Multi-Robot Cooperative Tasks using Combined Nature-Inspired

Techniques

Nunzia Palmieri

1,2

, Floriano de Rango

, Xin She Yang

and Salvatore Marano

Dept. Computer Engineering Modeling, Electronics, and Systems Science University of Calabria, Rende (CS), Italy

School of Science and Technology, Middlesex University The Burroughs, London, U.K.

Keywords: Swarm Intelligence, Swarm Robotics, Firefly Algorithm, Bio-Inspired Algorithm.

Abstract: In this paper, two metaheuristics are presented for exploration and mine disarming tasks performed by a

swarm of robots. The objective is to explore autonomously an unknown area in order to discover the mines,

disseminated in the area, and disarm them in cooperative manner since a mine needs multiple robots to disarm.

The problem is bi-objective: distributing in different regions the robots in order to explore the area in a

minimum amount of time and recruiting the robots in the same location to disarm the mines. While

autonomous exploration has been investigated in the past, we specifically focus on the issue of how the swarm

can inform its members about the detected mines, and guide robots to the locations. We propose two bio-

inspired strategies to coordinate the swarm: the first is based on the Ant Colony Optimization (ATS-RR) and

the other is based on the Firefly Algorithm (FTS-RR). Our experiments were conducted by simulations

evaluating the performance in terms of exploring and disarming time and the number of accesses in the

operative grid area applying both strategies in comparison with the Particle Swarm Optimization (PSO). The

results show that FTS-RR strategy performs better especially when the complexity of the tasks increases.

1 INTRODUCTION

In the robotics field, an important aspect of multiple

agents systems is the coordination that allows the

system accomplishes efficiently general tasks, such as

exploration, coverage and surveillance. Autonomous

robots, equipped with proper sensors, are deployed in

the environment to find the object of interest, i.e., fire

spots in the jungle, mines in unknown area, missing

black box from a crashed airplane, or to measure a

concentration of hazardous materials. The use of a

swarm of robots is utilized in these applications for

the expected benefits of reducing risks to humans,

lower cost, and improved efficiency (Bellingham and

Godin 2007).

Swarm robotics is a new approach to the

coordination of a multi robots system, that typically

consist of a population of simple agents interaction

locally with each other and with the environment. The

benefit of cooperation can be significant in situation

where global knowledge of the environment does not

exist. Individuals within the group interact according

to the swarm intelligence algorithms by exchanging

information that is useful for performing the tasks

collectively.

In our collective construction task, there are some

mines randomly distributed in an unknown area. The

robots should first search for these mines

individually, but for disarming task, multiple robots

are needed to work together. The problem is not a

pure exploration: on one hand, it is required for robots

to cover as much area as possible in the minimum

amount of time, avoiding any overlapping area. On

the other hand, the problem needs to allocate more

robots in the same area to disarm a mine. The problem

is a bi-objective optimization problem where robots

have to make decisions whether to explore the area or

to help other robots to disarm the detected mines.

Because the problem of the unknown lands with

the constraint to disarm mine is a NP hard problem,

we proposed a combined approach using two bio-

inspired meta-heuristic approaches such as Ant

Colony Optimization (ACO) and Firefly algorithm

(FA) to perform the coordination task among robots.

Basically, each robot consists of two phases

during the task: searching and disarming. When there

is no detected mine, the robot status should be in the

searching phase, where robots are exploring the area

and searching for mines, taking into account the

quantity of pheromone perceived in the cells. Once

Palmieri, N., Rango, F., Yang, X. and Marano, S..

Multi-Robot Cooperative Tasks using Combined Nature-Inspired Techniques.

In Proceedings of the 7th International Joint Conference on Computational Intelligence (IJCCI 2015) - Volume 1: ECTA, pages 74-82

ISBN: 978-989-758-157-1

mine is detected either by the robot itself or by its

neighbours, the robot status should be switched to the

disarming phase, under specific condition. The

strategy for the exploration task is designed according

to the main ideas of the ant system (Dorigo et al.,

2006). While the robots navigate, they deposit a

specific substance, the pheromone (the analogue of

the pheromone in biological ant systems), into the

environment. At each time/iteration, each robot

receives information from the pheromone and makes

a navigation decision: it chooses the area in which it

perceives a less quantity of pheromone because this

area has a greater probability to be unexplored (De

Rango and Palmieri, 2012; De Rango et al., 2015).

