Cogent Engineering

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/oaen20

Modeling and control of a 3-DOF articulated

robotic manipulator using self-tuning fuzzy sliding
mode controller

Aderajew Ashagrie, Ayodeji Olalekan Salau & Tilahun Weldcherkos |

To cite this article: Aderajew Ashagrie, Ayodeji Olalekan Salau & Tilahun Weldcherkos | (2021)
Modeling and control of a 3-DOF articulated robotic manipulator using self-tuning fuzzy sliding
mode controller, Cogent Engineering, 8:1, 1950105, DOI: 10.1080/23311916.2021.1950105

To link to this article: https://doi.org/10.1080/23311916.2021.1950105

© 2021 The Author(s). This open access

article is distributed under a Creative
Commons Attribution (CC-BY) 4.0 license.

Published online: 15 Jul 2021.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=oaen20
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

SYSTEMS & CONTROL | RESEARCH ARTICLE

Modeling and control of a 3-DOF articulated
robotic manipulator using self-tuning fuzzy
sliding mode controller
Received: 01 April 2021 Aderajew Ashagrie1*, Ayodeji Olalekan Salau2 and Tilahun Weldcherkos1
Accepted: 15 June 2021
Abstract: Robotic manipulators are highly coupled multi-input multi-output (MIMO)
*Corresponding author: Aderajew
Ashagrie, Department of Electrical nonlinear systems with uncertainties and highly time-varying dynamic capabilities.
and Computer Engineering, Addis These characteristics make the trajectory control of a robotic manipulator very
Ababa Science and Technology
University (AASTU), Addis Ababa, challenging. This paper presents the modeling and trajectory tracking control of
Ethiopia
E-mail: aderajewashagrie@gmail. a 3-DOF robotic manipulator using self-tuning fuzzy sliding mode controller (ST-
com FSMC). The stability of the system was investigated using the Lyapunov direct
Reviewing editor: method. The controller was implemented using MATLAB/Simulink and its perfor­
Eldaw Eldukhri, Electrical
Engineering, King Saud University, mance was evaluated. The simulation results show that the proposed controller
Riyadh, Saudi Arabia removed chattering phenomena from the control effort and reduced the tracking
Additional information is available at error (average steady-state error) to 0.0036 rad. However, in the case of the con­
the end of the article
ventional controllers, the average steady-state error increased to 0.0413 rad,
0.0044 rad, and 0.0053 rad for the Proportional-integral-derivative (PID) controller,
Sliding mode controller (SMC), and Fuzzy sliding mode controller (FSMC), respec­
tively. The conventional controllers (PID, SMC, and FSMC) were designed for the

ABOUT THE AUTHORS PUBLIC INTEREST STATEMENT

Aderajew Ashagrie received a BSc. in Electrical Robots are in widespread use in recent times.
and Computer Engineering from Arba Minch This is because most industries make use of
University and M.Sc. in Electrical and Computer robots to increase productivity and product
Engineering (Industrial Control and quality. Robots are also used to get jobs done
Instrumentation Engineering) from Addis Ababa faster with greater precision and reliability than
Science and Technology University, Ethiopia. He is humans. Robot manipulators are created from
currently working as a lecturer at Arba Minch a sequence of link and joints. They are highly
University. complicated nonlinear systems with uncertain­
Dr. Ayodeji Olalekan Salau received a B.Eng. in ties. These characteristics make the trajectory
Electrical/Computer Engineering from the Federal control of a robotic manipulator difficult. To solve
University of Technology, Minna, Nigeria. He the problem of handling uncertainties with the
received an M.Sc. and Ph.D. degree in Electronic aim of increasing the robustness of the system to
and Electrical Engineering from the Obafemi give an improved output, a self-tuning fuzzy
Ayodeji Olalekan Salau Awolowo University, Ile-Ife, Nigeria. In 2020, Dr. sliding mode controller (ST-FSMC) was proposed.
Salau’s paper was awarded the best paper of This control method was implemented to give
the year 2019 in Cogent Engineering. He is cur­ a high-performance controller in the presence of
rently working at Afe Babalola University in the uncertainties. In addition, various control meth­
Department of Electrical/Electronics and ods were successfully designed and comparative
Computer Engineering. analysis was carried out. Simulations were per­
Tilahun Weldcherkos received a BSc. in formed to demonstrate that the proposed ST-
Electrical and Computer Engineering from Arba FSMC performances better than other conven­
Minch University, and MSc. in Industrial Control tional controllers (PID, SMC, and FSMC) because
and Instrumentation Engineering from Addis the proposed controller can adjust itself to the
Ababa Science and Technology University, systems uncertainty.
Ethiopia. He is currently working as a lecturer in
Arba Minch University.

© 2021 The Author(s). This open access article is distributed under a Creative Commons
Attribution (CC-BY) 4.0 license.

Page 1 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

purpose of comparison with the proposed ST-FSMC. The simulation results show that
the designed controller (ST-FSMC) has a superior tracking performance, is robust
and is not sensitive to applied model parameter variations as compared to other
conventional controllers.

Subjects: Algorithms & Complexity; Industrial Engineering & Manufacturing; Systems &
Control Engineering; Fuzzy Systems; Electrical &Electronic Engineering

Keywords: Fuzzy sliding mode controller; sliding mode controller; ST-FSMC; PID; DOF

1. Introduction
Robots are in widespread use in recent times to get jobs done faster with greater precision and
reliability than humans. Nowadays, most industries make use of robots to increase productivity,
precision, and product quality. A robot makes use of a robotic manipulator to perform tasks which
are most times difficult to perform. These tasks include: space exploration, robotic surgery or
welding. The robotic manipulator is an electrically actuated Three Degrees of Freedom (3-DOF)
arm-like serial manipulator comprising several segments which may be sliding or jointed. They are
used in industrial applications with articulated geometric types because articulated robotic manip­
ulators are most commonly used in factories worldwide (Gambhire et al., 2021).

Articulated manipulators are the most common industrial robotic structures providing more
than 50% of annual installations around the world (Wang et al., 2014). This type of robot is utilized
in numerous industrial processes such as welding, painting, pick and place, dispensing, packaging
as well as several handling applications which include machine tool tending, metal casting, and
general material handling. It has a wide workspace and operates in a 3-D space.

Robotic manipulators are also employed for jobs that are too dirty, hazardous, and highly
repetitive or boring to be suitable for humans (Ghaleb & Aly, 2018). Due to these reasons, robotic
manipulator research is one of the most interesting areas in industrial technology applications.
However, most of these applications are confined to slow motion operations without interactions
with the environment. This is primarily due to the limited performance of the available controllers
in the market. To help improve the operational speed for better accuracy, advanced control
strategies are required. Because a controller is the main part of this sophisticated system, the
main requirements to control robotic manipulators are stability and robustness. The main chal­
lenges in its motion and control are the complexity of its dynamics and the presence of uncer­
tainties in the system. Consequently, modeling and analysis of a robotic manipulator and in
addition implementing control methods are significant to obtain precision, good accuracy, and
increased productivity.

This paper presents the modeling and design of a controller for a 3-DOF robotic manipulator
which requires minimum effort with a minimum error. Since, trajectory-tracking control of robotic
manipulators is a challenging task due to the highly coupled MIMO and highly time-varying
dynamic nature of the system. Furthermore, there are always numerous uncertainties in the
system model such as parameter uncertainty and external disturbance which cause unstable
performance of the control system. In addition to this, most research works have been observed
to neglect actuator dynamics. This resulted in reduced trajectory tracking control performance. In
any control system, stability and robustness are the two important characteristics that a robot
manipulator requires. Consequently, it is very important to model and evaluate robot manipulators
by utilizing proper control methods. In industrial applications, typical PID controllers are used to
control such complex systems. PID controllers are easy to implement and require the tuning of
only three parameters, and have relatively acceptable tracking performance for a robot. But the
linear PID controller is not suitable to accomplish the desired tracking performance of nonlinear
dynamic systems.

Page 2 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Different researches have been done to reduce the shortcomings of PID control using
a nonlinear controller, and in addition, different approaches have been proposed for designing
nonlinear controllers such as FLC, SMC, and FSMC. All these three controllers have been applied
successfully in many applications, but they also have some limitations. Therefore, this paper also
addresses the problem of performance degradation or loss of system stability due to neglected
actuator dynamics by taking into account actuator dynamics in the system to improve the
performance response.

