Bitlis Eren Üniversitesi Fen Bilimleri Dergisi

BİTLİS EREN UNIVERSITY JOURNAL OF SCIENCE

ISSN: 2147-3129/e-ISSN: 2147-3188
VOLUME: 13 NO: 1 PAGE: 192-204 YEAR: 2024
DOI:10.17798/bitlisfen.1370223
Design and Implementation of an Optimized PID Controller for Two-
Limb Robot Arm Control
Said MÜFTÜ1*, Barış GÖKÇE2
1
Pamukkale University, Department of Mechatronics Engineering, Denizli,20160, Türkiye
2
Necmettin Erbakan University, Department of Mechatronics Engineering, Konya, 42090, Türkiye
(ORCID: 0000-0001-5621-7805) (ORCID: 0000-0001-6141-7625)

Keywords: DC motor, Abstract

Optimization, PID Control, Advances in technology have increased the use of robot arms and led to more
Robot Arm. research and development on robot arms. Controllability, which is the main focus of
the studies on robot arms, generally provides speed and precision to robot arms. In
this study, a two-limbed robot arm is controlled using the MATLAB support package
for Arduino Hardware, and a microcontroller is used to optimize the control of this
robot arm with a PID controller. In addition, in the direct current (DC) brushed motor
system, the transfer function was obtained using values from the motor data sheet.
Feedback is provided to this control system with a Hall-effect encoder. For the square
reference tracking of the gripper end of the two-armed robot arm, the controller
parameters were obtained by particle swarm (PSO), artificial bee colony (ABC), and
chaos game optimization (CGO) algorithms, and these parameters were applied to
the robot arm. The CGO algorithm, which is one of the methods in the literature and
has become popular in recent years, was used for the first time to determine the PID
parameters. It has been shown that the CGO algorithm can be used to determine the
coefficients of the PID controller.

1. Introduction determining the coefficients that make up this

controller has created a new problem.
Two-limbed robotic arms are a widely used PID control is used to minimize the error
manipulator in industry for many different purposes. between the reference and the measured value
It can perform complex work with rotational determined in robotic systems [2]. In a study
movement at two separate connection points. With conducted in 2001, it was concluded that the rate of
this robotic system, which can work quickly, using a PID controller for a system that requires
precisely, and without fatigue, many applications in control is more than 90%. It was observed that this
the industry, such as material handling, mass rate decreased by 50% in 2017, but it is still the most
production in automotive, and welding processes, can preferred controller [3]. PID controllers, which date
be realized. The importance of these robot arms, back to the 1890s, are generally used to control robot
which save manpower, is increasing day by day, with arms [4].
the ability to do sensitive work at the same time. There are proportional (𝑘𝑝 ), derivative (𝑘𝑑 ),
The use of robot arms in industry started with and integral (𝑘𝑖 ) constants in the PID controller. With
a crane-like robot made by P. Taylor in 1937 [1]. the proportional gain coefficient (𝑘𝑝 ), the error in the
Since 1937, there have been many developments in process output can be directly controlled; with the
robotic systems. Studies on the controllability of derivative coefficient (𝑘𝑑 ), it can control the rate of
robot arms, which is one of these developments, have change of the error, and the integral (𝑘𝑖 ) controls the
increased. Many different controllers have been sum of the error over time. It is aimed at obtaining
developed for the control of robot arms, and the closest result to the reference determined by these

*
Corresponding author: [email protected] Received: 02.10.2023, Accepted: 08.03.2024