The algorithm for exploration has been previously

validated (De Rango and Palmieri, 2012) and this

paper presents the analysis of the recruiting strategies

in order to disarm the mines. The first is based on the

exploration strategy and uses the pheromone to attract

the robots in the area where the mine is placed. The

second strategy is based on the new recent bio-

inspired technique called Firefly Algorithm (FA)

where the robots that detect the mines become the

fireflies and try to attract the other robots according

with a certain formula (Yang, 2009; Yang, 2010).

These strategies were compared to the well known

Particle Swarm Optimization in order to evaluate the

better coordination mechanism for this problem. This

contribution can be effective because the recruiting

strategy can affect the exploration task and the overall

bi-objective exploring and recruiting tasks.

The paper is organized as follows. Section 2

introduces the related work. Section 3 describes the

firefly algorithm. In Section 4 we present the problem

statement. In Section 5 we present the distributed

cooperative algorithms for a multi-robot disarming

task. Section 6 presents the simulation results using a

java-based platform and Section 7 analyses the

quality of the solutions. To conclude the paper,

Section 8 outlines the main research conclusions and

discusses topics for future work.

2 RELATED WORK

Multi-robot exploration has received much attention

from the research community. Swarm robotic

searching algorithm is one of the most concerns of the

researchers besides those basic tasks. The swarm

intelligence shows great ability in scalable, flexibility

and robustness and is suitable for real life applications

with the aid of various existing strategies. Within the

context of swarm robotics, most work on cooperative

exploration is based on biologically behaviour and

indirect stigmergic communication (rather than on

local information, which can be applied to systems

related to GPS, maps, wireless communications).

This approach is typically inspired by the behaviour

of certain types of animals, like the ants, that use

chemical substances known as pheromone to induce

behavioural changes in other members of the same

species (Russell, 1999; Sugawara et al., 2004; Garnier

et al., 2007; Ducatelle et al., 2011, Masàr, 2013).

Other authors experiment with chemical

pheromone traces, e.g. using alcohol (Fujisawa et al.,

2008) or using a special phosphorescent glowing

paint (Mayet, 2010). Another approach is the

pheromone robotics where robots spread out over an

area and indicate the direction to a goal robot using

infrared communication (Payton et al., 2001). In our

approach, during the exploration the robots sign/mark

the crossed cell through the scent that can be detected

by the other robots; the robots choose the cell that has

the lowest quantity of substance to allow the

exploration of the unvisited cells in order to cover the

overall area in less time (De Rango and Palmieri,

2012).

The self-organizing properties of animal swarms

such as insects have been studied for better

understanding of the underlying concept of

decentralized decision-making in nature, but it also

gave a new approach in applications to multi-agent

systems engineering and robotics. Bio-inspired

approaches have been proposed for multi-robot

division of labour in applications such as exploration

and path formation, or cooperative transport and prey

retrieval. Within the context of swarm robotics, most

work on cooperative tasks is based on social

behaviour like Ant Colony Optimization (Dorigo et

al., 2006), Particle Swarm Optimization (Meng and

Gan, 2008) Bee Algorithm (Jevtic et al., 2012).

For sharing information and accomplishing the

tasks there are, basically, three ways of information

sharing in the swarm: direct communication

(wireless, GPS), communication through

environment (stigmergy) and sensing. More than one

type of interaction can be used in one swarm, for

instance, each robot senses the environment and

communicates with their neighbour. Balch (Balch,

2005) discussed the influences of three types of

communications on the swarm performance and Tan

(Tan and Zheng, 2013) presents an accurate analysis

of the different type of communication and the impact

in a behaviour of swarm.