2. Related works
Various modeling and control strategies have been presented in different researches for trajectory
tracking control of robotic manipulators. In this section, a review of the researches which focus on
trajectory-tracking control of robotic manipulators and other related controllers used in the control
of robot manipulator systems is presented. In general, the studies in the literature reviewed in this
paper focused on modeling and design of the system for either a linear or a nonlinear controller.

In Ohri et al. (2011), the authors present a comparison of the conventional PID controller and sliding
mode control (SMC) for robotic manipulators. Both techniques have good performance, but the SMC is
insensitive to parametric variations and has good disturbance rejection. The simulation results of the
study show that the performance of the SMC is better than PID under both sine and cosine trajectory.
As the payload is changed or during uncertainty, the tracking error in the PID controller is increased,
but in the case of SMC, the performance remains the same, which clearly shows that the SMC is more
robust than the PID controller. In this paper, the authors assume that the actuators do not have
dynamics of their own, and arbitrary forces can be commanded at the joint of the robotic manipulator
for simplicity. The results obtained were promising, but the authors didn’t consider the actuator
dynamics and ignored the effects of joint friction which brought about low controller performance.

In Getachew (2018), the authors proposed SMC for a two degrees of freedom (2-DOF) robotic
manipulator system and presented a comparative study of PID controllers. Based on simulation results
and analysis, the authors concluded that the SMC controller has superior performance and is more
robust when contrasted with the PID controller for sinusoidal trajectory tracking. The results obtained
show a good performance response when the system is under a disturbance of 25% of the input
amplitude. The problem with this was, as the system disturbance increased to 40% of the input
amplitude, the sliding surface and the derivative of the sliding surface of the second joint introduced
a chattering which caused a chattering in the control effort of the second joint. Finally, the authors
suggested that in future works, other chattering elimination methods can be designed which may
overcome any amount of disturbance introduced in the robot.

In Alassar et al. (2010), the authors introduced a fuzzy logic controller (FLC) for manipulating a 5-DOF
robotic manipulator based on the independent joint control approach. The proposed controller was
developed to overcome the shortcomings of the linear PID controller. Simulation was performed using
MATLAB/Simulink to compare the performance of the controller in terms of time response. The results
obtained were promising as the FLC was used to control complex nonlinear dynamic systems. However,
the design of the FLC highly depended upon expert knowledge or trial and error. Furthermore, it didn’t
guarantee stability and robustness due to the linguistic expressions of fuzzy control.

A design of computed torque control (CTC) methodology for PUMA-560 robot manipulator was
presented in (Piltan et al., 2012a). Computed torque control (CTC) is a powerful nonlinear controller
that is extensively utilized in robot manipulator control which calculates the required torques for all
manipulator joints using a nonlinear feedback control law based on feedback linearization. The
simulation results indicate that the CTC controller works very well when all dynamic and physical
parameters are known. But still, there was a problem in uncertain dynamic system models. In most
physical systems like robot manipulators, the parameters are unknown or time-variant. Consequently,
the authors recommended designing a sliding mode controller (SMC) to solve this problem.

Page 3 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

In (Piltan et al., 2012b), a novel fuzzy sliding mode controller (FSMC) was introduced to control
the bus suspension system by combining fuzzy logic with an SMC. This method eliminates the two
major drawbacks of SMC; firstly by using the boundary layer instead of the signum and FLC in place
of the equivalent dynamic part of SMC to eliminate the chattering and equivalent dynamic
dependency problem, respectively. The authors concluded from the outcome of the simulation
that FSMC has better performance for the bus suspension system and is more robust as compared
to the PID controller.

In Rahmdel et al. (2012), the authors introduced an intelligent control approach called the FSMC
with global stabilization and saturation function (boundary layer method) for the trajectory track­
ing control of a robot manipulator. The FSMC used in their work has to chatter in it. This chattering
effect of the FSMC was eliminated by applying a saturation function instead of a signum function.
The authors used several conventional methods such as PD, CTC, and SMC to evaluate and contrast
the proposed control method. Finally, the results of the simulation indicate that the FSMC can
guarantee robustness against continuous disturbance and has a strong ability to eliminate
chattering.

A nonlinear SMC was utilized in a spread-out scope of regions such as in robotics, process
control, and aerospace applications in Shaei (2010). The main reason this controller was used is
because of its performance. Also it solves some main challenging issues in the systems control
such as resistivity to external disturbance and uncertainty. However, this controller has two
significant limitations: the chattering and sensitivity problem; therefore the controller is very
sensitive to noise. In Chang et al. (2008), an SMC with a boundary layer technique to minimize
or eliminate chattering was presented. In the boundary layer technique, the basic idea is to replace
the saturation function with a discontinuous function in a small neighborhood of the switching
surface. But, this replacement caused an increase in the error performance, and also increased the
rising time. Similarly, in Kapoor & Ohri (2013), the authors presented an SMC with saturation and
hyperbolic tangent functions instead of signum function. Unfortunately, this approach was effec­
tive only in specific cases. When hard uncertainties are not present, some problems are attenuated
at the cost of a loss of robustness. Furthermore, a nonlinear FSMC was used to reduce or
compensate the negative effect of saturation function in the system. However, both SMC and
FSMC have difficulty in handling unstructured model uncertainties, because the sliding surface gain
is chosen by trial and error, which means that the SMC and FSMC must have prior knowledge of the
systems uncertainty. In contrast to other papers, this work combines the fuzzy-based tuning
method with FSMC to overcome the problem of handling unstructured model uncertainties to
increase the robustness of the system. The fuzzy-based tuning method was used to adjust the
sliding surface slope (λ) as well as a switching gain (K) to obtain the best performance of the
system. The coefficients which have a significant impact on the discontinuous part of the control
system were investigated. If these coefficients are tuned properly, the SMC can reject disturbances
more desirably. Accordingly, the proposed controller (Self-tuning-FSMC) has powerful resistance
and can solve the problems of both structured and unstructured uncertainties (Kapoor & Ohri,
2017; Medhaar et al., 2006). In (Anh et al., 2018), an adaptive fuzzy sliding mode controller
(AFSMC) to demonstrate the stability of a robotic manipulator system using Lyapunov stability
theorem was presented. Similarly, an AFSMC was proposed in (Zhang et al., 2020) to improve the
trajectory tracking accuracy of a 3-DOF serial manipulator with parameters uncertainties.

Past researches on the design of various control techniques indicate that actuator dynamics
plays an important role in the overall robot dynamics and poor attention to this aspect may cause
imperfection in the final control system, particularly in the case of high-velocity moment, and
highly varying loads as reported in Tao & Lee (2016). Moreover, the saturation of the actuator was
not given much attention in the dynamic modeling of the robot manipulator. The authors used an
electrical actuator to drive all joints of the robot manipulator, and the injected force/torque of the
joints is the load torque of the actuator, which is considered in the dynamic model of the overall
system. Additionally, in the case of torque-based control design, the torque directly controls the

Page 4 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

actuators which cause them to face implementation issues in real-time applications (Karamali
Ravandi et al., 2018). On the other hand, the voltages of the electrical actuators are utilized to
control commands in practical applications. In (Dhyani et al., 2020), the authors presented
a redundant 7-DOF manipulator. It possesses more resources than those strictly required to
execute a given task. This provides the robot manipulator with an increased capacity to work in
real-world applications. However, the inverse kinematics and control of redundant manipulators
are becoming increasingly complicated with each added degree of freedom. In addition, the
system dynamics are complex and may involve un-modelled uncertainties; thereby causing the
corresponding path planning problem to become increasingly challenging.

To address these challenges, this paper proposes a self-tuning fuzzy slide mode controller (ST-
FSMC) with powerful resistance to solve both structured and unstructured uncertainty problems. In
addition, the paper also addresses the problem of performance degradation or loss of system
stability due to neglected actuator dynamics by taking into account actuator dynamics in the
system to improve the performance response. To meet the predefined aim of this paper, authors
primarily concentrated on the study of the literature about the theoretical and structural back­
ground of the robotic manipulator and its controller. Thereafter with the knowledge gained, we
have compared the performance of various controllers with the aim of overcoming the short­
comings of the traditional linear and nonlinear controllers with the proposed controller (ST-FSMC).

