You are on page 1of 11

Journal of King Saud University – Computer and Information Sciences xxx (xxxx) xxx

Contents lists available at ScienceDirect

Journal of King Saud University –


Computer and Information Sciences
journal homepage: www.sciencedirect.com

Multi-robot control for a static polygon formation using


Neighbor-Leader algorithm
Bayadir A. Issa, Abdulmuttalib Turky Rashid
Electrical Engineering Department, University of Basrah & Southren Technical university, Basrah, Iraq

a r t i c l e i n f o a b s t r a c t

Article history: One of the biggest open challenges to the robotic system is controlling robots formation. Its importance
Received 8 April 2020 has been noted in a wide variety of applications, from truck platoons to robot-based standalone parcel
Revised 8 July 2020 delivery. In this paper, a new central approach to controlling a polygon-shaped formation called the
Accepted 1 August 2020
Neighbor-Leader Algorithm is proposed for a group of neighbor robots randomly distributed in an
Available online xxxx
unknown environment. Firstly, a localization procedure for multi-robots is used to estimate the real-
time positions and orientations of a leader robot and its neighbor robots. This information is collected
Keywords:
using an equipped RP LIDAR sensor on each robot then the collected information is sent to the leader
Neighbor-Leader
Formation
one. Secondly, neighbor robots are rearranged into a regular distribution by moving them around the lea-
RP LIDAR sensor der one in circular paths to produce a new distribution with equal angles between each two neighbor
Mobile robot robots. In the last step, the direction of each neighbor robot is changed toward the leader robot. After that,
all neighbor robots move towards the leader until they reach a virtual circle that passes through the
required polygon vertices. The simulation results illustrate the integrity of this algorithm.
Ó 2020 The Authors. Production and hosting by Elsevier B.V. on behalf of King Saud University. This is an
open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction methods of controlling formation are proposed, including behav-


ioral approach, virtual structures, and a leader–follower approach
Mobile robotics formations have received a lot of attention in (Penghao and Zhiyong, 2016). In the behavior-based configuration,
the earlier decade due to their ability to work together and perform group actions (or tasks) consist of several low-level actions (or sub-
multiple types of work as a whole more professionally than one tasks), where individual robots need to perform low-level actions
robot (Chen et al., 2010; Rashid et al., 2015). In addition, they to achieve a group. Actions created to achieve different goals
can provide the strength and the ability to confront the errors (Rashid and Yakub, 2019). The main advantage of this strategy is
expected for military applications and rescue missions. Building that the planned responses are mainly combined by carrying loads
multi-robot systems is less expensive than building a large and of activities that depend on associates responding to the connected
complex single robot (Farias et al., 2016; Rashid, 2016). The forma- robot. However, this approach has drawbacks, the mathematical
tion of multi-robots is evolving rapidly as they are applied in many analysis of this methodology is complex, therefore, the simultane-
fields such as collaborative monitoring and the remote transport ousness of the robotic formation of the desiderated structure can-
architecture of multi-mobile robots (Toibero et al., 2008). These not be approved (Lawton et al., 2003; Marina et al., 2017). In a
tasks require the robots to keep some desired configurations on virtual structure approach, the controller is taken in three stages.
the move. Formation control refers to the problem of controlling In the first stage, the dynamics required for the hypothetical struc-
the relative position and orientation of robots within a group while ture are distinguished. Second, the ideal movement of the virtual
allowing movement as a whole (Issa and Rashid, 2019). Various structure is translated into the desired movement of all robots.
Finally, one next controller for all robots is obtained from robotic
tracking (Sugihara and Suzuki, 1996). The principle favorable posi-
Peer review under responsibility of King Saud University. tion of this methodology is that a solitary scientific guideline inter-
prets the whole tangible info space into the actuator yield space
without the requirement for numerous standards or practices.
Likewise, the obtained behaviors can be consolidated by utilizing
Production and hosting by Elsevier vector activities. A few properties, for example, robustness and

https://doi.org/10.1016/j.jksuci.2020.08.001
1319-1578/Ó 2020 The Authors. Production and hosting by Elsevier B.V. on behalf of King Saud University.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Please cite this article as: B. A. Issa and A. T. Rashid, Multi-robot control for a static polygon formation using Neighbor-Leader algorithm, Journal of King
Saud University – Computer and Information Sciences, https://doi.org/10.1016/j.jksuci.2020.08.001
2 B.A. Issa, A.T. Rashid / Journal of King Saud University – Computer and Information Sciences xxx (xxxx) xxx