In this paper, we considered the spatial and

temporal dispersion of the pheromone to make the

scenario more realistic (De Rango and Palmieri,

2012). While walking, the robots leave pheromone,

Multi-Robot Cooperative Tasks using Combined Nature-Inspired Techniques

which marks the cells they took. This chemical

substance can be detected by other robots. After a

while, the concentration of pheromone decreases due

to the evaporation and diffusion associated with the

distance and with the time; in this way we can allow

continuous coverage of an area via implicit

coordination. The other robots, through proper

sensors, smell the scent in the environment and move

in the direction with a minimum amount of

pheromone that corresponds to an area less occupied

and probably an unexplored area. On the other hand,

in order to deactivate the mines, the first robot that

detects a mine (recruiter) in a cell, sprays another

scent smelled by the robots; in this case the robots

move into the cells with a higher concentration of

pheromone and reach the area where to deactivate the

mines. In this attraction strategy of the recruiter,

another recent and novel bio-inspired approach

inspired by other insects such as fireflies has been

investigated in this work so as to see the effectiveness

of the algorithm and potential use of different insect

behaviour on the robot coordination task and their

performance. The algorithm inspired by fireflies is

called Firefly algorithm (FA) and is summarized in

the next section.

3 FIREFLY ALGORITHM

The firefly algorithm is a nature-inspired meta-

heuristic algorithm developed in 2008 by Xin-She

Yang to solve optimization problems (Yang, 2009;

Yang, 2010; Yang, 2014). The algorithm is based on

the social flashing behavior of fireflies in nature. The

key ingredients of the method are the variations of

light intensity and formulation of attractiveness. In

general, the attractiveness of an individual is assumed

to be proportional to their brightness, which in turn is

associated with the encoded objective function.

In the firefly algorithm, there are three idealized

rules, which are based on some of the major flashing

characteristics of real fireflies. They are:

1. All fireflies are unisex, so that one firefly will

be attracted to other fireflies regardless of their

sex;

2. The degree of attractiveness of a firefly is

proportional to its brightness, which decreases

as the distance from the other firefly increases

due to the fact that the airabsorbs light. For any

two flashing fireflies, the less bright one will

move towards the brighter one. If there is not

a brighter or more attractive firefly than a

particular one in the neighborhood, it will then

move randomly;

3. The brightness or light intensity of a firefly is

determined by the value of the objective

function of a given problem.

The distance between any two fireflies i and j, at

positions X

and X

, respectively, can be defined as the

Cartesian or Euclidean distance as follows:

















∑



,



,









(1)

where x

i,k

is the k-th component of the spatial

coordinate X

of the i-th firefly and D is the number of

dimensions.

In the firefly algorithm, as the attractiveness function

of a firefly j, one should select any monotonically

decreasing function of the distance to the chosen

firefly, e.g., the exponential function:













(2)

where r

is the distance defined as in Eq. (1), β

is the

initial attractiveness at r

, and γ is an absorption

coefficient at the source which controls the decrease

of the light intensity.

The movement of a firefly i which is attracted by a

more attractive (i.e., brighter) firefly j is governed by

the following evolution equation:

































(3)

where the first term on the right-hand side is the

current position of the firefly, in our case a mine, the

second term is used for considering the attractiveness

of the firefly to light intensity seen by adjacent

fireflies, and the third term is used for the random

movement of a firefly in case there are not any

brighter ones. The coefficient α is a randomization

parameter determined by the problem of interest,

while σ is a random number generator uniformly

distributed in the space [0, 1].

Furthermore, we look at equation (3), thus non linear

equation provides much richer characteristics. Firstly,

if γ is very large, then attractiveness decreases too

quickly, this means that the second term in (3) became

negligible, leading to the standard simulated

annealing (SA). Secondly, if γ is very small ( i.e. →

0 ), then the exponential factor 







→1 and FA

reduces to a variant of particle swarm optimization

(PSO). Also, the randomization term can be extended

to other distributions such as Lévy flight.

Furthermore, FA uses non linear updating equation,

which can produce rich behavior and higher

convergence than linear updating equation used for

ECTA 2015 - 7th International Conference on Evolutionary Computation Theory and Applications

example in standard PSO. Regarding the parameters

setting, parametric studies suggest that 

=1 can be

used for most application; γ should related to the

scaling L. In general, we can set 





(Yang, 2014).

4 PROBLEM STATEMENT

We consider an environment assuming that it is

discretized into equally spaced cells that contains a

certain number of mines. Each cell has the potential

to consider three states: free, occupied by mine,

occupied by robot. Robots can move among cells and

they can have just local information about robots

(neighbors) or regions to explore (neighbor cells) in

order to provide a scalable strategy.

