten robots. Under
the priority rule BSB (Fig. 4), the robots communicate with one another, indi-
cating that the priority rule is effective for foraging behavior. In stark contrast,
under the priority rule SSB (Fig. 5), we observed many clusters of colliding robots
on the experimental field (indicated by black circles in Fig. 5), which developed
during the run (ca. 5 [min] after the onset). In the cluster, the robots were stuck
at each other and could not disengage from it. This comparison highlights the
fact that an alteration of the reaction rule at only one state can drastically affect
the swarm performance.
Figure 6 shows the performance of the swarm (number of times of foraging per
20 min trial, n = 10 each) for each priority rule and the swarm size (i.e., number
of robots). For statistical analyses, we paired these priority rules based on the
presence/absence of rule B at each of the three states, so as to assess how rule B
affects the swarm performance at each of these states. Table 2 shows the results
of multiple regression analyses with the model described in Section 3.2. (Note
that x
2
and thus β
2
correspond to different states for different pairs.). Briefly,
for each comparison, β
0
corresponds to the extrapolated intercept at swarm
size = 2 under the reaction rule N = S, β
1
corresponds to the slope of swarm
performance against swarm size under the reaction rule N = S, β
2
measures how
the alteration of the reaction rule N from S to B affects the swarm performance
6
Fig. 4. Snapshots of swarm behavior under
the priority rules BSB
Fig. 5. Snapshots of swarm behavior under
the priority rules SSB
0
10
20
30
40
4 7 10
4 7 10
4 7 10
4 7 10
4 7 10
4 7 10
4 7 10
4 7 10
SSS SBS
SSB
SBB BSS
BBS
BSB BBB
Priority rules
Number of robots
Priority rules
Number of times of foraging
Fig. 6. Experimental result (Horizontal axis is number of robots, vertical axis is number
of foraging)
(measured at the extrapolated intercept at swarm size = 2), and β
1×2
measures
how the alteration of the reaction rule N from S to B affects the slope against
7
Table 2. Statistical analyses of the effects of reaction rule change at each stage assessed
by multiple regressions.
Test Comparison β
0
t β
1
t β
2
t β
1×2
t
Rule change
at S
1
SSS to BSS 3.57 7.224*** -0.267 -3.007** -2.07 -2.960** 0.867 6.911***
SBS to BBS 9.65 17.612*** -0.450 -4.573*** 0.117 0.151 4.283 30.778***
SSB to BSB 0.85 0.663 1.483 6.440*** 10.12 5.578*** 2.517 7.726***
SBB to BBB 11.75 13.826*** -0.417 -2.730** -0.900 -0.749 3.633 16.832***
Rule change
at S
2
SSS to SBS 3.57 7.616*** -0.267 -3.170** 6.083 9.185*** -0.183 -1.541
BSS to BBS 1.50 2.633* 0.600 5.863*** 8.267 10.259*** 3.233 22.341***
SSB to SBB 0.85 0.673 1.483 6.536*** 10.90 6.099*** -1.900 -5.920***
BSB to BBB 10.96 12.497*** 4.000 25.38*** -0.117 -0.094 -0.783 -3.514***
Rule change
at S
3
SSS to SSB 3.57 3.122** -0.267 -1.300 -2.716 -1.681 1.750 6.031***
BSS to BSB 1.50 1.964 0.600 4.374*** 9.467 8.765*** 3.400 17.527***
SBS to SBB 9.65 13.50*** -0.450 -3.505*** 2.100 2.077* 0.033 0.184
BBS to BBB 9.77 13.65*** 3.833 29.846*** 1.083 1.071 -0.617 -3.395**
The coefficients of the regression model βs are shown, which were tested against β = 0
using t -tests. *: p < 0.05, **: p < 0.01, ***: p < 0.001.
swarm size. Our aim is to assess the effects of the change of a reaction rule at each
state, so hereafter we consider only β
2
and β
1×2
. In the comparisons of reaction
rules at S
1
, the signs of β
2
varied among comparisons, whereas β
1×2
was always
positive (and significantly different from 0). In contrast, in the comparisons of
reaction rules at S
2
and S
3
, the signs of β
2
were mostly significantly positive (at
least not different from 0), whereas β
1×2
varied among comparisons.