192
 S. Müftü, B. Gökçe / BEU Fen Bilimleri Dergisi 13 (1), 192-204, 2024

three coefficients that make up the controller. contribute to existing studies with this study. It is seen
If these coefficients are chosen smaller or larger than as a great advantage that the CGO algorithm has
the required value, the targeted result cannot be certain parameters, such as other optimization
achieved, and the control performance can be algorithms based on swarm intelligence, and it does
increased if the correct parameter is selected. not contain any parameters other than these. The
Therefore, determining the coefficients of the PID parameters found were tested on the realized robot
controller is of great importance for the robot to arm, and the results were compared both as a
perform the desired task. simulation and on the real robot arm.
There are many methods used to determine
PID control parameters. In their most general form, 2. Material and Method
classical techniques can be classified as analytical,
parametric, frequency response, adaptive tuning, and The two-legged robot arm is the simplest robotic arm
metaheuristic algorithms such as the Ziegler-Nichols in robotic systems. Two separate drive elements are
and Cohen Coon methods. Metaheuristic required to perform. In this study, a 1524006SR DC
optimization algorithms can be divided into swarm motor from Faulhaber company, an L298 motor
intelligence, Evolutionary Immunity, neural, driver relay to drive motors, and an Arduino Mega as
probability, and physics-related algorithms [4]. In this a microcontroller were used. The parts of the designed
study, swarm intelligence algorithms are used to solve robot arm were obtained from the 3D printer. The
the problem of determining the parameters of the PID designed two-limbed robot arm can be seen in Figure
controller. 1 a-), and the produced two-limbed robot arm can be
In this study, a PID controller was designed seen in Figure 1 b-).
for the control of the two-limb robot arm, and it was The 1524006R model of Faulhaber has a DC
aimed at obtaining the most appropriate controller motor with a 76:1 rotation ratio and a 2-channel field
parameters with the PSO, ABC, and CGO algorithms, effect encoder. This brand has clearly presented all
which are metaheuristic optimization algorithms. the necessary parameters to its users to obtain the
PSO and ABC have been used in studies on existing transfer function of the DC motor required for control.
robotic system control, but it is thought that CGO will

Figure 1. Two-limbed robotic arm (a) Designed two-limbed robot arm (b) two-limbed robot arm produced.

The distance between the shaft of the motor of arm length can be drawn, while the gripper end of
fixed to the floor of the robot arm and the motor the robot arm can be guided in any way between the
connected to the gripper end is L1, 17.75 cm, and the two-limb system and an outer circle with a radius of
other has a length of 8.25 cm. With an arm connected L1+L2 and an inner circle with a radius of L1-L2. The
to the end of a single motor, only a circle with a radius working area of the robot arm is shown in Figure 2.

193
 S. Müftü, B. Gökçe / BEU Fen Bilimleri Dergisi 13 (1), 192-204, 2024

Figure 2. The working area of a two-limbed robotic arm.

2.1. Controller Design of a Two-Limb Robot Arm the position and direction change with the values
given to the variables, while the inverse kinematics
In robotic systems, there are two mathematical approach allows the values of the variables to be
approaches to moving the end point of the robot arm obtained for the robot arm to reach a certain position.
from the starting point to a determined point. These Inverse kinematic analysis is of great importance for
are expressed as forward and reverse kinematics. The the control of the robot. The schematic representation
forward kinematics approach allows us to calculate of the robot arm realized in the study is given in
Figure 3.

Figure 3. Schematic representation of the two-limb robot arm.

The geometric kinematic analysis of the two- 𝑦 = 𝑙1 𝑠𝑖𝑛𝜃1 + 𝑙2 𝑠𝑖𝑛(𝜃1 + 𝜃2 ) (2)

legged robot arm can be easily calculated. The
projection along the x-axis in equation (1) is indicated Kinematically, equations (1) and (2) are used
along the y-axis in equation (2). This kinematic in the gripper end position problem of the robot at any
analysis gives the position of the robot's endpoint in moment, while inverse kinematics is the most used
the x and y axes versus the two-limb robot's variables mathematical expression in robotic systems, as it
𝜃1 and 𝜃2 . creates the result of variable values against the entered
position value. Inverse kinematic values can be found
as a result of many operations
𝑥 = 𝑙1 𝑐𝑜𝑠𝜃1 + 𝑙2 𝑐𝑜𝑠(𝜃1 + 𝜃2 ) (1)

194
 S. Müftü, B. Gökçe / BEU Fen Bilimleri Dergisi 13 (1), 192-204, 2024

performed in the robotic system. In equations (3) and We can consider the equivalent circuit of the
(4), the inverse kinematics expression of the variables DC motor as in equation 5 [6].
𝜃1 and 𝜃2 ., which are the variables of the two-limbed
robot, are stated, respectively [5].
𝜃(𝑠) 𝐾𝑡
= (𝑅 (5)
𝑦 𝑙2 𝑠𝑖𝑛𝜃2 𝑉(𝑠) 𝑎 +𝑠+𝐿𝑎) (𝐽∗𝑆∗𝐵)𝑠+𝐾𝑡 ∗𝐾𝑏 ∗𝑠
𝜃1 = 𝑎𝑡𝑎𝑛 (𝑥 ) + 𝑡𝑎𝑛−1 (𝑙 +𝑙 ) (3)
1 2 𝑐𝑜𝑠𝜃2