stability can be demonstrated utilizing the theoretical tools from sizes specified by the leader robot. The neighbor robots organize
physics, control theory and graph theory. However, the strategy with leader robot by using WIFI network communication, and cre-
of the virtual structure approach is frequently used to structure ate formations. To work in this scheme, the leader robot needs only
aggregate practices that require multi-robot formation which pro- the native data which consists of the neighbor robots positions rel-
duce inaccurate results and cannot be used in many applications ative to it, which is obtained from lidar sensor of each one. The sys-
(Cristescu et al., 2012; Han et al., 2017). In a leader–follower tem consists of one leader robot and four neighbor robots
approach, one robot is organized as a leader, travels along a certain distributed randomly in unknown environment. Each neighbor
path, and other robots are designed as followers (Khalaji and robot have simple capabilities of movement such as rolling wheels,
Zahedifar, 2019). The configuration control issue in a leader–fol- measuring distance and angle through RP LIDAR sensor attached
lower formation turns into two simple problems: one is to control with it. It used to collect information about other neighbor robots
the tracking of leader, and the other is to organize the formation and the environment. RP LIDAR is a full 360 degree of scanning,
with other robots. The tracking robot needs to be shaped according low cost and can take more than 8000 distance data per second
to the leader’s position, maintain the required formation, and the scanning rate 2 to 10 HZ (Rashid and Ali, 2012; Consolini and
adjust its position. Therefore, to adjust the configuration process, Morbidi, 2008). The RP LIDAR is used in environment mapping,
simply select the relative position between the leader and the fol- so, the sensor joined with the mobile robot and control the stations
lower robots (Balasundaram and Muthuswamy, 2017; Rashid and were established to collect the information about the environment.
Ali, 2013) by using a standard control theory techniques which It is more accurate, cover large area and polled up to 100 times per
provide stability of error tracking (Vaitheeswaran et al., 2014). This second, so, it can be used in 2D and 3D unknown environment. All
approach has a simple structure and is easy to implement by two controls of the neighbor robots are attained over the leader robot
controllers, it only relies on the information from local sensors which adjusts their relative positions according to its position
but it has a drawback that a leader robot is a single unit whose and orientation. In addition to mobile capability, the robots have
fault may lead to failure of the whole formation. Leader-follower some communication devices with which they can communicate
formations can take many forms, starting with the line in which with leader robot over WIFI communication network. The struc-
each robot is subordinate to the previous robot and the next robot ture of this paper is as follows: Section 2 defines the related work.
leader. Instead, it can consist of one leader which is followed by all Section 3 defines the basic leader follower algorithm based on
other robots. Combining these two shapes leads to different geo- them. Section 4 explains the suggested procedures. Section 5
metrical shapes and can be distorted to avoid obstacles or pass explains the simulations results. And finally, Section 6 draws the
through narrow distances. The most sophisticated method of orga- conclusions of the paper.
nizing robots includes replacing the leader with the main system,
monitoring the robot, and giving tasks to the robot (Darmanin
and Bugeja, 2017). Two basic methods are used to maintain the for- 2. Related works
mation of multi mobile robots based on the leader–follower
approach named distance-orientation method in which the fol- Over the years, several common areas of application have been
lower must maintain a fixed distance and orientation angle proposed in Leader-Follower Formation. For example, surveillance,
between it and the leader in the organize structure. And the second search, and rescue applications that have received much attention
one namely distance-distance method which is based on the scope from the multi-robotics community for many years. This is due to
of sensing of mobile robots to control the relative distance between the importance of these applications in daily life. The monitoring
the followers and the leader one. With this condition, each follower application was originally reserved for patrols or surveys of inte-
must follow farther leader robots to obtain the desired formation rior areas. The quest for food is often synonymous with a group
shape and more communication network are required (Markom of robots inspired by natural colonial systems such as bees and
et al., 2015; Wei and Cagle, 2018). ants. This is because distributed teams often include only an impli-
This work concentrates on controlling the formation of multi- cit connection to collect randomly distributed items and transfer
mobile robots systems. A new algorithm to control multiple mobile them to their ‘‘home” sites. In configuration applications, robotics
robots that can self-assemble in a specific configuration is pro- teams must maintain a strict position while avoiding obstacles in
posed. This is one of the most important features to complete tasks their path. This problem is more complicated than the flock, as
that require cooperation. The ability to form configurations is espe- team members must avoid obstacles in large quantities unless they
cially important when the capabilities of individual robots are lim- leave the formation for an extended period of time. A common
ited. For example, robots may coordinate for synchronized search solution is the Leader-Follower approach, where the path planning
and rescue activities in the event of a disaster, collectively trans- algorithm is implemented on the commanding robot and the con-
port large objects, direct unknown areas, and plan and maintain straints of the configuration (distance from the leader) are fol-
configurations to construct buildings, introduce a new control lowed by the followers to maintain the training around them.
and management system for steadily organizing mobile robots in Recent research has also adopted computational intelligence meth-
the shape of the static polygon. More efficient use of network com- ods in planning applications. In (Schlanbusch et al., 2010) an algo-
munication capacity and reducing energy consumption because of rithm has been proposed to create a leader-followers formation
the communication only between the leader and each neighbor assuming that the required angle between the leader and followers
robot, so, the physical components of this process is reduced. The is measured within the followers rather than the leader which
only leading site in our system known to external user control sys- manages more time and higher costs for implementation. In
tems that release appropriate mobile software agents in the envi- (Yang and Fan, 2019) a leader–follower formation is presented
ronment to create the required formation. In most other with the assumption that the followers complete the leader’s mea-
configuration control algorithms, each robot needs information surement by tracking the point between the default point and the
about the entire configuration, which is not acceptable in real leader’s center at a fixed distance. The work is based on the 6DOF
applications. In our algorithm, on the other hand, only the leader model, where each follower generates a real-time reference based
robot needs the native neighbor information which is obtained on the relative position and translation between the leader and its
from RP LIDAR laser sensor attached to it. An iterative procedure followers. This ensures that the leader points to the target while
refines the relative positions of the neighbor’s robots in the collec- reconstructing the formation. This method has complicated and
tion to create the final formation with the exact polygon shape and inaccurate results. While in (González-Sierra et al., 2018) a