5 Discussion and Conclusion
We found that the alteration of reaction rules from S to B had substantial
(mostly positive) effects on group-foraging performance. Moreover, statistical
analyses revealed that the alteration affects the group-foraging performance in
different ways depending on the states. The alterations at S
2
and S
3
improved the
group-foraging performance. This result has an intuitive interpretation: robots
take these states on the pheromone trail, and efficient use of the pheromone
trail is directly linked to foraging success. Meanwhile, these alterations showed
no constant effect on an increasing group size. This result might be attributed
to the fact that the trail length on which robots can exist increases only as the
square-root of the field size. The alteration at S
1
improved the group-foraging
performance under increased group size (i.e., altered the slope of group-foraging
performance against group size from negative to positive), but not the perfor-
mance itself. In our observation, local clustering of the searching robots (Fig. 5)
was observed mainly under the priority rules SNN. Therefore, this result sug-
gests that the collision-induced priority rule has an important role in avoiding
the negative impact of overcrowding on the field.
8
In this paper, we studied howpriority rulesaffect the group-foraging per-
formance of pheromone-mediated swarm robots. The priority rules are induced
when two robots collide on the pheromone trail and/or on the field. Using real
robot experiments, we found that the alteration in the reaction rules, compo-
nents of a priority rule, can have substantial effects on the performance. Our
results highlight the importance of implementing “fine-tuning” algorithms to
improve the performance of complex swarm systems. In future study, the scala-
bility of the improving effects of priority rules on group performance should be
investigated by numerical simulations.
References
1. Fujisawa, R., Imamura, H., Hashimoto, T., Matsuno, F.: Communication using
pheromone field for multiple robots. In: Proc. IEEE/RSJ 2008 International Con-
ference on Intelligent Robots and Systems. (2008)
2. Bonabeau, E., Dorigo, M., Theraulaz, G.: Swarm Intelligence. Oxford University
Press (1999)
3. olldobler, B., Wilson, E.: The superorganism: the beauty, elegance, and
strangeness of insect societies. WW Norton & Company (2009)
4. Dorigo, M., St¨utzle, T.: Ant Colony Optimization. Bradford Books (2004)
5. Sugawara, K., Kazama, T., Watanabe, T.: Foraging behavior of interacting robots
with virtual pheromone. In: Proc. IEEE/RSJ Int. Conf. on Intelligent Robots and
Systems (IROS2004). Volume 3. (2004) 3074–3079
6. Garnier, S., Tache, F., Combe, M., Grimal, A., Theraulaz, G.: Alice in pheromone
land: An experimental setup for the study of ant-like robots. In: Pro c. IEEE Swarm
Intelligence Symposium (SIS2007). (1-5 April 2007) 37–44
7. Shimoyama, I., Kanzaki, R.: Biological type micromachine. Journal of society of
biomechanisms 22 (1998) 152–157 in Japanese.
8. Purnamadjaja, A.H., Russell, R.A.: Guiding robots’ behaviors using pheromone
communication. Autonomous Robots 23 (2007) 113–130
9. Fujisawa, R., Dobata, S., Kubota, D., Imamura, H., , Matsuno, F.: Dependency by
concentration of pheromone trail for multiple robots. In: Proc. Sixth International
Conference on Ant Colony Optimization and Swarm Intelligence. (2008) 283–290
10. Krieger, M., Billeter, J., Keller, L.: Ant-like task allocation and recruitment in
co operative rob ots. Nature 406 (2000) 992–995
11. Chang, D., Shadden, S., Marsden, J., Olfati-Saber, R.: Collision avoidance for
multiple agent systems. In: Decision and Control, 2003. Proceedings. 42nd IEEE
Conference on. Volume 1., IEEE (2003) 539–543
12. Burd, M., Archer, D., Aranwela, N., Stradling, D.: Traffic dynamics of the leaf-
cutting ant, atta cephalotes. The American Naturalist 159 (2002) 283–293
13. Dussutour, A., Fourcassi´e, V., Helbing, D., Deneubourg, J.: Optimal traffic orga-
nization in ants under crowded conditions. Nature 428 (2004) 70–73
14. Dussutour, A., Beshers, S., Deneub ourg, J., Fourcassi´e, V.: Priority rules govern
the organization of traffic on foraging trails under crowding conditions in the leaf-
cutting ant atta colombica. Journal of Experimental Biology 212 (2009) 499–505
15. Fujisawa, R., Shimizu, Y., Matsuno, F.: Effectiveness of tuning of pheromone trail
lifetime in attraction of robot swarm. In: Proc. on International Symposium on
System Integration, SII2011. (2011) D2–2
... Because the use of the trail inevitably puts robots (as it does ants) into traffic-jam-like overcrowding, an accessory regulation that supports efficient pheromone communication is required in both systems. As a solution to the robotic overcrowding, we heuristically introduced a set of collisionprocessing behaviors in the robots (29,40). These behaviors constitute an overall "traffic rule", such that inbound (food-to-nest) robots are always given priority over outbound (nest-tofood) robots. ...