𝑥 2 +𝑦 2 −𝑙1 2 −𝑙2 2
𝜃2 = + cos−1 ( 2𝑙1 𝑙2
) (4) The values in Table 1 are replaced by the
Rotor inertia J, viscous damping B, electrical constant
There are many systems used as drive 𝐾𝑏 , torque constant 𝐾𝑡 , Rotor inductance L rotor
elements in robot arms. One of them, DC motors, are resistance R in the transfer function found in Equation
electromechanical devices that convert electrical (16), and the transfer function for the Faulhaber DC
energy into mechanical energy. In control science, the motor is found. The transfer function for the DC
transfer function is used to model the behavior of a motor is given in Equation (6). If the unknowns in
DC motor. It connects the electrical voltage equation 5 are filled by means of table (1), equation
information entering the system with the output (6) is obtained.
mechanical position information. The Laplace
transform is used to model the nonlinear behavior 𝜃(𝑠) 5,8×10−3
= (5.1+70×10−6 (6)
around the equilibrium point of the DC motor. The 𝑉(𝑠) 𝑠)(0.66×10−7 𝑠+1)𝑠+5,8×10−3 𝑠
transfer function for the 1512406sr type DC motor of
the Faulhaber company used in this study was found If PID control is used correctly in robot arms,
with the values specified in the catalog section. In it enables robots to save energy with precision,
Table 1, the required features of the Faulhaber stability, and less movement. Since the study has two
1524006sr DC motor for TF are given. degrees of freedom, two PID control blocks were
used. The reference value for the PID control block is
Table 1. Faulhaber 1524006sr DC motor specifications. obtained from inverse kinematic analysis. Since each
PID control has three variables, there are six
Feature Name Value
parameter values in total. Angle obtained after the
Rotor Inertia 0.66 𝑔𝑐𝑚2
Viscous Damping 0 PID control block value is sent to the motors,
Electrical Constant 𝐴 respectively. The operation block of the angle values
0.172
𝑚𝑁𝑚
𝑚𝑁𝑚
sent to the motors against the entered reference x and
Torque Constant 5.8 y axes is given in Figure 4.
𝐴
Robor Inductance 70 𝜇𝐻
Terminal Resistance 1.1 Ω

Figure 4. Blok diagram of the robot arm.

195
 S. Müftü, B. Gökçe / BEU Fen Bilimleri Dergisi 13 (1), 192-204, 2024

For square movement, the robot arm started (8.19). Afterwards, it goes to (8.14), (3.14), and (3.19)
from the point where theta1 and theta2 limb angles points, respectively, and finishes the movement. The
were 0 degrees and followed the points determined on end point of the robot arm follows a square shape with
a coordinate axis. The determined points form a a corner length of 5cm. The reference dimension
square shape. The gripper end of the two-limbed robot placed on the coordinate axis of the square shape is
arm first starts from point (3.19) and goes to point indicated in Figure 5.

Figure 5. Square motion plan of a two-pronged robotic arm.

In order to achieve the desired movement in PID expression for control. The PID parameters
robot arms, it is necessary to utilize inverse obtained as a result of the optimization process were
kinematics and any controller. The trajectory created set as controller parameters in the real system. The
with the Simulink Support Package for Arduino block diagram of the realized system is shown in
Hardware was converted into an angle value with Figure 6.
inverse kinematics and sent to the

Figure 6. Block diagram of the two-limbed robot arm in the Simulink Support Package for Arduino Hardware.

196
 S. Müftü, B. Gökçe / BEU Fen Bilimleri Dergisi 13 (1), 192-204, 2024

In the system implemented with Arduino feedback to the PID control block. The 8-second
support from Simulink, 8 seconds are determined for temporal change reference for 𝜃1 and 𝜃2 separately is
square movement. During the sampling period, given in Figure 7.
angular position data from the encoders provided

Figure 7. Angular positions that a two-limbed robot arm must make for square motion tracking.