Since the appropriate dynamic model equations are very important in designing the robotic
controller, to develop the model we have divided the developmental stages into two main parts.
First, we developed the robot manipulator dynamic equations. This starts from the equation of
position and orientation description, forward and inverse kinematics, dynamic analysis and forces,
kinematic and potential energy derivations using the Lagrange equation. The second is to model
the actuator dynamics of the robot with some assumptions.

Finally, the overall nonlinear system model or manipulator and actuator dynamics are devel­
oped. Since in most of the past studies, the control input is a torque or force but in our case,
instead of using torque or force, we used the voltage of the actuator from the developed
mathematical model. This was done to improve the drawbacks noticed in previous works.

After the dynamic models have been developed, the proposed controller (Self-Tuning FSMC) for
a 3-DOF robot manipulator was designed with MATLAB. Finally, the performance of the proposed
method is tested and compared with the conventional controller by simulation using MATLAB
software, and then analysis and interpretation of the results of simulation were presented.

3. Dynamic and kinematic formulation of 3-DOF articulated robotic manipulator

The first important step in controlling any physical system is to acquire a complete and accurate
mathematical model of the system. This model is useful for simulation of the system before
implementing the system into a real-time application. In this section, the mathematical equations
for the kinematic and dynamic model of the three-link industrial robotic manipulator having three
revolute joints are discussed. The kinematic analysis has been classified into two types, forward
and inverse kinematics, and the dynamic behavior of the manipulator concerned with the relation­
ship between joint actuator torques and motion of links for the design of the control algorithms.

Finally, the overall nonlinear system model or manipulator dynamic and actuator dynamics are
developed. Since most of the literatures have presented the use of the control input, torque or
force, but in our case, instead of these, we used the voltage of the actuator from the developed
mathematical model. This is to improve on some of the drawbacks in the previous works.

3.1. Kinematic modeling of robot manipulator

Kinematics describe and analyze the motion of bodies without regard to forces or moments that
cause the motion. In kinematics, the position, orientation, velocity, and acceleration of the robot

Page 5 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

manipulator are analyzed from the point of spatial geometry. To describe the geometry, a link
frame based on Denavit-Hartenberg (DH) description is assigned to each link of the manipulator.
The kinematics are divided into forward (FK) and Inverse (IK) kinematics.

The forward kinematics is done using the Denavit–Hartenberg (DH) parameters. The DH para­
meters use only four factors which describe the movement of the robotic manipulator. These
parameters are:qis the rotation angle: this is along z-axis always,α is twist: this is the angle it takes
to map z-axis of the previous joint to the current joint,ais the distance of previous joint center to
next center in x-direction and d is the offset distance of previous joint center to next in z-direction.

The frame assignments for the 3-DOF articulated robotic arm showing the various joints and
their links are shown in Figure Figure 1 (Massaoudi et al., 2019). The DH parameters for the 3-DOF
robotic manipulator are presented in Table 1.

Subsequent to getting the DH-convention table, a series of homogeneous matrices are obtained
based on the number of the DOF. Therefore, the overall homogeneous transformation matrix for
the position and orientation of the end effector that is required is given as:

Figure 1. Frame assignments

for the 3-DOF articulated
robotic arm.

Table 1. DH parameter for 3-DOF articulated robotic arm

Link ai 1 αi 1 di qi
1 0 900 L1 q1
2 L2 0 0 q2
3 L3 0 0 q3

Page 6 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

2 3
C1 C23 C1 S23 S1 L3 C1 C23 þ L2 C1 C2
6 S1 C23 S1 S23 C1 L3 S1 C23 þ L2 S1 C2 7
3 6
T0 ¼ 4 7 (4)
S23 C23 0 L3 S23 þ L2 S2 þ L1 5
0 0 0 1

The homogenous transformation matrix defined in Eq. (4) is the one that defines the forward
kinematics of the 3-DOF robotic manipulator shown in Figure Figure 1. From this matrix, the
position and orientation of end effectors is a non-linear function of joint variables. Having derived
the forward kinematics of Figure Figure 1, then it’s possible to obtain the position and orientation
of the end-effector from the individual joint angle (q1 ,q2 ; andq3 ). The end-effector orientation
matrix is represented by the first three rows and columns of the transformation matrix in Eq. (5).

8 9 2 3
< px ¼ L3 C1 C23 þ L2 C1 C2 = C1 C23 C1 S23 S1
py ¼ L3 S1 C23 þ L2 S1 C2 ; Rd ¼ 4 S1 C23 S1 S23 C1 5 (5)
: ;
pz ¼ L3 S23 þ L2 S2 þ L1 S23 C23 0

3.2. Inverse kinematics

The inverse kinematics of the robotic manipulator are a way of finding the robot manipulator joint
variables given the position of the Cartesian coordinates of the end-effector. In this section, we
solved the inverse kinematics equations for the 3-DOF robotic manipulator using the geometrical
approach as given by Eqs. (6)-(8)

q1 ¼ a arctanðpy ; px Þ, where, q1 is defined only when

p2x þ p2y >0 (6)

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
p2x þp2y þðpz 1Þ2 l22 l23
q3 ¼ atan2ðS3 ; C3 Þ, where,C3 ¼ 2l2 l3 ,S3 ¼ � 1 C3 2 (7)

S3 2 ½ 0:173; 0:173� (Else, point P (end effector point) is out of workspace). This is due to joint
three constraint.

q2 ¼ atan2ðS2 ; C2 Þ, where, q2 is defined if and only if p2x þ p2y þ ðpz 1Þ2 >0
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðl2 þl3 C3 Þðpz l1 Þ l3 S3 ðC1 px þpy S1 Þ
S2 ¼ , C2 ¼ � 1 S2 2 2 ½ 1; 1� (8)
l3 2 S3 2 ðl2 þl3 C3 Þ2

3.3. Dynamic model of 3-DOF robotic manipulator

In this section, we derive the mathematical equations that explicitly describe the relationship
between the joint torques applied by the actuators and motion of the robot system. The best
technique that can be used for obtaining the dynamic model of the manipulator is the Euler-
Lagrangian approach which gives an energy difference between the kinetic energy and potential
energy. This approach is mostly used in deriving the complicated mathematical model of a robot
manipulator (Abbas, 2018). Therefore, we have used the Euler -Lagrange method in this paper.

Therefore, kinetic and potential energy analysis of each link of the manipulator must be carried
as shown in Figure Figure 2. The kinetic energy can be both rotational and translational; it is
_ The potential energy is due to conservative forces
a function of both position and velocity, KE ðq; qÞ.
exerted by gravity. The Euler-Lagrange equations for the n-DOF system are defined as:

_
d @Lðq; qÞ _
@Lðq; qÞ
ð Þ ¼ τi (9)
dt @qi @qi

Page 7 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 2. Force calculation of

joints.

where, i ¼ 1; 2 . . . :n; τi is the torque/ force exerted on each joint of the manipulator, qi is the link
position, and q_ i is the joint velocity.

The values of the physical parameters of the manipulator are presented in Table 2 (Korayem
et al., 2010; Pezeshki et al., 2012).

As there are 3-DOF, the Euler-Lagrangian equations give these three equations (Eqs. (10)–(12)).