... Among the 16 multilocus genotypes, the genotype {1,0,1;1} showed the highest swarm fitness, in which a traffic rule was established with outbound robots (b3 = 1) giving priority to inbound robots (b2 = 0) on the pheromone trail. This result was consistent with our previous real robot experiment (40). To confirm that the traffic rule helped the swarms to avoid overcrowding, we counted the number of collisions during each simulation run. ...
... The basic algorithm for the pheromone-mediated group foraging behavior (29) has been well validated by comparison of the dynamic properties between simulated and real robot systems (28,29,40). In brief, once a searching robot (state S1) finds food, it starts to secrete a chemical compound on the ground while returning to its nest with a virtual food item (state S2). ...
Preprint
Full-text available
The evolution of complexity is one of the prime features of life on Earth. Although well accepted as the product of adaptation, the dynamics underlying the evolutionary build-up of complex adaptive systems remains poorly resolved. Using simulated robot swarms that exhibit ant-like group foraging with trail pheromones, we show that their swarm intelligence paradoxically involves regulatory behavior that arises in advance. We focused on a "traffic rule" on their foraging trail as a regulatory trait. We allowed the simulated robot swarms to evolve pheromone responsiveness and behaviors simultaneously. In most cases, the traffic rule, initially arising as selectively neutral component behaviors, assisted the group foraging system to bypass a fitness valley caused by overcrowding on the trail. Our study reveals a hitherto underappreciated role of regulatory mechanisms in the origin of swarm intelligence, as well as highlights the importance of embodiment in the study of their evolution.
... Because the use of the trail inevitably puts robots (as it does ants) into traffic-jam-like overcrowding, an accessory regulation that supports efficient pheromone communication is required in both systems. As a solution to the robotic overcrowding, we heuristically introduced a set of collision-avoidance behaviors in the robots, which improved their group-foraging performance 29,40 . These behaviors constitute an overall traffic rule, such that inbound (food-to-nest) robots are always given priority over outbound (nest-to-food) robots. ...
... Among the 16 multilocus genotypes, the genotype {1,0,1;1} showed the highest swarm fitness, in which a traffic rule was established with outbound robots (b 3 = 1) giving priority to inbound robots (b 2 = 0) on the pheromone trail. This result was consistent with our previous real robot experiment 40 . To confirm that the traffic rule helped the swarms to avoid overcrowding, we counted the number of collisions during each simulation run. ...
... It should be noted here that the rectangular field setup used in our simulations was not the primary factor giving an advantage of having the traffic rule. Our previous study using real robots 40 showed that the same traffic rule {1,0,1;1} remained essential for the pheromonemediated group foraging with fewer (n = 4-10) robots on a field of different shape (3600 mm × 3600 mm). Our field setup might be applied not only to the group foraging behaviour (see Introduction) but also to the traffic flow inside ant nests, where numerous ants have to manage overcrowding along their underground galleries. ...
Article
Full-text available
The evolution of complexity is one of the prime features of life on Earth. Although well accepted as the product of adaptation, the dynamics underlying the evolutionary build-up of complex adaptive systems remains poorly resolved. Using simulated robot swarms that exhibit ant-like group foraging with trail pheromones, we show that their self-organizing capacity paradoxically involves regulatory behavior that arises in advance. We focus on a traffic rule on their foraging trail as a regulatory trait. We allow the simulated robot swarms to evolve pheromone responsiveness and traffic rules simultaneously. In most cases, the traffic rule, initially arising as selectively neutral component behaviors, assists the group foraging system to bypass a fitness valley caused by overcrowding on the trail. Our study reveals a hitherto underappreciated role of regulatory mechanisms in the origin of complex adaptive systems, as well as highlights the importance of embodiment in the study of their evolution.