Please cite this article as: B. A. Issa and A. T. Rashid, Multi-robot control for a static polygon formation using Neighbor-Leader algorithm, Journal of King
Saud University – Computer and Information Sciences, https://doi.org/10.1016/j.jksuci.2020.08.001
B.A. Issa, A.T. Rashid / Journal of King Saud University – Computer and Information Sciences xxx (xxxx) xxx 3

distributed nonlinear formation controller has been proposed so that, the robots form a static polygon formation by stabilizing
that the control inputs for each follower can be represented as a the distance between the leader and each robot on the circumfer-
product of the nonlinear term based on the error of distance under ence of the circle and rotating them in the same direction as the
the leader–follower structure. With this type of leader–follower leader.
structure, a number of robots are considered leaders, direct the
groups of other robots to specific destinations, which are called fol-
3. Leader-Follower algorithm
lowers. The stability of this control law is demonstrated through
the use of the candidate Lyapunov functions. And (Solot and
This algorithm draws revelation from (Farias et al., 2016) the
Ferlini, 2019) introduced a control strategy to form a leading-
leader–follower algorithm and creates a static polygon shape, as
follower with a distance dependent model of two kinematic mod-
(Rashid et al., 2019). For this reason, this section briefly presents
els of robots. Use bouncing techniques to control the follower’s
the concept of (Farias et al., 2016) leader–follower formation and
robots to maintain the required distance and direction with the
the configuration of the static polygon shape. The circular paths
robot leader. In (Qian and Xi, 2018) distributed technology is used
algorithm is used to predict the path of the follower robot that
to control the leader–follower formation of three turtle robots,
depends on the path of the leader one, and noting the leader posi-
each of which makes data-driven options provided by the onboard
tion to follow it. In this algorithm, the follower robot track contain
lidar and works as needed to track a human or other robot. A
numerous circular paths everyone is attained from three points
derivative integral (DI) station sliding mode control for leader–fol-
characterized by the prospect position of leader robot, the previous
lower formation is structured in (Yang et al., 2019) dependent on
and present positions of follower robot. Initially, the distance (L)
the relative pose. Since the formation organization experiences
among the leader and follower robots is calculated and at every
vulnerabilities, the nonlinear disturbance viewer is adopted to
step of movement the size of the motion step is rose by specific
manage the vulnerabilities. The sense of lyapunov proven the sta-
issue (l) as the distance between leader and follower is larger than
bility of the closed-loop control system. Another control structure
the partial of L, other its reductions by the similar issue l. While the
is proposed instead of the traditional back stepping method in
approach presented in (Rashid et al., 2019) is to localize multi-
(Chinelato et al., n.d.) in which an essential robot is structured to
robots through the organized formation. Formatting the configura-
such that it converges to the reference direction dependent on
tion allows one to localize the mobile robot with three static robots
the pose requirement of the leader’s without the need of leader’s
that act as source nodes. This mobile robot will be fixed once the
velocity and dynamics. Also lyapunov stability analysis is used to
step is completed and it will be used as one of the source nodes.
show that all closed-loop signals are uniformly ultimately
One fixed robot is selected and you can switch to the next mobile
bounded. While in (Tutuko et al., 2018) a control method called
robot. By repeating this process, the static formation is achieved
an SDRE (State-Dependent Riccati Equation) is proposed to control
without an external source signal. In the configuration format,
the formation of a multi-robot system with one leader robot and
the mobile robot moves through the shifting and switching
two followers. And in (Kambayashi, et al., 2019) an estimator is
sequences. In the shift sequence, each robot moves to the desired
proposed to use only the limited posture amounts to evaluate
path. In the switching sequence, the multi-robot is started and fol-
the relative position. A controller dependent on the assessed rela-
lows the desired path. As each robot moves, it must remain within
tive posture in the follower coordinate structure is proposed for
the range of the infrared sensor. In this section, formation coordi-
the unicycle-type mobile robots. In (Trinh et al., 2018), an algo-
nation will be used as a strategy to create a regular polygon shape.
rithm was proposed to control the formation of swarm robots
Every robot in a multi-robot system gets its data from its own sen-
using a three-dimensional moving agent. Mobile agents transport
sor and by communicating with other robots. First, all robots use a
physical robots and use indirect communication by spreading
block-matching algorithm to estimate the initial position and
pheromones. The spread of pheromones is also achieved by mobile
direction. The strategy then begins to choose the appropriate
agents. The algorithm applies two types of mobile software agents:
robots to move depending on the shape and location of these
ant and pheromone. Ants use local information to generate a pher-
robots. This process is accomplished by a block-matching algo-
omone. Local information consists of vector values that determine
rithm that selects robots that are far from the polygon formation
where robots occupy the neighborhood to form the required for-
site. In triangulation, robot motion estimation is performed by at
mation. This type of algorithm was expensive and required more
least three fixed robots, but the proposed strategy requires only
communication between all robots. In (Rashid and Ali, 2019) a
two robots. This reduces the number of transitional events when
direction established formation control of a gathering of self-
building a polygon formation. Each robot uses the algorithm of
sufficient robots with the leader first follower (LFF) structure in
the diagram of the binary tree to determine its path. These robots
an open dimensional space. Posture-just control laws that all-
stop at the boundaries of the communication range of the neigh-
inclusive stabilize LFF formations to the preferred formation are
boring fixed robot. Then another distant robot begins moving to
proposed and examined. Several static strategies for polygon shape
target formation and repeats the same procedure. This process
formation in global knowledge environments are implemented to
continues until all robots reach their targets.
minimizing the energy spent by the robots or the time for achiev-
ing the task have been investigated. These strategies use the clus-
ter matching algorithm instead of the triangulation algorithm 4. Neighbor-Leader algorithm
(Rashid et al., 2019).
The purpose of this paper is to present a different algorithm to In controlling formation algorithms, it is typically expected that
form leaders- followers’ static polygon formation based on a sim- ideal estimation data is accessible for the robots in the formation.
plification of orientation – distance algorithm for a group of mobile In a multi-robotic system, where maintaining the shape of the
robots randomly distributed around a leader in an unknown envi- formation is important, if the neighboring robots do not have fully
ronment. Each robot is equipped with an RP LIDAR sensor used to tuned sensors, then the strength of the required shape will not
collect information about the environment and other robots. The stabilize the group formation. Specifically, the configurations
collected information from RP LIDAR sensors is sent to the leader show undesirable group motion as well as desired shape distortion.
robot through direct contact among the leader one with other fol- This problem can persist if two adjacent robots do not bear the
lowers. Therefore, the leader can localize each robot to the correct same responsibility. For example, only one robot controls the error
position according to its location with respect to the leader. After distance between two adjacent robots in master and slave