1 1 1 1
τ1 ¼ ð m1 r12 þ m2 l22 C22 þ m3 l23 C223 þ m3 l22 C22 þ m2 l2 l23 C2 C23 Þq
€1 þ ð m2 l22 sinð2q2 Þ m3 l22 sinð2q2 Þ
2 4 4 4
m3 l2 l3 sinð2q2 þ q3 Þ m3 l2c2 sinð2ðq2 þ q3 ÞÞÞq_ 1 q_ 2 þ ð m3 l2 l3 C2 S23 m3 l2c3 sinð2ðq2 þ q3 ÞÞq_ 1 q_ 3
(10)

1
τ2 ¼ ð m2 l22 þ m3 ðl22 þ l2c3 þ l2 l3 C3 ÞÞq
€2 m2 l2 l3 S3 q_ 2 q_ 3 þ m3 ðl2 lc3 C3 þ l2c3 Þq
€3 m3 l2 lc3 S3 q_ 3 þ
4
1 l2 1
ð m2 l2c2 sinð2q2 Þ þ m3 2 sinð2q2 Þ þ l2 lc3 m3 sinð2q2 þ q3 Þ þ l2c3 sin 2ðq2
2 2 2
þ q3 ÞÞq_ 21 þ lC2 m2 gC2 þ l2 m3 gC2 þ lC3 m2 gC23
(11)

Table 2. List of parameters of robot manipulator

Parameter Notation Value
Length of link 1 l1 0.15 m
Length of link 2 l2 0.5 m
Length of link 3 l3 0.5 m
Mass of link 1 m1 0.5 kg
Mass of link 2 m2 0.5 kg
Mass of link 3 m3 0.5 kg
Gravity g 9:81m=s2

Page 8 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 3. Equivalent circuit of

a chopper fed DC motor with
gear and mechanical load.

1
τ3 ¼ m3 l23 q
€3 m3 ðl2 lc3 C3 þ l2c3 Þq €2 þ m3 lc3 S23 ðl2 C2 þ lc3 C23 Þq_ 2 2
3
þ m3 ðl2 lc3 Sq_ 2 q_ 3 þ glc3 C23 Þ (12)

According to Euler-Lagrange’s equation, the 3-DOF robotic manipulator dynamics are given by three
coupled nonlinear differential equations of motion. Presenting this in vector form, we obtain Eq. (13).

€ þ BðqÞ½q;
τ ¼ MðqÞq _ q� _ 2 þ GðqÞ
_ þ CðqÞ½q� (13)

where, τ is the actuator’s torque vector,MðqÞis nxn symmetric and positively defined inertia matrix,
i:eMðqÞ ¼ MðqÞT >0, BðqÞ is coriolis torque matrix, and CðqÞ is centrifugal torque matrix, and GðqÞ is
_ q�is
the force of gravity, and is an n × 1 matrix, qis an n × 1 vector, ½q; _ the vector of joint velocity and
_ 2 is the derived vector given by: ½q_ 2 1 ; q_ 2 2 ; q_ 2 3 �T (Pezeshki et al., 2012).
½q�

Figure 4. Uncontrolled dynamic

model Simulink diagram.

Page 9 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 5. Simulation of the

open-loop dynamic model with
U1 = U2 = U3 = 0 v.

In order to obtain the dynamic equations that reflect the reality of the physical system, it is
essential to model (at least approximately) the forces of friction (Craig, 2009). Friction phenomena
in robot manipulators may affect the accuracy of the system for position control, especially when
the manipulator is moving at low velocities. For most engineering applications, a static friction
model is sufficient; experimental works proved that a good static friction model can approximate
the actual frictional forces with 90% of conformability (Marton & Lantos, 2007).

To simplify the model, the friction model consists of only the coulomb and viscous terms. Thus,
this simplified friction model is known as a classic friction model. The advantage of this model is its
simplicity with only two parameters. It can be described using Eq. (14) (Bittencourt, 2007).

_ ¼ FV q_ þ FS sgnðqÞ
FðqÞ _ (14)

_ is the friction force, FS represents the direction-dependent Coulomb friction coefficient,

where, FðqÞ
FV is the viscous damping coefficient, and q_ is the velocity of the system.

The manipulator is subjected to external disturbances and approximate friction (Fateh, 2008).
_
Eq. (13) is modified to give Eq. (15) which includes friction (FðqÞ), external dynamics, and other
terms as a global uncertainty term denoted by vector (τd ).

€ þ BðqÞ½q;
τ ¼ MðqÞq _ q� _ 2 þ GðqÞ þ FðqÞ
_ þ CðqÞ½q� _ þ τd (15)

3.4. Overall system model including actuators

In a number of literatures, the design of a controller for the robot manipulator control system
without introducing the actuator dynamics has been presented (Hu & Woo, 2006; Wai, 2003), while

Figure 6. Simulation of the

open-loop dynamic model with
U1 = U2 = U3 = 12 v.

Page 10 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 7. Block diagram of

adaptive (self-tuning) fuzzy
sliding mode controller with
fuzzy non-linear parts.

actuator dynamics plays an important role in the overall robot dynamics. Proper selection of
actuators will dictate how effectively a robot can perform a specific task.

3.4.1. Joint torque calculations

Selecting the right motor for the joint is not a random selection. It was done based on calculations
of expected weights that each joint will handle. The basic specification we look for when selecting
a motor for a joint is the holding torque. For the motor to be able to hold the link and all
subsequent links connected to it. To estimate the torque required at each joint, the worst-case
scenario must be chosen. The diagram to represent this is shown in Figure Figure 2.

where, W1 , W3 , and W5 are the weights of link one, two, and three respectively. The effecting
point of those weights is at the center of each link. W2 and W4 are the weights of the second and
third joint (motor at each joint was also taken into account). W6 is the maximum weight of the
object that the robot will lift. L1 ,L2 , andL3 are the lengths of link 1, link 2, and link 3, respectively.
The required torques at each joints are:

From the joint torque calculations, the calculated torques at the three joints that are used for
lifting the respective joints are 39.6 Nm, 38.9 Nm, and 15.9 Nm for joint one, two, and three,

Figure 8. Triangular member­

ship function for input (s).

Page 11 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 9. Triangular member­

ship function for output (α).

respectively. From the calculated joint torques, the joint actuators were selected from the PMDC
motor manufacturer’s catalog of Electrocraft (Yin et al., 2017). This was due to its higher perfor­
mance, reliability, and adjustable speed control. The dc drives play a vital role in modern industrial
drives. Therefore in this paper, a PMDC motor is used to drive all joints of the robotic manipulator,
and the speed control of the dc motor was obtained by using a chopper as a converter to vary the
armature voltage. The performance of the DC machine improved due to the use of the dc chopper,
making it an essential component for rapid transit systems. Figure Figure 3 shows the DC motor
(Sahin, 2005).

The mathematical equation representing the electrical and mechanical dynamics of

a permanent magnet DC motor is given by Eq. (16) (Spong & Vidyasagar, 2008).

dia dθm
Va ¼ Ria þ L þ Kb (16)
dt dt

Figure 10. Angular position

tracking error using ST-FSMC
without disturbance and para­
meter variation.

Page 12 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Table 3. Performance of the ST-FSMC angular position of joints

Performance Angular position joint Angular position joint Angular position joint
1 2 3
Rising time(s) 0.4351 0.4283 0.3714
Settling time(s) 0.8261 0.8010 0.6906
Overshoot (%) 0 0 0
Steady state error 0.0008 0.0052 0.005
RMS error 0.4395 0.1603 0.3348
Chattering Free Free Free
Average Steady state 0.0037
error (rad)

where, Va is the armature voltage of the motor, R and L are armature equivalent resistance and
inductance, respectively. Kb is the back electromotive force constant, ia is the armature current, and
θm denotes the angular position of motor.

Hence, the electromagnetic torque is due to the current through the armature winding and is
proportional with this current. Mathematically, this can be written as:

τm ¼ Km ia (17)

where,τm is the generated motor torque, and Km is the diagonal matrix of the motor torque
constant.

It ought to be noted that the application control input for driving the manipulator is the
armature voltage of the dc motor. Therefore, by utilizing Eqs. (16) and (17), and ignoring
inductanceL, due to its tiny amount (Sahin, 2005), the Eq. (18) is obtained.