ABC, CGO meta-heuristic algorithms were used in

2.2. Controller Parameter Optimization of a
this study.
Two-Limbed Robot Arm

Optimization is the process of finding the best value 2.2.1. Particle Swarm Optimization Algorithm
of a given objective function with constraints. This
process is encountered in many fields, from PSO is a meta-heuristic optimization algorithm that
engineering systems to economics, from health to mimics particle swarm behavior to solve optimization
business. Optimization problems can be categorized problems. It was developed in 1995 by Dr. Kennedy
as continuous or discrete, dynamic or static, and Dr. Eberhart. PSO is applied in many different
constrained or unconstrained. These algorithms are a fields thanks to its easy implementation, fast solution,
sub-branch of artificial intelligence, and as the and high success rate. PSO is based on the principle
popularity of artificial intelligence increases, so does of particles searching for the optimal solution to a
its use for different problems [8]. given problem [10].
Meta-heuristic algorithms can often be The velocity and position of each particle are
classified according to their inspiration. Another expressed as a vector. There is also a fitness function
classification is the type of initial solution, i.e., that measures the fitness of each particle, which varies
multiple or single solution-based. Multiple solution- according to the problem. Each particle also keeps
based algorithms are usually called population-based, track of its past best position (pbest) and the best
while single solution-based algorithms are called position in the swarm (gbest). The particles move
trajectory-based [9]. Classifying metaheuristic towards the optimal solution by continuously
optimization algorithms according to the source of updating pbest and gbest. Finally, when the optimal
inspiration is usually classified as human, solution is obtained, the solution is achieved [11]. The
evolutionary, swarm logic, or science-based. PSO, steps that the particle swarm optimization algorithm
takes to solve a problem are shown in Figure 8 as a
flowchart.

197
 S. Müftü, B. Gökçe / BEU Fen Bilimleri Dergisi 13 (1), 192-204, 2024

solutions with other bees. This process is iteratively

repeated to optimize a given objective function [14].
More details about the ABC algorithm are given in
Figure 9 as a flowchart.

Figure 8. Flowchart of the particle swarm optimization

algorithm [12].

The ability of each particle to solve the

Figure 9. Flowchart of an artificial bee colony
problem and find the best solution means individual
optimization algorithm flowchart [14].
performance for the PSO algorithm. The PSO
algorithm tries to find the best solution by updating Finding the optimal solution by mimicking
the position and velocity of each particle. In this way, the behavior of the bees to search for and gather the
the individual performance of each particle plays an food source constitutes the individual performance of
important role in the process of solving the problem. the ABC algorithm. As the bees communicate with
The PSO algorithm encourages cooperation and each other, they work together to find the best
information sharing between particles to optimize solution. With the improvement in the performance of
individual performance and improve each particle's each bee, the problem-solving ability improves. As a
ability to solve the problem. In this way, the algorithm result, thanks to the organization and cooperation
is often used effectively in complex optimization exhibited by the bees, the ABC algorithm gives
problems. successful results in real-world optimization
problems.
2.2.2. Artificial Bee Colony Optimization
Algorithm 2.2.3. Chaos Game Optimization

ABC is a metaheuristic optimization algorithm The chaos game optimization algorithm is a

developed by Karaboga in 2005. It is inspired by the population-based optimization algorithm developed
behavior of bees as they forage for food sources and by Talatahari and Azizi. This algorithm incorporates
bring it back to their nests. The behavior of bees is certain chaos theory principles where fractals are
modeled mathematically. The parts of ABC can be generated using chaos game methodology. This
listed as relocation of bees, discovery of food sources,
game, which aims to generate fractal patterns based
evaluation of food sources, and information sharing
on randomly generated starting points, is designed
among bees [13].
The ABC algorithm is simulated in a search based on the hypothesis of generating the Sierpinski
space where bees form a colony to represent potential triangle geometric structure [15].
solutions. The bees evaluate randomly selected CGO is known for positive aspects such as
solution candidates and use this information to high computation time efficiency and easy
generate new solution candidates by sharing the best implementation to effectively solve constrained