Please cite this article as: B. A. Issa and A. T. Rashid, Multi-robot control for a static polygon formation using Neighbor-Leader algorithm, Journal of King
Saud University – Computer and Information Sciences, https://doi.org/10.1016/j.jksuci.2020.08.001
4 B.A. Issa, A.T. Rashid / Journal of King Saud University – Computer and Information Sciences xxx (xxxx) xxx


configuration which is called leader–follower configuration. The xi ¼ x0 þ Di  cos li ð2Þ
leader- follower formation leads to a directed topology that

describes the relationships between neighboring robots and ana- Y i ¼ y0 þ Di  sin li ð3Þ
lyzes the main properties of configuration control algorithms such
as stability, gravity, and convergence time. An analysis of the dis- Here, (x0, y0) are the coordinate axis of the leading robot, and (xi,
tribution of responsibilities between neighboring robots showed yi) are the initial coordinate axis of the robot i.
that increasing the number of robots in a team requires additional Step2: Change the neighbor robots orientation: In this step
costs to coordinate all sensor pairs on neighboring robots. In this each robot is located at circular trajectory around the leader robot
section, Neighbor-Leader algorithm to form a static polygon shape as shown in Fig. 2. The directions of the robots must be changed to
formation is introduced. This algorithm is used to predict the tra- align with the circular path on which the robots are located. This
jectories of the neighbor robots dependent on the posture of the process is implemented according to the following steps:
leader one. It takes into account the observed location of the leader
robot to go toward it. At first, the positions and the directions of 1. The neighbor robot i scan the environment to correct the angle
neighbor robots are calculated according to the leader position ai (Fig. 3) between its direction and the leader robot location.
and at each step of the neighbor robot movement the length of 2. Change the orientation of the neighbor robot i by j degrees.
the movement step is increases by a factor (j) until the distance
bi ¼ bi þ j ð4Þ
between leader and neighbor robots becomes equal to the radius
of the circle where the target of each robot is around the leader
3. Steps 1 and 2 are repeated until ai = 90 degrees.
to form the static polygon shape. The investigation of this algo-
rithm requires two parts: Step3: Rearrange the distance among the neighbor robots:
The uniform formation of the neighbor robots occur when the
4.1. Part one: Rearrange the angles among the neighbor robots angles between any two neighbor robots with respect to the leader
robot have the same value. This process is implemented according
In this part, the robots are randomly distributed. The coordina- to the following steps:
tion axis and directions for each robot are calculated relative to the
leader robot. This position information can be obtained from the RP 1- Compute the angle q between each two neighbor robots
LIDAR sensor installed on each robot. RP LIDAR provides direction with respect to the leader one.
and distance information for each robot in the path. From this
information, the position of each robot can be calculated using q ¼ 2p=n ð5Þ
the following procedure: where n is the number of the neighbor robots.
Step1: Computation of initial position: The coordinates and
directions of each robot adjacent to the leader robot must be calcu- 2- Compute the angles between the neighbor robots and the
lated according to the following procedure as shown in Fig. 1: leader one as shown in Fig. 4.