... In some research, chemical substances are used as pheromone markers [62], and alcohol is widely used to act as a virtual pheromone [63][64][65][66]. It is obvious that this idea is not suitable for an extensive underwater area. ...
Article
Full-text available
The advent of the swarm makes it feasible to dynamically monitor a wide area for maritime applications. The crucial problems of underwater swarm monitoring are communication and behavior coordination. To tackle these problems, we propose a wide area monitoring strategy that searches for static targets of interest simultaneously. Traditionally, an underwater robot adopts either acoustic communication or optical communication. However, the former is low in bandwidth and the latter is short in communication range. Our strategy coordinates underwater robots through indirect communication, which is inspired by social insects that exchange information by pheromone. The indirect communication is established with the help of a set of underwater communication nodes. We adopt a virtual pheromone-based controller and provide a set of rules to integrate the area of interest into the pheromone. Based on the information in the virtual pheromone, behavior laws are developed to guide the swarm to monitor and search with nearby information. In addition, a robot can improve its performance when using additional far-away pheromone information. The monitoring strategy is further improved by adopting a swarm evolution scheme which automatically adjusts the visiting period. Experimental results show that our strategy is superior to the random strategy in most cases.
... In the case of a collision from the back, the robot ignores the event. A detailed investigation of this collision processing in relation to swarm performance of real robots has been reported elsewhere (Fujisawa et al. 2012). ...
Article
Full-text available
In swarm robotics, communication among the robots is essential. Inspired by biological swarms using pheromones, we propose the use of chemical compounds to realize group foraging behavior in robot swarms. We designed a fully autonomous robot, and then created a swarm using ethanol as the trail pheromone allowing the robots to communicate with one another indirectly via pheromone trails. Our group recruitment and cooperative transport algorithms provide the robots with the required swarm behavior. We conducted both simulations and experiments with real robot swarms, and analyzed the data statistically to investigate any changes caused by pheromone communication in the performance of the swarm in solving foraging recruitment and cooperative transport tasks. The results show that the robots can communicate using pheromone trails, and that the improvement due to pheromone communication may be non-linear, depending on the size of the robot swarm.
Conference Paper
This paper describes an implementation of Lévy walk (or Lévy flight) to pheromone communicating swarm robots. Lévy flight is a special class of random walk in which the step length distribution is given by power law distribution. Lévy flight is known to maximize the efficiency of resource searches in uncertain environments. Using computer simulations, we show that the Lévy walk-like searching strategy can maximize the group foraging efficiency of the swarm robots using pheromone trails (mimicking ant group foraging), as well as maximize individual searching area. The Lévy walk was achieved by adjusting the probability per unit time with which an individual robot moves forward (otherwise it turns to right, to left, and reverse). We discuss the effect of swarming on optimal parameter values of Lévy walk. Optimization of individual searching strategies should be studied further, both in swarm robots and real organisms.
Article
Full-text available
Swarm intelligence is a relatively new approach to problem solving that takes inspiration from the social behaviors of insects and of other animals. In particular, ants have inspired a number of methods and techniques among which the most studied and the most successful is the general purpose optimization technique known as ant colony optimization. Ant colony optimization (ACO) takes inspiration from the foraging behavior of some ant species. These ants deposit pheromone on the ground in order to mark some favorable path that should be followed by other members of the colony. Ant colony optimization exploits a similar mechanism for solving optimization problems. From the early nineties, when the first ant colony optimization algorithm was proposed, ACO attracted the attention of increasing numbers of researchers and many successful applications are now available. Moreover, a substantial corpus of theoretical results is becoming available that provides useful guidelines to researchers and practitioners in further applications of ACO. The goal of this article is to introduce ant colony optimization and to survey its most notable applications
Conference Paper
Full-text available
In this paper, we consider a issue that the reliable and the inexpensive communication method in swarm robotics. The ants forage for preys by using pheromone trails. They lay down the pheromone trails between preys and a nest. By detecting the trail pheromone, they can find the preys. Though they do not have excellent intelligence, they can communicate with each other and cooperate by adding information to the environment, like a pheromone. This communication method has a merit that an agent does not need to memorize the place of the preys. We consider to answer the issue that ldquoHow do the swarm robots communicate using pheromone trail?rdquo. We construct a swarm behavior simulator and develop swarm robots that communicate using the pheromone trail. We demonstrate the effectiveness of the communication using the pheromone trail by computer simulations and experiments using swarm robots. To realize this purpose, we design a swarm behavior algorithm, based on 4 perceptual signs (stimuli) and 3 effector signs (actions). In the simulations, an experimental field is discretized by computational grids, and evaporation and diffusion are phenomena of the pheromone modeled by discretized equations. The proposed algorithm is demonstrated by the simulation. Simulation result shows that proposed algorithms act effectively. Based on the simulation results, we set three robots, one nest and one prey in the flat experimental field. We observe three robotspsila behavior and the state of the environment for 20 minutes. The robots laid down the pheromone trail between the nest and the prey, and reinforced the pheromone trail many times. This fact means that swarm robots can realize the function of the chemical, indirect, plastic and local communication like ants by using the pheromone trail.