The considered scenario is presented under this

assumption:

1) The robots are equipped with proper sensors

that are able to deposit and smell the chemical

substances (pheromones) leaved by the other

robots; for exploration task they make

probabilistic decision based on amount of

pheromone in the cells. The exploration

strategy is the same for the recruiting

strategies.

2) The robots are equipped with proper sensor to

detect the mines.

3) The robots can move on a cell-by-cell basis to

explore new cells or to go towards the mine.

The robots during the exploration spray a scent

(pheromone) into the cells to support the navigation

of the others. In the algorithm, the robots decide the

direction of the movement relying on a probabilistic

law inherited by swarm intelligence and swarm

robotics techniques. The scent evaporates not only

due to diffusion effects in the time, but also in the

space according to the distance; this allows a higher

concentration of scent in the cell where the robot is

moving and a lower concentration depending on the

distance.

Let M be the matrix of size mxn representing the

coverage area of size mxn. Let M(i,j) be the cell in the

matrix with row i and column j. Let z be the number

of mines on a set MS to distribute on the grid in a

random fashion (e.g., it is applied a uniform

distribution on X and Y axes). The MS set is

characterized by the coordinates of the mines. For

example,

 





12,7,10,5,4,3MS

indicates that there

are 3 mines in the area with the coordinates (3,4),

(5,10) and (7,12). The robots can be placed on the

same initial cell or can be randomly distributed on the

grid area. It is assumed that each robot in a cell M(i,j)

can move just in the neighbor cells through discrete

movements. Let t

be the time necessary for a robot to

consider a cell, and let t

be the time necessary to

disarm a mine once it has been detected. It is assumed

that a fixed number of robots (rd

min

) are necessary to

disarm a mine; this means that for the exploration task

robots can be distributed among the area because each

robot can independently explore the cells, whereas for

the mine detection, more robots need to be recruited

in order to perform the task. M (i, j )_a is a variable

representing the number of robots (accesses) that

passed through the cell (i,j).

For the problem we define an bi-objective

function as both the time to detect and the disarming

the mine through the exploration on the overall grid.



tmin

and





min

(4a)

subject to





1_, ajiM

njmi ...1;...1





Mji ,





min

_, rdajiM 

with



MSji ,

This is a bi-objective optimization problem and its

solutions will result in a Pareto front. However, in

order to solve this problem more effectively, for

simplicity, we will combine these two objectives to

form a single objective optimization problem so as to

minimize the overall total time as follows:



















idetot

ttT

minmin

(4b)

subject to

,_1∀1,..;1,…/



,



∈





,













,



∈

The law used by the robots to choose the cells during

the movement is presented below (De Rango and

Palmieri 2012).

We consider a robot in a cell s and it will attribute

to the set of next cells v

following a probability as:















sNvsvp

sNi















(5)

where (p(v

|s) represents the probability that the

robot, that is in the cell s, chooses the cell v

; N(s) is

the set of neighbors to the cells, τ

vi,t

is the amount of

pheromone in the cell v

; ɳ

vi,t

is the heuristic

Multi-Robot Cooperative Tasks using Combined Nature-Inspired Techniques

parameter introduced to make the model more

realistic. In addition,



and θ are two parameters

which affect respectively the pheromone and

heuristic values.

Taking into account the spatial dispersion of the scent

and the temporal dispersion in the amount of

pheromone in the cell v where the robot will move

during the exploration is:



,









,







(6)

In order to explore different areas of the environment,

the robots choose the cell with a minimum amount of

pheromone (MINIMUM_TRACE_FOLLOWER),

corresponding to cells that probably are less

frequented and therefore not explored cells. The

chosen cell will be selected according with eq. (5):





svpv

inext

|min



sNv





(7)

5 BIO-INSPIRED APPROACH

FOR THE DISARMING TASK

The purpose of the problem is to discover all mines

disseminated in the area and to disarm them. In the

first strategy the first robot that detects a mine

becomes a recruiter,which on the basis of the

recruiting startegy can spray a scent in order to

inform the other robots about the presence of a mine

and to recruit other robots for the disarming task

(indirect communication).

Alternatively, in the second startegy, the robots are

equipped by wireless module and the recruiters can

send a packet where putting the information about the

mine position (this can be useful for the FA based

strategy).