1. The leader RP LIDAR scans the environment to calculates bi (the ci ¼ tan1 ððyi  yo Þ=ðxi  xo ÞÞ ð6Þ
rotation angle between the leader direction and the robot i
position) and Di (the distance among the leader robot and the 3- Rearrange the angles between the neighbor robots and the
robot i). leader one in ascending manner.
2. Calculate the direction of the robot i with respect to the leader 4- Compute the orientation angle between each two neighbor
location as: robots.

li ¼ ui þ hi ð1Þ

where ui is the orientation of the leader robot.

3. The coordinates of the robot i are computed as follows:

Fig. 2. Each neighbor robot is located at a circular trajectory around the leader
Fig. 1. Computation of initial position. robot.

Please cite this article as: B. A. Issa and A. T. Rashid, Multi-robot control for a static polygon formation using Neighbor-Leader algorithm, Journal of King
Saud University – Computer and Information Sciences, https://doi.org/10.1016/j.jksuci.2020.08.001
B.A. Issa, A.T. Rashid / Journal of King Saud University – Computer and Information Sciences xxx (xxxx) xxx 5

4.2. Part two: rearranged the distances between the neighbor robots
and the leader

In this part, the robots must change their orientation toward the
leader one and move until reach to the circular path that pass
through the desired polygon shape as in the following steps:
Step1: Change the neighbor robots orientation: At the end of
the first stage, all the neighbor robots are arranged in uniform
angles with each other and have different distances from the leader
one. The direction of each neighbor robot is perpendicular to the
leader location. These orientation must be changed by 90 degrees
to be in direct path to the leader robot location. This process imple-
ments according to the following steps:

1- According to Fig. 3 the orientation of the neighbor robot i is:

ri ¼ biþ li ð11Þ

2- The neighbor robot i orientation is increased by j at each


step.
Fig. 3. Each neighbor robot orientation must be in perpendicular to the leader
location. ri ¼ ri þ j ð12Þ

3- The process is repeated until bi = 180 degrees as shown in


Fig. 6.
Step2: Complete the polygon shape construction: After
changing the neighbor robots orientation, The next step is to move
these robots toward the leader one at distances represented by the
circular path that run through the vertices of the polygon shape.
This process is implemented according to the following steps:

1. Compute the new position of the neighbor robot i

xiþ1 ¼ xi þ m  cosri ð13Þ

yiþ1 ¼ yi þ m  sinri ð14Þ

2. Compute the distance between the neighbor robots and the lea-
der as:
 
Di ¼ sqrt ðyi  yo Þ2 þ ðxi  xo Þ2 ð15Þ

Fig. 4. Computing the angles between the neighbor robots and the leader one.

#i ¼ ciþ1  ci ð7Þ
where #i is the angle between the robots i and i + 1.

5- If the angle #i is less than the angle q then increase the angle
ci.

ci ¼ ci þ j ð8Þ

6- Compute the new position of neighbor robot i.

xiþ1 ¼ xi þ m  cosðci Þ ð9Þ

yiþ1 ¼ yi þ m  sin ðci Þ ð10Þ

7- Repeat steps 5 and 6 for current neighbor robot and for all
robots until all the angles between the neighbor robots
and the leader one are equal and have the value q#i ¼ q as
shown in Fig. 5. Fig. 5. Organize neighbor robots with equal distances and angles around the leader.

Please cite this article as: B. A. Issa and A. T. Rashid, Multi-robot control for a static polygon formation using Neighbor-Leader algorithm, Journal of King
Saud University – Computer and Information Sciences, https://doi.org/10.1016/j.jksuci.2020.08.001
6 B.A. Issa, A.T. Rashid / Journal of King Saud University – Computer and Information Sciences xxx (xxxx) xxx

Fig. 6. Direct the orientation of each neighbor robot toward the leader one.
Fig. 7. Arrange the orientation angles between the neighbor robots as the leader
orientation.