198
 S. Müftü, B. Gökçe / BEU Fen Bilimleri Dergisi 13 (1), 192-204, 2024

optimization problems. The biggest difference from similarities of the solution candidates and keeps the
other optimization algorithms is that it is parameter- best solution.
free. In other words, it does not need any additional
parameters other than parameters such as population
size, and maximum iteration. Thanks to this feature, 3. Results and Discussion
it overcomes challenging problems. The algorithm There are six controller parameters in total for the
starts optimization by initially generating random realized two-legged robot arm. Particle swarm,
search candidates, and this initialization process is artificial bee colony, and chaos game optimization
performed depending on the population size, the algorithms were used to find the most suitable values
number of decision variables, and the bounds of the of these parameters. The simulation results obtained
solution space [15]. More details about the CGO from each metaheuristic optimization algorithm are
compared. As a result of the simulation process, the
algorithm are given in Figure 10 as a flowchart.
controller parameters found for trajectory tracking in
the real robot arm are specified.
In all three simulations, the maximum
number of function evaluations was carried out at 40
and the number of populations at 50 in order to
perform the most appropriate tracking of the square
trajectory for the robot arm. The social and cognitive
constants of the PSO algorithm were determined to be
2, and the inertia weight was determined to be [0.9,
0.4] [7]. The limit value for the ABC is determined to
be 100. These coefficients were used at the same
values for each step throughout the study. If we put
the PID parameters in order
𝑘𝑝1 , 𝑘𝑖1 , 𝑘𝑑1 , 𝑘𝑝2 , 𝑘𝑖2 , 𝑘𝑑2 limit [0 0 0 0 0 0], and
upper limit [20,000 15,000 18,000 20,000 15,000
18,000]. The simulation results obtained as a result of
the optimization process for frame motion tracking
are shown in Figures 11 for the chaos game, 12 for the
particle swarm, and 13 for the artificial bee colony.
It is clearly seen in Figure 8, Figure 9, and
Figure 10 that CGO and ABC algorithms give better
results than PSO algorithms. The ABC and CGO
algorithms gave similar results. Figure 14 shows that
the CGO algorithm gives better results than the ABC
algorithm. In Figure 14, (A) is a close view of CGO,
and (B) is a close view of the ABC algorithm at the
point (8,14) forming the square. It is noted that CGO
is closer to the reference value of red dots than ABC.
Figure 10. Flowchart of Chaos game optimization
In the simulation environment, all three algorithms
algorithm flowchart [16]. gave successful results, but CGO gave better results
with differences. For this problem, each algorithm is
In the CGO algorithm, the success of each successful but the best ranking is CGO, ABC, and
solution constitutes the individual performance of the PSO. The reason for this is the parameters in the PSO
algorithm. The CGO algorithm tries to find the best and ABC algorithms, and it is thought that the CGO
solution by randomly moving each solution candidate algorithm is more successful in generating new
on a fractal structure. While solving the optimization values.
problem, the CGO algorithm compares the

199
S. Müftü, B. Gökçe / BEU Fen Bilimleri Dergisi 13 (1), 192-204, 2024

Figure 11. CGO reference tracking for square motion.

Figure 12. PSO reference tracking for square motion

Figure 13. ABC reference tracking for square motion

200
 S. Müftü, B. Gökçe / BEU Fen Bilimleri Dergisi 13 (1), 192-204, 2024

Figure 14. Comparison of ABC and CGO reference

Mean Square Error (MSE) and Root Mean and 𝜃2 that make up the two-limbed robot arm. The
Squared Error (RMSE) were used to compare the MSE and RMSE values obtained by CGO, PSO, and
error parts of the optimization process of the angles 𝜃1 ABC are given in Table 2.

Table 2. MSE and RMSE error metrics of CGO, PSO, and ABC algorithms for reference tracking.
𝜃1 angle 𝜃2 angle
MSE RMSE MSE RMSE
CGO 32.33 5.68 154.38 12.42
PSO 58.98 7.67 238.79 15.45
YAKA 32.74 5.72 155.36 12.46

As a result of the optimization process, six were obtained for each optimization algorithm. The
optimization parameters 𝑘𝑝1 , 𝑘𝑖1 ,𝑘𝑑1 ,𝑘𝑝2 ,𝑘𝑖2 ,𝑘𝑑2 PID parameters found are given in Table 3.

Table 3. The PID parameters found for square motion.