We assumed that the robots are not able to

communicate in a multi-hop manner but just via a

direct message (single-hop) (using for example a

wireless radio). Each robot can only communicate

with its neighbors. Two robots are defined as

neighbours if the distance between them is less than a

pre-specified communication range. In the following

section, the recruitment issue is formalised in order to

apply the proposed combination of the two bio-

inspired techniques.

5.1 Ant-based based Team Strategy for

Robots Recruitment (ATS-RR)

For this strategy we assume that the robots are

equipped by sensor that perceived a pheromone,

different by the pheromone used for the exploration.

The robots communicate with others through the

environment (indirect communication).

We considered that the mine disarming time is equal

to the total evaporation time of substance (scent); in

this way when the mine is disarmed, the robots

involved in this operation will not be affected by scent

trails.

We assume t is the time in which the robot r

detected a mine and it deposits the substance. The

robot r continues to spray until all necessary robots

reach its position.

If m is the time needed to disarm the mine, the law for

the evaporation of the scent is the following:

ttt







(8)

where ζ

is the substance sprayed when the robot

detetects a mine. At the beginning ζ

= ζ

t0.

In this way after m steps ζ should be zeo so the scent

will not affect any more the movement of the other

robots. This assures that all robots will cover other

new space and disarm other mines completing the

task in an efficient and distributed manner.

5.2 Firefly based Team Strategy for

Robots Recruitment (FTS-RR)

For this task we considered the following assumption:

1) The robots are equipped by wireless module;

infact when a robot detects a mine, it becomes

a firefly and tries to attract other robots

sending messages via broadcast

communication to the robots in its wireless

range.

2) The robots, that receive messages by different

robots (fireflies), evaluate the light of fireflies

and choose the best firefly (at minimum

distance) and move toward firefly according to

a modified Discrete Firefly Algorithm.

When a robot finds a mine, during the exploration

task, it applies the FTS-RR and becomes the recruiter

of the other robots in order to disarm the mine. For

this purpose, in FTS-RR strategy, it becomes a firefly

and it tries to attract the other robots on the basis of

the mine position. The original version of FA is

applied in the continuous space, but in our case we

modified the algorithm in order to fit with our

problem. In our case, the robots can move in a

discrete space because they can go just in the

contiguous cells step-by-step. This means that when

a robot perceives at a distance the presence of a firefly

(the recruiter robot) and it is in a cell with coordinates

ECTA 2015 - 7th International Conference on Evolutionary Computation Theory and Applications

and y

, it can move according with the FA attraction

rules such as expressed below:







































)







































)

(9)

where x

and y

represent the coordinates of detected

mine translated in terms of row and column of the

matrix area. r

is the Euclidean distance between

mine (or recruiter) and robot that moves towards the

mine. The robot movement is conditioned by mine

(recruiter) position in the second term of the formula

(9) and by a random component in the third term. This

last term is useful to avoid that more robots go

towards the same mine if more mines are distributed

on the land (this avoids the local minimum in order to

approach to a global optimum) Fig.1.

In order to modify the FA to a discrete version, the

robot movements have been considered through three

possible value updates for each coordinates:



1,0,1 

such as expressed in Eq.(10). A robot r that























1



































0















1



































0















0



































0



(10)























1



































0















1



































0















0



































0



(11)

a) b)

Figure 1: Robots during the exploration receive two

recruiting calls because they are in an overlapped area. a)

In ATS-RR strategy the robots will choose the cell where

they perceive a greater quantity of pheromone; b) the FTS-

RR strategy tries to coordinate better the robots in the

disarming task considering the distance from the recruiter.

is in the cell of coordinates x

and y

such as depicted

in Fig.2 can move in eight possible cells according

with the three possible values attributed to x

and y

For example if applying eq.10 and eq.11 and the

result is {-1,+1} , the robot will move in the cell with

coordinates { x

i-1

; y

i+1

} such as depicted in Fig.2

Figure 2: Possible movements for a robot on the basis of the

and y

values.