Conference Paper
Full-text available
In this paper, we discuss the concentration dependency of pheromone communication in swarm robotics. Instead of a pheromone trail and the insect antenna, we used ethanol and an alcohol sensor. This experimental system has a trade-off problem; high concentrations of the pheromone yield high signal strength but the signal duration is short, while low pheromone concentrations yield low signal strength but a long signal duration. We examined the optimal pheromone concentration for a swarm of robots. For this purpose, we developed a swarm behaviour algorithm and swarm robots that communicate using a pheromone trail. In addition, we discuss the effects of the pheromone concentration.
Conference Paper
We focus on swarm behavior of ants. They communicate with each other using pheromone. They forage food using pheromone trail to attract many ants to foods. They also realize adaptive foraging by adjusting the property of the pheromone trail according to changes in the environment. It is effective to apply the swarm behavioral mechanism to the robotics. We propose adjustment method of the trail duration time based on the food quantity. We obtained simulation and experimental results that the proposed adjustment of the trail duration time is effective.
Keywords: Swarm Intelligence ; Collective Robotics ; SWARM_BOTS ; Evolutionary Robotics Note: Proceedings of the ANTS 2004, 4th International Workshop Sponsor: swarm-bots, OFES 01-0012-1 Reference LIS-BOOK-2004-002 URL: http://iridia.ulb.ac.be/~ants/ants2004/ Record created on 2006-01-12, modified on 2016-08-08
Chapter
Ant Colony Optimization (ACO) is a stochastic local search method that has been inspired by the pheromone trail laying and following behavior of some ant species [1]. Artificial ants in ACO essentially are randomized construction procedures that generate solutions based on (artificial) pheromone trails and heuristic information that are associated to solution components. Since the first ACO algorithm has been proposed in 1991, this algorithmic method has attracted a large number of researchers and in the meantime it has reached a significant level of maturity. In fact, ACO is now a well-established search technique for tackling a wide variety of computationally hard problems.
Conference Paper
In multi-robot system, communication is indispensable for effective cooperative working. In this system, direct communication by physical methods such as light, sound, radio wave is quite general. But in biological system, especially in the insect world, not only the physical but also the chemical communication methods can be observed. As the chemical methods have some unique properties, it is challenging to apply such a method to the cooperative multi-robot system. Unfortunately, to treat real chemical materials for the robots is not easy for now because of some technical difficulties. In this paper, we propose virtual pheromone system in which chemical signals are simulated with the graphics projected on the floor, and in which the robots decide their action depending on the color information of the graphics. We examined the performance of this system through the foraging task, which is one of the most popular tasks for multi-robot system and is generally observed in ant societies.
Article
This paper describes an ongoing project to investigate the uses of pheromones as a means of communication in robotics. The particular example of pheromone communication considered here was inspired by queen bee pheromones that have a number of crucial functions in a bee colony, such as keeping together and stabilizing the colony. In the context of a robotic system, one of the proposed applications for robot pheromones is to allow a group of robots to be guided by a robot leader. The robot leader could release different chemicals to elicit a range of behaviors from other members of the group. A change of the operating temperature of tin oxide gas sensors has been implemented in order to differentiate different chemicals. This paper provides details of the robots used in the project and their behaviors. The sensors, especially the method of using the tin oxide gas sensors, the robot control algorithms and experimental results are presented. In this project, pheromones were used to trigger congregating behavior and light seeking in a group of robots.