Controller Parameter CGO PSO ABC

𝑘𝑝1 19853 15.016 17667

PID 𝜃1 𝑘𝑖1 799 17675 0

𝑘𝑑1 1082 3446 0

𝑘𝑝2 19775 19374 20000

PID 𝜃2 𝑘𝑖2 0 9252 1835

𝑘𝑑2 1118 3776 1157

Each metaheuristic optimization algorithm, has been tested separately on the real system. Table 2
CGO, PSO, and ABC, which determines the PID shows the differences in the controller parameters
parameters obtained as a result of the simulation obtained by the CGO, PSO, and ABC algorithms.
performed on the robot arm controlled in real time These differences come from the structure of the
thanks to the Simulink support package for Arduino, optimization algorithms.

201
 S. Müftü, B. Gökçe / BEU Fen Bilimleri Dergisi 13 (1), 192-204, 2024

The angle value that each motor should make given in Figure 15 (a) for 𝜃1 and Figure 15 (b) for
for 8 seconds in order for the two-limbed robot arm to 𝜃2 for the CGO algorithm, Figure 16 (a) for 𝜃1 and
perform the reference motion is given in Figure 7. The Figure 16 (b) for 𝜃2 for the PSO algorithm, Figure 17
results obtained by realizing the PID controller (a) for 𝜃1 , and Figure 17 (b) for 𝜃2 for the ABC
parameters specified in Table 2 on the real system are algorithm.

Figure 15. Reference tracking of PID parameters obtained by CGO for square orbit at angle.
(a) θ1 (b) θ2 .

Figure 16. Reference tracking of PID parameters obtained by PSO for square orbit at angle (a) θ1 (b) θ2 .

Figure 17. Reference tracking of PID parameters obtained by ABC for square orbit at angle (a) θ1 (b) θ2 .
202
 S. Müftü, B. Gökçe / BEU Fen Bilimleri Dergisi 13 (1), 192-204, 2024

algorithm gave more successful results than the PSO

When the experimental data obtained from
and ABC algorithms in the PID controller parameters
real-time measurements of the DC motors, which are
for the two-limbed robot arm. The success of
the actuators of the two-limbed robot arm, are
metaheuristic optimization algorithms is not the same
analyzed in Figures 15, 16, 17; PSO and ABC
for every problem. The success of an optimization
algorithms gave very close results to the reference as
algorithm in one problem does not mean that it will
mentioned in the literature. For the square motion,
be successful in another. This study has shown in the
which is one of the most difficult motions for a two-
literature that the CGO metaheuristic optimization
limbed robot arm, all three algorithms gave very
algorithm can be used to determine the coefficients of
successful results in real-time control. However, the
a PID controller.
recently popular and promising CGO algorithm gave
slightly better results than the PSO and ABC The controller part is of great importance in
algorithms. This is the same as the simulation result. the studies to be carried out on robotic systems.
This is due to the fact that the CGO algorithm finds Among the methods to be used for the problem of
the parameters of the PID controller closer to the determining controller parameters, methods with a
correct values. small number of parameters should be chosen. Since
the angular loss of any axis affects the holder end, the
4. Conclusion and Suggestions material to be used with the smallest error rate should
be selected. Motor selection should be made
Many studies in the literature on robot control have according to the torque value required for the system.
been done with servo motors or stepper motors. The If the motor to be used as a drive element is a DC
DC motor used in this study gave a very successful motor, the step range of the encoder should be high,
result and demonstrated that a DC motor with certain the value of the motor speed should be low, and the
parameters can be used for robot control. The transfer cycle ratio of the reducer should be high.
function of the DC motor was obtained from the
catalog values, and the controller feedback was Acknowledgment
provided by the encoder. There are many algorithms
to determine the parameters of the PID controller used This study was conducted at Necmettin Erbakan
to control the two-limbed robot arm. In this study, the University, Graduate School of Natural and Applied
CGO algorithm is used for the first time to determine Sciences, Mechatronics Master's thesis written within
the PID parameters. The CGO, PSO, and ABC the scope of the master's program at the Department
algorithms are compared as a result of the simulation of Engineering prepared.
of the transfer function of the DC motor. The Contributions of the authors
parameters of the PID controller obtained as a result
of the simulation were tested on the two-limbed robot All authors contributed equally to the study.
arm.
As a result of the simulation process, the CGO Conflict of Interest Statement
algorithm gave a result closer to the reference than the
PSO and ABC algorithms. In order for the two-limbed There is no conflict of interest between the authors.
DC motors to perform square motion, the change in
the angle value of the DC motors for 8 s was Statement of Research and Publication Ethics
calculated, and the control was performed for each
motor. When the motors run for 8 s, the end part of The study is complied with research and publication
the robot draws a square motion. Thanks to the ethics.
metaheuristic optimization algorithms used, the CGO

203
 S. Müftü, B. Gökçe / BEU Fen Bilimleri Dergisi 13 (1), 192-204, 2024

References

[1] W. S. Barbosa, M. M. Gioia, V. G. Natividade, R. F. Wanderley, M. R. Chaves, F. C. Gouvea and F. M.