6 PERFORMANCE

COMPARISON

In this section, we evaluate the performance of the

two proposed algorithms in comparison with the well

known Particle Swarm Optimization, focusing on the

time to cover all unknown area and disarm all mines

and the number of accesses in the cells in order to see

the effectiveness of the joint exploration task (space

distribution) and disarming task (space

concentration).The specific FTS-RR parameters were

set as follows:



=1;



=0.2; ϒ=1/L where L is max{m,

n} where m and n are the number of rows and columns

of the matrix M, respectively. For the ATS-RR, the

parameter values were set as follow: =1, θ=1; ɳ=0.9.

We considered different scenarios by varying the

minimum number of robots necessary to disarm a

mine, the total number of robots in the rescue area,

the dimension of grid and the number of mines. To

highlight the performance benefits, we use random

positions of the mines and the robots in the area by

varyng the number of robots so as to investigate the

performance of all strategies.

In Figures 3, 4 and 5 the number of mines and the grid

area have been fixed, respectively, to 3 and 20x20.

In Fig.3a and 3b, a comparison of the three algorithms

is depicted. In particular, it is shown the time to

complete both tasks measured as the number of

iterations and the number of accesses in the cells. The

convergence time and number of accesses in the cells

was averaged over 50 independent simulation runs in

order to enter in the 5% of confidence interval.

It is possible to see that the FTS-RR strategy performs

better mainly when the number of robots is low. This

is due to the better robot recruitment when a mine is

discovered that is able to balance the robot

coordination and movements among all mines. On the

Overlapped area

Mine

Robot

Area in which robots perceive the mine

pheromone

rij

Overlappedarea

Wirel e ssrange

Mine

Robot

x

‐1,y

‐1x

‐1,y

 x

‐1,y

+1

x

,y

‐1r x

,y

+1

+1,y

‐1x

+1,y

+1

Multi-Robot Cooperative Tasks using Combined Nature-Inspired Techniques

other hand, when the number of robots increases and

the number of minimum robots to disarm a mine is

equal to 2, ATS-RR, FTS-RR and PSO-RR are

similar because the high number of robots in

comparison to the number of mines allows to

complete both tasks in a lower time and no significant

difference between the strategies is so evident. The

number of accesses in the cells was plotted in Fig.3b.

Simulations show that FTS-RR is able to balance

better the robots in the recruitment phase considering

that the exploration phase is common to all

algorithms. This determines that a lower average

number of accesses in the cells can be obtained in

FTS-RR in comparison with ATS-RR and PSO-RR.

The same considerations can be made when the

number of robots needed to disarm a mine is 3(Fig.

4). An interesting result is shown in Fig 5. In these

cases, FTS-RR performs better for both low and high

numbers of robots in the convergence time especially

in comparison with the ATS-RR. This is due to the

most effective recruitment strategy that is able to

better distribute robots when, in the overlapping area,

more recruiters can engage robots for disarming. In

this case, the use of distance and the firefly algorithm

allows robots to spread over different mines avoiding

going towards the same mines to disarm. The overall

effect is a reduction in the task execution time.

Figure 3: Comparison between ATS-RR, FTS-RR and PSO

evaluating with 3 mines and 2 robots per mine to disarm

and increasing number of robots: a) number of iterations; b)

number of accesses in the cell.

Figure 4: Comparison between ATS-RR, FTS-RR and PSO

evaluating with 3 mines and 3 robots per mine to disarm

and increasing number of robots: a) number of iterations; b)

number of accesses in the cell.

Figure 5: Comparison between ATS-RR, FTS-RR and PSO

evaluating with 3 mines and 4 robots per mine to disarm

and increasing number of robots: a) number of iterations; b)

number of accesses in the cell.

ECTA 2015 - 7th International Conference on Evolutionary Computation Theory and Applications

Concerning the number of accesses is much lower in

the FTS-RR. Increasing the complexity of the task the

difference between the three different algorithms, in

terms of overall time to complete the tasks, is greater

(Fig.6 and Fig. 7). This means that the best

performing of recruiting task can affect indirectly the

discovery task leading to a better distribution of

robots among mines to disarm and consequently to

explore the novel un-explored cells.

Figure 6: Comparison between ATS-RR, FTS-RR and

PSO, evaluating with 5 mines and 4 robots per mine to

disarm and increasing number of robots in a 30x30 gird

map.

Figure 7: Comparison between ATS-RR, FTS-RR and PSO

in terms of overall time, evaluating with 10 mines and 4

robots per mine to disarm and increasing number of robots

in a grid 40x40.