Va ¼ RKm 1
½gr MðqÞ þ Jm gr 1 € þ RKm1 ½gr BðqÞðq;
�q _ qÞ _ 2 � þ RKm1 ½βm gr
_ þ gr CðqÞðqÞ 1
(18)
þ R 1 Km Kb gr 1
�q_ þ RKm1 gr GðqÞ þ R 1 Km ½gr FðqÞ
_ þ gr τd �

Finally, the complete nonlinear dynamic model representation of the 3-DOF robotic manipulator is
given as:

2 3 2 3 1 82 3
€1
q 0:005r12 þ 0:029 þ 0:035C22 þ 0:0067C223 þ 0:028C2 C23 0 0 < Va1
>
6€ 7 6 7 6 7
4 q2 5 ¼ 4 0 0:076 þ 0:02778C3 0:0092 þ 0:014C3 5 4 Va2 5
>
:
€3
q 0 0:017 þ 0:02526C3 0:0707 Va3
2 32 3 2 32 2 3 2 39
b11 b12 0 q_ 1 q_ 2 0 0 0 q_ 1 0:25 þ 2:66q_ 1 þ 0:04sgnðq_ 1 Þ >
=
6 76 7 6 76 7 6 7
4 0 0 b23 54 q_ 1 q_ 3 5 4 c22 0 c23 54 q_ 2 2 5 4 0:81C2 þ 0:272C23 þ 0:25 þ 2:66q_ 2 þ 0:04sgnðq_ 2 Þ 5
>
;
0 0 0 q_ 2 q_ 3 c31 c32 0 q_ 23 0:248C23 þ 0:021 þ 2:43q_ 3 þ 0:04sgnðq_ 3 Þ
(19)
3.4.2. Open loop system dynamic model description of the 3-DOF robotic manipulator
The open-loop dynamic model without using the controller was obtained using Matlab/Simulink as
shown in Figure Figure 4. To check the systems validity, fixed input voltages are selected for the
manipulator’s joints, and afterward, the block is simulated.

The open-loop uncontrolled dynamic model shows the impact of the input on the output state
when a joint angle for computing input voltages U1, U2, and U3 for joints one, two, and three are
utilized. Then the models simulation results are obtained for various input levels. Results in Figure 5

Page 13 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 11. 3D-helical trajectory

tracking using ST-FSMC.

show that when all the input voltages of the manipulator are at an equilibrium or all three input
voltages are at 0 v, then both the joints two and three are oscillated due to the gravitational
impact of the earth. From the simulation results in Figures 5 and 6, it’s clear that the robot
manipulator system is a highly coupled, nonlinear, and unstable system. Therefore, this has
brought about the need to design a controller with an optimal performance.

4. ST-FSMC Controller design

Adaptation law plus FSMC can give a system capability of online-tuning and improve the systems
trajectory tracking performance. The fuzzy adaptive technique can train the parameters variation
by expert knowledge. Fuzzy adaptive method was used to solve the complicated uncertain

Figure 12. Angular position

tracking errors using PID con­
troller without disturbance and
parameter variation.

Page 14 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 13. Control effort of joint

one using PID without distur­
bance and parameter variation.

Figure 14. Control effort of joint

two using PID without distur­
bance and parameter variation.

nonlinear system (Darajat & Istiqphara, 2020), the integration of this adaptive method with the

Page 15 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 15. Control effort of joint

three using PID without distur­
bance and parameter variation.

Figure 16. Angular position

tracking errors using SMC with
signum function without dis­
turbance and parameter
variation.

Page 16 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 17. Angular position

tracking errors using SMC with
boundary layer (sat function)
without disturbance and para­
meter variation.

Figure 18. Control effort of joint

one using SMC with saturation
function without disturbance
and parameter variation.

Page 17 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 19. Control effort of joint

two using SMC with saturation
function without disturbance
and parameter variation.

FSMC, therefore, helps the system to enhance its performance by online tuning of nonlinear and
time-variant parameters (Jalali et al., 2013).

In a fuzzy controller, adaptation is divided into a direct adaptive technique and an indirect
adaptive technique. The direct adaptive technique adjusts/tunes the controller inputs parameters,
and an indirect supervisory technique adjusts/tunes the parameters of the control system depend­
ing on the performance error (Huang et al., 2018; Kim et al., 2020; Trigatti et al., 2018).
A supervisory controller is a kind of adaptive controllers which searches to see the current
performance of the system and adjusts the controller parameter to increase the systems
performance.

In this paper, a saturation function is utilized rather than a signum function in the discontinuous
part of the SMC to eliminate the chattering effect on the control effort. Moreover, the adaptive
fuzzy inference scheme is used to avoid perturbation. The ST-FSMC is proposed to discard the
unstructured model uncertainty. Lamda (λ) and (K) are the coefficients having an important
impact on the discontinuous part of the control system. In the proposed controller for the self-
tuning part, these two significant gains are adjusted. Therefore, if these coefficients are tuned
properly, the SMC can discard perturbations more desirably. Accordingly, Self-tuning-FSMC has
powerful resistance and can solve uncertainty problems. The block diagram of the proposed self-
tuning FSMC is presented in Figure 7.

To ensure that the system is asymptotically stable during the reaching phase, Lyapunov’s
stability theorem was applied. Lyapunov-based method was utilized to obtain conditions of the
sliding surface. It gives a positive scalar function called the Lyapunov candidate function presented
in Eq. (21). For the system state variable, the control law that will diminish this function has been
assumed as:

Page 18 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 20. Control effort of joint

three using SMC with saturation
function without disturbance
and parameter variation.

V>0 (20)

For picking the Lyapunov candidate function, there is no standard law. To investigate the stability
of the system, the positive Lyapunov candidate function for each separate dynamic system best
suitable function was chosen as follows:

1 2 1
VðsÞ ¼ s ; VðsÞ ¼ s2 >0 (21)
2 2

_
VðsÞ ¼ ss_ ¼ sð KsignðsÞÞ ¼ jsjK<0; . . . dots::if . . . . . . K>0 (22)

where, sands_ are defined as a sliding function and its derivative, and K is a design constant that
must be higher than zero to meet the Lyapunov stability condition in the Eqs. (21) and (22).

The FLC adjusts the system to different disturbances with one input, s and one output, α as
shown in Figures 8 and 9. This Fuzzy Logic supervisor increases the trajectory tracking performance
by online tuning of the controller parameters using the output, α, through the gain, G1, and G2. The
scaled factors are: G1 ¼ 0:9, G2 ¼ 10 for the first joint; G1 ¼ 1, G2 ¼ 5 for the second joint, and
G1 ¼ 1, G2 ¼ 5for the third joint.
5. Results and discussion
In this section, the performance and stability of the proposed control method were analyzed with
and without uncertainty (external disturbance and parameter variation). The simulations were
performed using MATLAB/Simulink. To confirm the performance of the proposed controller, the
simulation results were compared with the conventional controllers.

Page 19 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

For experimental purpose, the desired trajectory for the robotic manipulator is picked arbitrarily
to check the tracking capability of the proposed controller (STFSMC) and other conventional
controllers. Based on the results, ST-FSMC achieved a better performance than all other conven­
tional controllers.

The robustness of the designed controller was also tested based on its power of disturbance
elimination and sensitivity to parameter variation (mass uncertainty). This was also obtained for
the conventional controllers. The value of disturbance applied to the desired trajectory of the joint
variable was predefined. The mass term was considered as parameter variation. Based on the
results, ST-FSMC was found to be more robust than PID, SMC, and FSMC.

5.1. Performance of proposed controller (ST-FSMC)

The response of ST-FSMC for the 3-DOF robot manipulator is shown in Figure 10 for angular
position tracking error performance of joints one, two, and three respectively. Figure 10 shows
that the tracking error converges to zero. In Table 3, the performance of the ST-FSMC angular
position of joints with minimum time and without overshot as needed for the design of the ST-
FSMC was presented. The results show that the end effector of the robot manipulator asymptoti­
cally follows the desired trajectory with minimum acceptable error. This indicates that the ST-FSMC
control system is stable.

Figure 11 shows the trajectory tracking performance of the end-effector of the robot manip­
ulator using ST-FSMC for 3D-helical trajectory. Figure 11 shows 3D-helical desired trajectory which
is produced when the position of the end effector is starting initially from point ð100; 0; 15Þcm
following the trajectoriesð70 sinðtÞ; 50 cosðtÞ; 15 þ 3:5tÞcm in theX, Y, and Zcoordinates,
respectively.