Gonçalves, “Industry 4.0: examples of the use of the robotic arm for digital manufacturing processes,”
International Journal on Interactive Design and Manufacturing (IJIDeM), vol. 14, pp.1569-1575, 2020.
[2] S. K. Valluru and M. Singh, “Optimization strategy of bio-inspired metaheuristic algorithms tuned PID
controller for PMBDC actuated robotic manipulator,” Procedia Computer Science, vol.171, pp. 2040-
2049, 2020.
[3] K. J. Åström and T. Hägglund. “The future of PID control,” Control engineering practice, vol. 9, no.
11, pp.1163-1175, 2001.
[4] S. B. Joseph, E. G. Dada, A. Abidemi, D. O. Oyewola, and B. M. Khammas, “Metaheuristic algorithms
for PID controller parameters tuning: Review, approaches and open problems,” Heliyon, vol. 8, no. 5,
p.e09399, 2022.
[5] S. Kucuk and Z. Bıngul, Robot kinematics: Forward and inverse kinematics, London, UK: INTECH
Open Access Publisher, 2006, pp. 117-148.
[6] S. Maheriya and P. Parikh, “A review: Modelling of Brushed DC motor and Various type of control
methods,” Journal for Research, vol. 1, 2016.
[7] X. Hu, “Particle swarm optimization tutorial, titan.csit.rmit.edu, 2006. [Online]”. Available:
https://titan.csit.rmit.edu.au/ e46507/publications/pso-tutorial-seal06.pdf. [Accessed: March 22, 2021].
[8] J. Tang, G. Liu and Q. Pan, “A review on representative swarm intelligence algorithms for solving
optimization problems,” Applications and trends. IEEE/CAA Journal of Automatica Sinica, vol. 8, no.
10, pp.1627-1643, 2021.
[9] E. Boğar, “Optimizasyon kuramında yeni bir metasezgisel yaklaşım: Ergen kimlik arama algoritması
(AISA) ve mühendislik uygulamaları,” Doktora tezi, Pamukkale Üniversitesi Fen bilimleri Enstitüsü,
Denizli, 2021.
[10] R. Eberhart and J. Kennedy, “Particle swarm optimization,” in Proceedings of the IEEE international
conference on neural networks vol. 4, pp. 1942-1948, November 1995.
[11] D. Bratton and J. Kennedy, “Defining a standard for particle swarm optimization,” In 2007 IEEE swarm
intelligence symposium, pp. 120-127, April 2007.
[12] M. Azab, “Global maximum power point tracking for partially shaded PV arrays using particle swarm
optimization,” International Journal of Renewable Energy Technology, vol. 1, no.2, pp. 211-235, 2009.
[13] D. Karaboga and B. Basturk, “A powerful and efficient algorithm for numerical function optimization:
artificial bee colony (ABC) algorithm,” Journal of global optimization, vol. 39, pp. 459-471, 2007.
[14] H. O. Erkol, “Ters sarkaç sisteminin yapay arı kolonisi algoritması ile optimizasyonu,” Politeknik
Dergisi, vol. 20, no. 4, pp.863-868, 2017.
[15] S. Hassan, B. Abdelmajid, Z. Mourad, S. Aicha and B. Abdenaceur, “An advanced MPPT based on
artificial bee colony algorithm for MPPT photovoltaic system under partial shading condition.
International,” Journal of Power Electronics and Drive Systems, vol. 8, no. 2, p. 647, 2017.
[16] S. Talatahari and M . Azizi, “Chaos game optimization: a novel metaheuristic algorithm,” Artificial
Intelligence Review, vol. 54, no. 2 pp. 917-1004, 2021.
[17] M. Barakat, “Novel chaos game optimization tuned-fractional-order PID fractional-order PI controller
for load–frequency control of interconnected power systems,” Protection and Control of Modern Power
Systems, vol. 7, no. 1, 2022.

204