7 SOLUTION QUALITY

ANALYSIS

To validate the quality of solutions and results of the

three metaheuristics we have also considered the p

values of Student t-tests. The t-tests were used to

analyze the relationships between the results obtained

from the three metaheuristics. The parameter of

interest is the p-value.

Table I, Table II and Table III show the p-value

obtained from the t-tests using all above simulation

results by considering each parameter (the number of

interactions and the number of accesses in the cell)

for all considered scenario.

Table 1: Results of p values in the t Test for ATS-RR and

FTS-RR.

Table 2: Results of p values in the t Test for ATS-RR and

PSO-RR.

Table 3: Results of p values in the t Test for FTS-RR and

PSO-RR.

By analyzing the experimental results, it can be

observed a significant difference between the ATS-

RR and FTS-RR and ATS-RR and the PSO-RR. In all

considered scenario the p-value < 0.05, so there is a

strong statistic evidence of the difference between the

strategies. Regarding to the PSO-RR and the FTS-RR

analyzing the results (Table III) it can be observed

that the p-value <0.05 except for the Scenario 1 and

2. For other Scenario the p-value <0.05. This

confirms that the FTS-RR exhibits superior

performance when the complexity of tasks, in terms

of dimension of operative area, number of mines and

number of robots need to disarm a mine, increases.

8 CONCLUSION

Novel bio-inspired self-organizing coordination

algorithms are proposed for a distributed multi robot

coordination in a mined region.

For this purpose, two different strategies for the

mine disarming task combined with an Ant-based

space discovery strategy are proposed. The first

strategy ATS-RR isbased on Ant Colony

optimization, and the other one is based on the Firefly

Algorithm (FTS-RR). We compare both startegies

with the well known Particle Swarm Optimization.

By extensive simulations,we have concluded that the

FTS-RR can perform better in terms of the time to

Scenario

4

(Fig.6)

Scenario

5

(Fig.7)

Number

of

Iteration

Numberof

Accessin

thecells

Number

of

Iteration

Numberof

Accessin

thecells

Number

of

Iteration

Numberof

Accessinthe

cells

Number

of

Iteration

Number

of

Iteration

pvalue 0,0362 0,0261 0,0447 0,00342 0,02976 0,00172 0,001153 1,32E‐05

Scenario1

(Fig.3a,Fig.3b)

Scenario2

(Fig.4a,4b)

Scenario

3

(Fig.5a,5b)

Scenario

4

(Fig.6)

Scenario

5

(Fig.7)

Number

of

Iteration

Numberof

Accessin

thecells

Number

of

Iteration

Numberof

Accessin

thecells

Number

of

Iteration

Numberof

Accessinthe

cells

Number

of

Iteration

Number

of

Iteration

pvalue 0,0552 0,0328 0,0476 0,00048 0,0347 0,00368 0,003667 0,00030

Scenario1

(Fig.3a,Fig.3b)

Scenario2

(Fig.4a,4b)

Scenario3



(Fig.5a,5b)

ATS‐RRVSPSO‐RR

Scenario

4

(Fig.6)

Scenario

5

(Fig.7)

Number

of

Iteration

Numberof

Accessin

thecells

Number

of

Iteration

Numberof

Accessin

thecells

Number

of

Iteration

Numberof

Accessinthe

cells

Number

of

Iteration

Number

of

Iteration

pvalue 0,0623 0,0522 0,0524 0,001524 0,0495 0,0045 0,00667 0,00364

FTS‐RRVSPSO‐RR

Scenario1

(Fig.3a,Fig.3b)

Scenario2

(Fig.4a,4b)

Scenario3

(Fig.5a,5b)

Multi-Robot Cooperative Tasks using Combined Nature-Inspired Techniques

complete the task and the number of acccesses in the

cell, leading to a better distribution of robots with a

better combined time for both mine disarming and

exploration tasks expecially when the complexity of

the task increases.

Possible directions for future work can be followed.

Firstly, it would be very useful to vary the ACO and

FFA parameters and then evaluate their performance.

Secondly, it will also be fruitful to study the energy

consumed by the robots. Finally, mobile targets can

be a much better and realistic extension of the current

work. It can be expected that this present work can

form a basis for further extension and research.