3. Repeat 1, 2 until Di = L as shown in Fig. 7.where L represents the


formation distance between the neighbor robots and the leader
as shown in Fig. 6.
Step3: Complete the neighbor robots orientation: Correct the
orientation of all neighbor robots to make their directions as the
leader one by the following equations and as shown in Fig. 7 and
the complete static polygon formation shown in Fig. 8:

p
#i ¼  li ð16Þ
2

@ i ¼ p þ #i þ u i

bi ¼ bi þ p ð17Þ

5. Simulation results

In this paper, Simulation of the proposed algorithm is used to


investigate the Neighbor-Leader formation of mobile robots using
Visual Basic 2010. The configuration strategy is simulated in an
unknown environment and is explored as a static polygon forma-
tion. The algorithm suggested in this paper is compared with the
leader–follower algorithm in (Farias et al., 2016) and static polygon Fig. 8. Static polygon formation.
formation in (Rashid et al., 2019). The formation with the leader–
follower algorithm is implemented in the global knowledge
environment, where each robot knows its location including the  Number of neighbor robots (n): this metric is used to measure
distances to the other robots to predict the trajectory of the fol- the time required to complete the static polygon formation
lower robot dependent on the trajectory of the leader one. While when increasing the number of robots.
the other formation using two algorithms: power saving algorithm  The formation completing time (t): this metric is used to mea-
which is based on the permutation algorithm to determine the tar- sure the percentage of the completing time with respect to
get of each robot. The ideal arrangement occurs in the case when the number of the neighbor robots.
the sum of the distances between robots and goals is the smallest.
Furthermore, the shortest time formation algorithm is also imple- Fig. 9 shows the simulation. Fig. 9(a)–(f) represent the Screen-
mented in global knowledge environment and using permutation shots of simulations for our algorithm at different time steps to
algorithm. The best arrangement in this case occurs when the max- investigate the static polygon formation in unknown environment.
imum distance in this arrangement is the minimum with respect to The main goal of this simulation is to show the relation between
the other arrangements. The simulations are performed using a dif- the number of robots and the accomplishment percentage.
ferent topology representing different network sizes (n) from 3 to 7 Figs. 10–12 show the comparison among the Neighbor Leader,
robots. The robots are randomly placed in a space of 500  500 the leader–follower and the save energy formation algorithms to
pixels. In this simulation, two parameters are used. compute the percentage of the formation completing time in a

Please cite this article as: B. A. Issa and A. T. Rashid, Multi-robot control for a static polygon formation using Neighbor-Leader algorithm, Journal of King
Saud University – Computer and Information Sciences, https://doi.org/10.1016/j.jksuci.2020.08.001
B.A. Issa, A.T. Rashid / Journal of King Saud University – Computer and Information Sciences xxx (xxxx) xxx 7

Fig. 9. The formation of the robots. (a-e) Screenshots of the simulations at different time steps. (f) Final formation.

Fig. 10. Comparison among the Neighbor- Leader, the leader follower and the save energy formation algorithms to compute the percentage of the formation completing time
in astatic environment with three neighbor robots.

Please cite this article as: B. A. Issa and A. T. Rashid, Multi-robot control for a static polygon formation using Neighbor-Leader algorithm, Journal of King
Saud University – Computer and Information Sciences, https://doi.org/10.1016/j.jksuci.2020.08.001
8 B.A. Issa, A.T. Rashid / Journal of King Saud University – Computer and Information Sciences xxx (xxxx) xxx

Fig. 11. Comparison among the neighbor leader, the leader follower and the save energy formation algorithms to compute the percentage of the formation completing time
in astatic environment with five neighbor robots.

Fig. 12. Comparison among the Neighbor Leader, the leader follower and the save energy formation algorithms to compute the percentage of the formation completing time
in astatic environment with seven neighbor robots.

static environment with three, five and seven neighbor robots The main goal of this simulation is to show the relation between
respectively. The comparisons show that as the number of robots the number of robots and the accomplishment time.
increases in the environment, the accomplishment percentage From Fig. 14 It is found that as the number of the neighbor
increases, too. The Neighbor-Leader formation algorithm has the robots increases, the execution time also increases for all the
optimal performance comparing to the other two algorithms. neighbor leader, the leader follower and the save energy formation
Fig. 13 shows the screenshots of simulations with different number algorithms. The Neighbor Leader formation algorithm produces the
of robots. This process is achieved by using the Neighbor-Leader optimal performance comparing to the other two algorithms.
algorithm to move far robots from their unknown places to their The last simulation is used to compare the total paths that used
targets to shape the Neighbor-Leader static polygon formation. by all the robots to complete the formation for all the Neighbor

Please cite this article as: B. A. Issa and A. T. Rashid, Multi-robot control for a static polygon formation using Neighbor-Leader algorithm, Journal of King
Saud University – Computer and Information Sciences, https://doi.org/10.1016/j.jksuci.2020.08.001
B.A. Issa, A.T. Rashid / Journal of King Saud University – Computer and Information Sciences xxx (xxxx) xxx 9

Fig. 13. The multi-robot formation. (a, b) three robots (c, d) five robots (e, f) six robots.

Fig. 14. Time of formation for different number of robots.