5.2. Performance comparison of proposed controller with other conventional controller

without disturbance and parameter variation
The comparison of the designed controller (ST-FSMC, FSMC, SMC, and PID controller) for trajectory
tracking performance is discussed as follows. Figure 12 shows the plot of the angular position
tracking error using PID for joints one, two, and three. The results show that the steady-state error
increased to 0.04 rad, 0.036 rad and 0.0476 rad (i.e., the average steady-state error was increased
to 0.0413 rad), while the overshoot increased, rise time decreased by 0.1495 sec, 0.0975 sec, and
0.257sec, and settling time is increased by 5.56 sec, 0.69 sec, and 1.3 sec for joint one, two, and
three respectively when the PID controller is used.

Figures 13, 14, and 15 show the control effort of joint one, joint two, and joint three of the PID
controller, respectively. The results indicate that a very high control effort is needed to track the
desired trajectory using PID relative to the proposed controller (ST-FSMC) control effort. The
maximum voltage used to track the desired trajectory with PID needs 48 v for all three joints.
This is increased by 23.5 v, 29 v, and 16.8 v for joint one, joint two, and joint three respectively.

Figure 16 shows the plot of the angular position tracking error using pure SMC for joints one, two,
and three respectively. Based on the results obtained, the performance of the pure SMC is almost
the same as that of ST-FSMC for all three joints. The results show the performance of angular
position of each joint in case of SMC with signum function (pure SMC) is also the same performance
as compared to ST-FSMC, but it introduces high-frequency oscillation (chattering phenomena) on
the control effort for tracking the desired trajectory. The chattering observed in the corresponding
results with pure SMC is evident in the difficulty involved in the practical design of controller using
this technique. This is one of its draw-backs.

To overcome the problem of chattering in pure SMC, a saturation function was replaced in place
of signum function. Figure 17 shows the plot of the angular position tracking error for joints one,
two, and three. The results show that the steady-state error for SMC with boundary layer is

Page 20 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

0.006 rad for joint one, 0.0032 rad for joint two, and 0.0036 rad for joint three (i.e., the average
steady-state error increases to 0.0044), and the overshoot remains the same, while both the rise
time and settling time were increased when SMC with saturation function was used.

We can conclude based on the simulation result that due to the replacement of the saturation
function with signum function, the error performance increased against the considerable chatter­
ing elimination. Figures 18, 19, and 20 show the control effort of joints one, two, and three of SMC
with saturation function respectively. The maximum voltage required to track the desired trajec­
tory by SMC with saturation function is 25 v for joint one, 21.4 v for joint two, and 34.26 v for joint
three. This is increased by 0.5 v, 2.4 v, and 3.06 v respectively compared to ST-FSMC.

Figure 21 shows the results obtained for the angular position tracking error using FSMC. The
simulation results show that the performance of FSMC is almost similar to that of SMC with
saturation function in terms of overshoot size, rising and settling time, and steady-state error.
The steady-state error of the joint one for the FSMC is 0.0052, it is 0.0065 for the SMC with
saturation. The average steady-state error of three joints for the FSMC is 0.0053, for the SMC it is
0.0044 with saturation. Also the rising and settling time for the FSMC is almost similar to that of
SMC with saturation function.

Figures 22, 23, and 24 show the control effort of joint one, two, and joint three of FSMC
respectively. It requires high control effort relative to the ST-FSMC to track the desired trajectory.
The maximum voltage used to track the desired trajectory by FSMC needs 25 v for joint one, 24.5 v
for joint two, and 31.2 v for joint three. This increases by 0.6 v, 5.5 v, and 0 v respectively.

According to the simulation result, the performance of ST-FSMC and pure SMC is better than
FSMC, SMC with saturation function and PID controller for controlling robotic manipulators. The

Figure 21. Angular position

tracking errors using FSMC
without disturbance and para­
meter variation.

Page 21 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

performance of the ST-FSMC was found to be superior to that of FSMC, SMC, and PID controllers
with relatively low control effort required to track the desired trajectory.

5.3. Disturbance rejection and sensitivity to parameter variation of the controller

In the previous section, we showed the trajectory tracking performance without parameter varia­
tion and external disturbance in the system. In this section, the results with parameter variation
and disturbance are presented and discussed.

5.3.1. Sensitivity to parameter variation of PID, SMC, FSMC, and ST-FSMC

In this section, the internal parameter variation (link parameter variation) is considered and then
the performance of the ST-FSMC, FSMC, SMC, and PID is compared. For parameter variation, all the
parameters are kept the same as presented in Table 2, except for m1 , m2 ; andm3 which are
changed to 2 kg each, that means considering an uncertainty in the value of the system to be
controlled.

With reference to the simulation results obtained in Figures 25–28, we also compared the
performance of the proposed controller with another conventional controller in the presence of
structured uncertainty (link parameter variation). According to the simulation result, the perfor­
mance of ST-FSMC results with parameter variation and without parameter variation is almost the
same (i.e., rising time, settling time, overshoot, and steady-state error are approximately equal)
relative to other controllers, while the average steady-state error for the three joints was increased
by 0.0023. However, in the case of PID, SMC with the boundary layer, and FSMC controller, the
system is highly affected by the parameter variation relative to ST-FSMC. For the PID controller with
parameter variation, the rising time was increased by 0.05 sec, 0.06 sec, and 0.1 sec, while the
settling time also increased by 1 sec, 1.3 sec, and 2.5 sec for joints one, two, and three respectively.

Figure 22. Control effort of joint

three using FSMC without dis­
turbance and parameter
variation.

Page 22 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 23. Control effort of joint

three using FSMC without dis­
turbance and parameter
variation.

Figure 24. Control effort of joint

three using FSMC without dis­
turbance and parameter
variation.

The average steady-state error for three joints is increased by 0.0427. For SMC with saturation
function and parameter variation, the rising time was increased by 0.51 sec, 0.002 sec, and 0.0733

Page 23 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 25. Angular position

tracking errors using PID with
parameter variation.

sec, while the settling time was increased by 1.02 sec, 0.02 sec, and 0.08 sec for joints one, two,
and three respectively. The average steady-state error for the three joints was increased by 0.0336.
For FSMC with parameter variation, the rising and setting time are almost the same as that of FSMC

Figure 26. Angular position

tracking errors using SMC with
parameter variation.

Page 24 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 27. Angular position

tracking errors using FSMC with
parameter variation.

Figure 28. Angular position

tracking errors using ST-FSMC
with parameter variation.

without parameter variation, but the overshoot increased, and also the average steady-state error
for the three joints is increased by 0.0073.

Page 25 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 29. Disturbance torques

due to vibration.

5.3.2. Disturbance rejection capability of PID, SMC, FSMC, and ST-FSMC

Robot manipulators are often tasked with working in environments with vibrations and are subject
to load uncertainty. In this section, the proposed controllers are tested in terms of rejecting the
uncertain disturbance torques caused by external vibration and payload variation. The Figures 29,
30, and 31 show the disturbance torques due to vibration, the disturbance torques due to payload
variation effects, and the sum of disturbance torques due to vibration and payload variation
respectively.

Figures 32, 33, 34, and 35 show the plot of angular position tracking errors for each joint using
PID, SMC FSMC, and ST-FSMC respectively under uncertain external disturbances. The results show
the performance of the angular position of each joint. ST-FSMC showed a high capability in
handling the newly introduced uncertain external disturbance as compared to other conventional
controllers as shown in Figure 35 of the angular position tracking error for joints one, two, and
three using STFSMC under external disturbance. It’s clear that the results show that the designed
controller (ST-FSMC) makes the system insensitive to parametric variations (it has good perfor­
mance with variation in mass parameters). It is also robust and performs well even in the presence
of external disturbances with minimum tracking error as compared to PID, SMC, and FSMC

Figure 30. Disturbance torques

due to payload variation effect.

Page 26 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 31. Disturbance torques

due to vibration and payload
variation combination.

controllers with considerable control effort. Generally from the results, the ST-FSMC is more robust
than the FSMC, SMC, and PID controller.

5.4. Application of robot manipulators in 3-D industrial target task

5.4.1. Robot manipulator for painting and welding in automation industry

One of the tasks performed in a vehicle automation industry is painting and welding the upper part
of the vehicles body. The trajectory tracking performance of a robot manipulator was discussed in

Figure 32. Angular position

tracking errors using PID under
uncertain external disturbance.