REFERENCES

Balch, T. (2005). Communication, diversity and learning:

cornerstones of swarm behaviour. In: Swarm robotics,

lecture notes in computer science, vol.3342. Springer;

p. 21e30.

Bellingham, J. D. and Godin, M. (2007). Robotics in

Remote and Hostile Environments. Science, vol. 318,

pp. 1098-1102.

De Rango, F. and Palmieri, N. (2012). A swarm-based robot

team coordination protocol for mine detection and

unknown space discovery. In 8th Int. Conf. on Wireless

Communications and Mobile Computing (IWCMC), p.

703-709.

De Rango, F., Palmieri N.,Yang X-S, and Marano. (2015).

Bio-inspired Exploring and Recruiting Tasks in a Team

of Distributed Robots over Mined Regions. In

International Symposium on Performance Evaluation

of Computer and Telecommunication System

(SPECTS).

Dorigo, M., Birattari, M. and Stutzle, T. (2006). Ant colony

optimization. IEEE Comp. Intell. Mag., Vol. 1, No. 4,

28–39.

Ducatelle, F., Di Caro, G.A., Pinciroli, C. and Gambardella

L.M,. (2011). Selforganized cooperation between

robotic swarms. Swarm Intelligence, 5(2):73–96.

Fujisawa, R., Dobata, S., Kubota, d., Imamura, H., and

Matsuno, F. (2008). Dependency by concentration of

pheromone trail for multiple robots. In Proceedings of

ANTS 2008, 6th International Workshop on Ant

Algorithms and Swarm Intelligence, volume 4217 of

LNCS, pages 283–290. Springer.

Garnier, S., Tache, F., Combe, M., Grimal, A., and

Theraulaz, G. (2007). Alice in pheromone land: An

experimental setup for the study of ant-like robots. In

Proc. of the IEEE Swarm Intelligence Symp. (SIS),

pages 37–44, Washington, DC, USA.

Jevtic, A., Gutiérrez, A., Andina, D. and Jamshidi, M.

(2012). Distributed Bees Algortithm for Task

Allocation in Swarm of Robots. IEEE Systems Journal,

Vol 6 NO 2.

Masár, M. (2013). A biologically inspired swarm robot

coordination algorithm for exploration and

surveillance. In Proc. 17th IEEE Int. Conf. on

Intelligent Engineering Systems INES, Budapest, pp.

271-275, ISBN 978-1-4799-0830-1.

Mayet, R., Roberz, J., Schmickl, T., & Crailsheim, K.

(2010). Antbots: a feasible visual emulation of

pheromone trails for swarm robots. In Proceedings of

the 7th international conference on swarm intelligence

(ANTS) (pp. 84–94).

Meng, Y., Gan, J. A. (2008). A distributed swarm

intelligence based algorithm for a cooperative

multirobot construction task. Swarm Intelligence

Symposium, IEEE Transactions, pp. 1-6.

Payton, D., Daily, M., Estowski, R., Howard, M., & Lee, C.

(2001). Pheromone robotics. Autonomous Robots,

11(3), 319–324.

Russell, R. (1999). Ant trails - An example for robots to

follow. In Proc. of IEEE Int. Conf. on Robotics and

Automation (ICRA), pp. 2698–2703.

Sugawara, K., Kazama, T., and Watanabe, T.(2004).

Foraging behavior of interacting robots with virtual

pheromone. In Proc. of the IEEE/RSJ Int. Conf. on

Intelligent Robots and Systems (IROS), pp. 3074–3079.

Tan, Y. and Zheng, Z-Y. (2013). Research Advance in

Swarm Robotics. Defence Technology 9(1):18–39.

Yang, X.S. (2009). Firefly algorithms for multimodal

optimization. Lectures Notes in Computer Science,

5792(2009) 169-178.

Yang, X.S. (2010). Firefly algorithm, stochastic test

functions and designoptimisation. Int. Journal of Bio-

Inspired Computation, 2(2) 78-84

Yang, X.S. (2014). Cuckoo Search and Firefly Algorithm

Theory and Applications. Book on Studies in

Computational Intelligence, ISBN: 978-3-319-02140-

ECTA 2015 - 7th International Conference on Evolutionary Computation Theory and Applications