Leader, the leader follower and the save energy formation algo- the save energy formation algorithms to compute the average of
rithms. Minimizing the total path means reducing in the power the total path for different number of the neighbor robots. The
used by these robots to reach to their targets. Fig. 15 shows the comparisons show that as the number of the robots increase in
comparison among the neighbor leader, the leader follower and the environment, the average of the total path increase, too. The

Please cite this article as: B. A. Issa and A. T. Rashid, Multi-robot control for a static polygon formation using Neighbor-Leader algorithm, Journal of King
Saud University – Computer and Information Sciences, https://doi.org/10.1016/j.jksuci.2020.08.001
10 B.A. Issa, A.T. Rashid / Journal of King Saud University – Computer and Information Sciences xxx (xxxx) xxx

Fig. 15. Comparison the average of the total path for the neighbor leader, the shortest time and the save energy formation algorithms with respect to the number of the
neighbor robots.

Fig. 16. The average value and standard deviation of the distance between the robots positions and their target.

save energy formation algorithm has the best performance than 6. Conclusions
the other two algorithms. For a more measurable evaluation, sim-
ulations in this situation have been recurrent for an altered dis- In this paper, a novel organizing remote adjacent robots forma-
tances, extending from 0 to 200 at steps of 5. For each average tion strategy for the static polygon shape formation is proposed in
value of distance at a time to reach the target, 10 different realiza- an unknown environment by using several numbers of mobile
tions have been considered, where at each realization a different robots localized and distributed randomly. Simulation results are
scattering of the neighbor robots in the environment has been implemented in an environment with a different number (3 to 7)
taken into account. For each static polygon formation algorithm, of robots using the Neighbor-Leader, the leader follower and the
average value and standard deviation of the distance between save energy formation algorithms. The results show that the
the robots positions and their target have been calculated and Neighbor-Leader algorithm has a better efficiency to complete
described in Fig. 16. The presented algorithm has better perfor- the formation since it uses the Neighbor-Leader algorithm to deter-
mances than the other two algorithms. mine the execution time and the accomplishment percentage. The

Please cite this article as: B. A. Issa and A. T. Rashid, Multi-robot control for a static polygon formation using Neighbor-Leader algorithm, Journal of King
Saud University – Computer and Information Sciences, https://doi.org/10.1016/j.jksuci.2020.08.001
B.A. Issa, A.T. Rashid / Journal of King Saud University – Computer and Information Sciences xxx (xxxx) xxx 11