Page 27 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 33. Angular position

tracking errors using SMC under
uncertain external disturbance.

the previous sections of this paper utilizing the proposed control system with a 3D-helical trajec­
tory as the desired trajectory and various conventional controllers. The trajectory planning of the
3-DOF robot manipulator in welding and painting the vehicles body was explored in this section to
improve the quality of painting and welding in the automobile industry. In summary, the

Figure 34. Angular position

tracking errors using FSMC
under uncertain external
disturbance.

Page 28 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 35. Angular position

tracking errors using ST-FSMC
under uncertain external
disturbance.

advancement of robotized welding and painting is very remarkable, and is now one of the most
important aspects of the automobile industry. Because of these advantages for humans, it is
excellent for use in the automotive industry to improve both painting and welding operations. As
shown in Figure 36, the 3D-square wave trajectory was used as the desired trajectory for welding
and painting of vehicles body.

The trajectory tracking performance of the 3D-square wave trajectory is shown in Figures 37 and
38. The actual trajectory is compared with the desired trajectory using PID, SMC, FSMC, and the
proposed ST-FSMC controllers. Integral square error (ISE), integral absolute error (IAE), integral
time square error (ITSE), and integral time absolute error (ITAE) indices were used as performance
measures. The results are summarized in the Tables 4, 5, 6 and 7.

Figure 36. 3D-square wave tra­

jectory of the manipulator for
welding and painting of vehi­
cles body.

Page 29 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Figure 37. 3D-square wave tra­

jectory planning of the 3-DOF
robot manipulator for welding
and painting of vehicles body.

Figure 38. Trajectory tracking

control of the 3-DOF robot
manipulator.

Table 4. Performance Index for PID controller

Trajectory ISE IAE ITSE ITAE
Dynamics
X-axis 1.073 0.512 0.08145 0.09059
Y-axis 3.089 0.8813 0.2428 0.1614
Z-axis 0.3796 0.2866 0.02333 0.09409

Table 5. Performance Index for SMC controller

Trajectory ISE IAE ITSE ITAE
Dynamics
X-axis 0.4223 0.1854 0.01181 0.008821
Y-axis 0.4672 0.336 0.03469 0.06729
Z-axis 0.1693 0.1605 0.02211 0.9636

Page 30 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Table 6. Performance Index for FSMC controller

Trajectory ISE IAE ITSE ITAE
Dynamics
X-axis 1.798*10−1 0.01295 8.721*10−1 0.06423
Y-axis 0.1796 0.1866 0.004333 0.0019409
Z-axis 5.429*10−1 0.006993 0.0002874 0.03558

Table 7. Performance Index for ST-FSMC Controller

Trajectory ISE IAE ITSE ITAE
Dynamics
X-axis 3.69*10−2 0.001133 1.829*10−1 0.00513
Y-axis 0.01251 0.1020 0.001016 0.000258
Z-axis 8.438*10−2 0.005329 1.376*10−1 0.02348

The performance indices shown in Tables 4–7 show that a smaller error is obtained by ST-FSMC
than when PID, SMC, and FSMC controllers were used in welding and painting the vehicles body.
Finally, it can be said that the proposed controller for the 3-DOF robot manipulator control has
a good performance and good transient response when excited with both 3-D helical and
3-D square wave trajectories.

6. Conclusion
In this paper, a self-tuning fuzzy sliding mode controller (ST-FSMC) was proposed. The ST-FSMC, PID,
SMC, and FSMC were successfully designed and comparative analysis was carried out. Simulations
were performed to demonstrate that the ST-FSMC is more effective in terms of the response time
performance than other conventional controllers (PID, SMC, and FSMC) for a given desired trajectory
tracking because the proposed controller can adjust itself to the system parameter’s variations and
external disturbances. The simulation results further show the high ST-FSMC tracking performance
and ability of the system to solve the problem of handling unstructured uncertainties and achieve
stability, robustness as well as reliability. The stability of the designed controller was verified using
Lyapunov`s Direct Method. Also, various evaluation criteria were chosen to compare the performance
of the controller designed (ST-FSMC) with other controllers such as PID, SMC, and FSMC. The system
performance in both SMC and FSMC were observed to have difficulty in handling uncertain external
disturbances (unstructured uncertainties) because the sliding surface gain was chosen by trial and
error, which means that the SMC and FSMC must have prior knowledge of the system uncertainty. In
order to solve the problem of handling unstructured uncertainties to increase the robustness of the
system and give an improved output, FSMC and Fuzzy base tuning controller methods were com­
bined. This control method was implemented in the robotic manipulator to give a high-performance
nonlinear controller in the presence of uncertainties. This paper assumed that the working environ­
ment of the system is clean (i.e., no obstacles around the working environment). Therefore, 3-DOF
was used for reducing redundancy to perform a given task cyclically. In future works, authors intend
to extend the system to better degree of freedom by considering the presence of obstacles to execute
various tasks in different working environments by applying the same or any other control methodol­
ogy addressed in this research work. Redundant manipulators are extensively used because of their
ability to perform a prescribed end-effector motion (tracking problem), while taking into account
extra goals such as obstacle avoidance or the minimization of energy consumption, or other cost
functions. This is particularly important for manufacturing tasks, which often require robots to per­
form a repetitive task while keeping energy and time requirements as low as possible.

Page 31 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Funding Journal of Dynamics and Control, 9(1), 363–378.