other algorithms depend on several algorithms to perform the Cristescu, S., Ionescu, C., et al., 2012. Leader-Follower String Formation using
Cascade Control for Mobile Robots, Barcelona, Spain, pp. 1092–1098, pp. 3–6.
same task which needs more computation time. From the results,
Han, Q., Ren, S., et al., Bearing-Based Localization for Leader-Follower Formation
it is found that the accomplishment percentage is increasing as Control, PLOS J. Long Wang, Peking University, China, April 20, 2017, pp. 1–16.
the number of neighbor robots increase for all algorithms. Also, Khalaji, A.K., Zahedifar, R., 2019. Lyapunov-Based Formation Control of Underwater
as the number of robots increases, the execution time also Robots, Robotica in Cambridge University, pp. 1–18.
Balasundaram, M., Muthuswamy, S., 2017. Systematic approach for role assignment
increases and the Neighbor-Leader algorithm has a better effi- in mobile robots using leader-follower approach: experimental observations.
ciency than the other algorithms. In computing the average of Int. J. Rob. Autom. 32 (3), 224–232.
the total path for completing the formation for all the Neighbor Rashid, A.T., Ali, A.A., et al., 2013. Path planning with obstacle avoidance based on
visibility binary tree algorithm. Rob. Auton. Syst. 61, 1440–1449.
Leader, the leader follower and the save energy formation algo- Vaitheeswaran, S., B.M.K, et al., 2014. Leader follower formation control of ground
rithms, it is found that as the number of the robots increase in vehicles using camshaft based guidance, Int. J. Multimedia Its Appl. (IJMA) 6 (6).
the environment, the average of the total path increases, too. The Darmanin, R., Bugeja, M., 2017. A review on multi-robot systems categorized by
application domain. In: 25th Mediterranean Conference on Control and
save energy formation algorithm has the optimal performance Automation, pp. 1–7.
than the other two algorithms because this algorithm chooses Markom, M., Adom, A., et al., 2015. Mobile Robotics Mapping using RP LIDAR
the best distribution among the robots and their target locations. Scanner. In: 2015 IEEE International Symposium on Robotics and Intelligent
Sensors (IRIS), October 2015, pp. 1–7.
Wei, P., Cagle, L., 2018. LiDAR and camera detection fusion in a real-time industrial
Declaration of Competing Interest multi-sensor collision avoidance system, MDPI Electronics J., 7 (84) 1–32.
Rashid, A.T., Ali, A.A., et al., 2012. Multi-robot collision-free navigation based on
reciprocal orientation. Rob. Auton. Syst. 60 (10), 1221–1230.
The authors declare that they have no known competing finan-
Consolini, L., Morbidi, F., et al., 2008. Leader-follower formation control of
cial interests or personal relationships that could have appeared nonholonomic mobile robots with input constraints. Automatica 44 (5),
to influence the work reported in this paper. 1343–1349.
Schlanbusch, R., Kristiansen, R., et al., 2010. Attitude Reference Generation for
Leader-Follower Formation With Nadir Pointing Leader, in: The American
References Control Conference, June 2010, pp. 1–7.
Yang, Xiaowu, Fan, Xiaoping, 2019. A distributed formation control scheme with
Chen, Jian, Sun, Dong, Yang, Jie, Chen, Haoyao, 2010. Leader-follower formation obstacle avoidance for multiagent systems. Mathematical Problems Eng. 2019,
control of multiple non-holonomic mobile robots incorporating a receding- 1–16.
horizon scheme. Int. J. Robotics Res. 29 (6), 727–747. González-Sierra, J., Hernández-Martínez, E.G., Ferreira-Vazquez, Enrique D., Flores-
Rashid, A.T., Frasca, M., Ali, A.A., Rizzo, A., Fortuna, L., 2015. Multi-robot localization Godoy, J.J., Fernandez-Anaya, G., Paniagua-Contro, P., 2018. Leader-follower
and orientation estimation using robotic cluster matching algorithm. Rob. control strategy with rigid body behavior. IFAC-PapersOnLine 51 (22), 184–189.
Auton. Syst. 63, 108–121. Solot, A., Ferlini, A., 2019. Leader-Follower Formations on Real Terrestrial Robots.
Farias, G., Fabrega, E., et al., 2016. Development of an easy-to-use multi-agent Association for Computing Machinery, Beijing, China.
platform for teaching mobile robotics. IEEE Trans. J. 4, 1–13. Qian, D., Xi, Y., 2018. Leader–Follower formation Maneuvers for multi-robot
Rashid, A., 2016. Leader follower tracking with obstacle avoidance using circular systems via derivative and integral terminal sliding mode, MDPI J.
paths algorithm. Basrah J. Eng. Sci. 16 (2), 29–47. Yang, Z., Zhu, S., Chen, C., Feng, G., Guan, X., Leader-follower formation control of
Toibero, J., Roberti, F., et al., 2008. Formation control for non-holonomic mobile nonholonomic mobile robots with bearing-only measurements, J. Franklin
robots: a hybrid approach, Recent Adv. Multi Robot Syst., INTECH J., 234–248, Institute, No. 7, 2019.
www.intechopen.com. Chinelato, C., Filho, L., S D R E based Leader - Follower Formation Control of Multiple
Issa, B., Rashid, A., 2019. Survey of multi-mobile robot formation control. Int. J. Mobile Robots, Tema J. 15 (2) 195–202.
Computer Appl. 181 (48), 0975–8887. Tutuko, B., Saparudin, S.N., Fitriana, G., 2018. Enhancement of non-holonomic
Penghao, W., Zhiyong, G., 2016. Leader-Follower Formation Control of Multi-Robot leader-follower formation using interval type-2 fuzzy logic controller, IJOE – 14
Systems Using the Dynamic Surface Approach. In: Proceedings of the 35th (9).
Chinese Control Conference, Chengdu, China, July 27–29, 2016, pp. 7757–7762. Kambayashi, Y., Yajima, H., et al., 2019. Formation control of swarm robots using
Rashid, M., Yakub, Z., et al., 2019. Comprehensive review on controller for leader- mobile agents, Vietnam J. Computer Sci., 10 (1142) 1–30.
follower robotic system. Indian J. Geo Marine Sci. 48 (7), 985–1007. Trinh, M., Zhao, S., Sun, Z., Zelazo, D., Anderson, B., Ahn, H., 2018. Bearing-based
Lawton, J.R.T., Beard, R.W., Young, B.J., 2003. A decentralized approach to formation formation control of a group of agents with leader-first follower structure, IEEE
maneuvers. IEEE Trans. Robot. Automat. 19 (6), 933–941. Trans. Automatic Control.
Marina, H.G., Jayawardhana, B., et al., 2017. Distributed algorithm for controlling Rashid, A., Ali, A., 2019. Polygon shape formation for multi-mobile robots in a local
scale-free polygonal formations. In: IFAC 2017 World Congress, Toulouse, knowledge environment, Basrah J. Eng. Sci. 19 (2) 39–46.
France. Rashid, A., Ali, A., et al., 2019. Polygon shape formation for multi-mobile robots in a
Sugihara, Kazuo, Suzuki, Ichiro, 1996. Distributed algorithms for formation of global knowledge environment, Iraqi J. Electrical Electronic Eng. 15 (1) 76–88.
geometric patterns with many mobile robots. J. Robotic Syst. 13 (3), 127–139.

Please cite this article as: B. A. Issa and A. T. Rashid, Multi-robot control for a static polygon formation using Neighbor-Leader algorithm, Journal of King
Saud University – Computer and Information Sciences, https://doi.org/10.1016/j.jksuci.2020.08.001

You might also like