The authors received no direct funding for this research. https://doi.org/10.1007/s40435-020-00638-7
Getachew, D. (2018). Sliding mode control of a two degree of
Author details freedom robot arm using permanent magnet synchro­
Aderajew Ashagrie1 nous motor. http://localhost:80/xmlui/handle/
E-mail: [email protected] 123456789/12687
ORCID ID: http://orcid.org/0000-0003-4411-2410 Ghaleb, N. M., & Aly, A. A. (2018). Modeling and control of
Ayodeji Olalekan Salau2 2-DOF robot arm. International Journal of Emerging
ORCID ID: http://orcid.org/0000-0002-6264-9783 Engineering Research and Technology, 6(11), 24–31.
Tilahun Weldcherkos1 https://www.ijeert.org/papers/v6-i11/3.pdf
ORCID ID: http://orcid.org/0000-0002-7898-2791 Hu, H., & Woo, P. (2006). Fuzzy supervisory sliding-mode
1
Department of Electrical and Computer Engineering, and neural network control for robotic manipulators.
Addis Ababa Science and Technology University (AASTU), IEEE Transactions on Industrial Electronics, 53(3),
Addis Ababa, Ethiopia. 929–940. https://doi.org/10.1109/TIE.2006.874261
2
Department of Electrical/Electronics and Computer Huang, J., Hu, P., Wu, K., & Zeng, M. (2018). Optimal
Engineering, Afe Babalola University, Ado-Ekiti, Nigeria.. time-jerk trajectory planning for industrial robots.
Mechanism and Machine Theory, 121, 530–544. https://
Citation information doi.org/10.1016/j.mechmachtheory.2017.11.006
Cite this article as: Modeling and control of a 3-DOF Jalali, A., Piltan, F., Gavahian, A., Jalali, M., & Adibi, M.
articulated robotic manipulator using self-tuning fuzzy (2013). Model-free adaptive fuzzy sliding mode con­
sliding mode controller, Aderajew Ashagrie, Ayodeji troller optimized by particle swarm for robot
Olalekan Salau & Tilahun Weldcherkos, Cogent manipulator. International Journal of Information
Engineering (2021), 8: 1950105. Engineering and Electronic Business, 5(1), 68. https://
doi.org/10.5815/ijieeb.2013.01.08
References Kapoor, N., & Ohri, J. (2013). Fuzzy sliding mode controller
Abbas, Z. (2018). Motion control of robotic arm manipula­ (FSMC) with global stabilization and saturation func­
tor using pid and sliding mode technique. PhD thesis. tion for tracking control of a robotic manipulator.
Capital University. Journal of Control and Systems Engineering, 1(2), 50.
Alassar, A. Z., Abuhadrous, I. M., & Elaydi, H. A. (2010). https://doi.org/10.18005/JCSE0102003
Comparison between FLC and PID controller for 5DOF Kapoor, N., & Ohri, J. (2017). Sliding Mode Control (SMC) of
robot arm. 2nd IEEE International Conference on robot manipulator via intelligent controllers. Journal
Advanced Computer Control, 5, 277–281. Shenyang, of the Institution of Engineers (India): Series B, 98(1),
China. 83–98. https://doi.org/10.1007/s40031-016-0216-x
Anh, H. P. H., Kien, C. V., Son, N. N., & Nam, N. T. (2018). New Karamali Ravandi, A., Khanmirza, E., & Daneshjou, K. (2018).
approach of sliding mode control for nonlinear uncer­ Hybrid force/position control of robotic arms manipu­
tain pneumatic artificial muscle manipulator enhanced lating in uncertain environments based on adaptive
with adaptive fuzzy estimator. International Journal of fuzzy sliding mode control. Applied Soft Computing, 70,
Advanced Robotic Systems, 1–11. https://doi.org/10. 864–874. https://doi.org/10.1016/j.asoc.2018.05.048
1177/1729881418773204 Kim, J., Jin, M., Park, S. H., Chung, S. Y., & Hwang, M. J.
Bittencourt, A. C. (2007). Friction change detection in (2020). Task space trajectory planning for robot
industrial robot arms. MSc Thesis KTH Electrical manipulators to follow 3-d curved contours.
Engineering and Computer Science, Sweden, 1-79. Electronics, 9(9), 1424. https://doi.org/10.3390/
Chang, Y., Yen, H., & Wu, H. (2008). An intelligent robust electronics9091424
tracking control for electrically-driven robot systems. Korayem, M. H., Alamdari, A., Haghighi, R., &
International Journal of Systems Science, 39(5), Korayem, A. H. (2010). Determining maximum
497–511. https://doi.org/10.1080/ load-carrying capacity of robots using adaptive
00207720701832747 robust neural controller. Robotica, 28(7), 1083.
Craig, J. J. (2009). Introduction to robotics: Mechanics and https://doi.org/10.1017/S0263574710000081
control, E. Pearson Education India. Marton, L., & Lantos, B. (2007). Modeling, identification,
Darajat, A. U., & Istiqphara, S. (2020). Control of and compensation of stick-slip friction. IEEE
two-link robot manipulator with uncertainty para­ Transactions on Industrial Electronics, 54(1),
meter using self-tuning sliding mode control. 3rd 511–521. https://doi.org/10.1109/TIE.2006.888804
International Conference on Mechanical, Electronics, Massaoudi, F., Elleuch, D., & Damak, T. (2019). Robust
Computer, and Industrial Technology (MECnIT), control for a two DOF robot manipulator. Journal of
Medan, Indonesia. https://doi.org/10.1109/mec Electrical and Computer Engineering, 2019, 1–11.
nit48290.2020.9166643 https://doi.org/10.1155/2019/3919864
Medhaar, H., Derbel, N., & Damak, T. (2006). A decoupled
Dhyani, A., Panda, M. K., & Jha, B. (2020). Design of an
fuzzy indirect adaptive sliding mode controller with
evolving fuzzy-PID controller for optimal trajectory
application to robot manipulator. International
control of a 7-DOF redundant manipulator with
Journal of Modelling, Identification and Control, 1(1),
prioritized sub-tasks. Expert Systems with
23–29. https://doi.org/10.1504/IJMIC.2006.008644
Applications, 162, 113021. https://doi.org/10.1016/j.
Ohri, J., Vyas, D. R., & Topno, P. N. (2011). Comparison of
eswa.2019.113021
robustness of PID control and sliding mode control of
Fateh, M. M. (2008). On the voltage-based control of robot
robotic manipulator. International Symposium on
manipulators. International Journal of Control,
Devices MEMS Intelligent Systems and
Automation, and Systems, 6(5), 702–712. https://doi.
Communication (ISDMISC), Bhopal, India, 5–10.
org/10.1007/s12555-017-0035-0
Pezeshki, S., Badalkhani, S., & Javadi, A. (2012). Performance
Gambhire, S. J., Kishore, D. R., Londhe, P. S., & Pawar, S. N. analysis of a neuro-pid controller applied to a robot
(2021). Review of sliding mode based control tech­ manipulator. International Journal of Advanced Robotic
niques for control system applications. International Systems, 9(5), 163. https://doi.org/10.5772/51280

Page 32 of 33
Ashagrie et al., Cogent Engineering (2021), 8: 1950105
https://doi.org/10.1080/23311916.2021.1950105

Piltan, F., Yarmahmoudi, M. H., Shamsodini, time-varying uncertainties. 59th IEEE International
M., Mazlomian, E., and Hosainpour, A. (2012a). PUMA- Midwest Symposium on Circuits and Systems
560 robot manipulator position computed torque (MWSCAS), 283–294.
control methods using Matlab/Simulink and their Trigatti, G., Boscariol, P., Scalera, L., Pillan, D., &
integration into graduate nonlinear control and Gasparetto, A. (2018). A look-ahead trajectory plan­
Matlab courses. International Journal of Robotics and ning algorithm for spray painting robots with
Automation, 3(3), 167–191 non-spherical wrists. IFToMM symposium on
Piltan, F., Rahmdel, S., Mehrara, S., and Bayat, R. (2012b). mechanism design for robotics, 235–242.
Sliding mode methodology vs. Computed torque Wai, R. (2003). Tracking control based on neural network
methodology using matlab/simulink and their inte­ strategy for robot manipulator. Neurocomputing, 51,
gration into graduate nonlinear control courses. 425–445. https://doi.org/10.1016/S0925-2312(02)
International Journal of Engineering, 6(3), 142–177. 00626-4
https://www.cscjournals.org/manuscript/Journals/ Wang, Y., Shi, R., & Wang, H. (2014). ESO-based fuzzy
IJE/Volume6/Issue3/IJE-369.pdf sliding-mode control for a 3-DOF serial-parallel
Rahmdel, S., Bairamai, M., & Fahham, M. M. H. R. (2012). hybrid humanoid arm. Journal of Control Science and
Design chattering-free performance-based fuzzy Engineering, 1–9. https://doi.org/10.1155/2014/
sliding mode controller for bus suspension. 304590
International Journal of Engineering Sciences Yin, H., Li, Y., & Li, J. (2017). Decomposed dynamic
Research (IJESR), 796–802. control for a flexible robotic arm in consideration
Sahin, H. (2005). Design of a secondary packaging robotic of nonlinearity and the effect of gravity. Advances
system. PhD Graduate School of Natural and Applied in Mechanical Engineering, 9(2),
Sciences of Middle East Technical University, p. 156. 168781401769410. https://doi.org/10.1177/
Shaei, S. E. (2010). Sliding mode control of robot manip­ 1687814017694104
ulators via intelligent approaches. Citeseer. Zhang, H., Fang, H., Zhang, D., Luo, X., & Zou, Q. (2020).
Spong, M. W., & Vidyasagar, M. (2008). Robot dynamics Adaptive fuzzy sliding mode control for a 3-DOF
and control. John Wiley & Sons, 2008, 352. parallel manipulator with parameters uncertainties.
Tao, C. M., & Lee, T. (2016). Adaptive fuzzy sliding mode Complexity, 2565316, 16. https://doi.org/10.1155/
controller for linear systems with mismatched 2020/2565316

© 2021 The Author(s). This open access article is distributed under a Creative Commons Attribution (CC-BY) 4.0 license.
You are free to:
Share — copy and redistribute the material in any medium or format.
Adapt — remix, transform, and build upon the material for any purpose, even commercially.
The licensor cannot revoke these freedoms as long as you follow the license terms.
Under the following terms:
Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made.
You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.
No additional restrictions
You may not apply legal terms or technological measures that legally restrict others from doing anything the license permits.

Cogent Engineering (ISSN: 2331-1916) is published by Cogent OA, part of Taylor & Francis Group.
Publishing with Cogent OA ensures:
• Immediate, universal access to your article on publication
• High visibility and discoverability via the Cogent OA website as well as Taylor & Francis Online
• Download and citation statistics for your article
• Rapid online publication
• Input from, and dialog with, expert editors and editorial boards
• Retention of full copyright of your article
• Guaranteed legacy preservation of your article
• Discounts and waivers for authors in developing regions
Submit your manuscript to a Cogent OA journal at www.CogentOA.com

Page 33 of 33