Received May 8, 2019, accepted June 16, 2019, date of publication June 19, 2019, date of current version

July 9, 2019.
Digital Object Identifier 10.1109/ACCESS.2019.2923916

Robust Navigational Control of a Two-Wheeled

Self-Balancing Robot in a Sensed Environment
CELESTINE IWENDI 1 , (Senior Member, IEEE), MOHAMMED A. ALQARNI 2 ,
JOSEPH HENRY ANAJEMBA 3 , AHMED S. ALFAKEEH4 , ZHIYONG ZHANG1 , (Member, IEEE),
ALI KASHIF BASHIR 5 , (Senior Member, IEEE)
1 Department of Electronics, BCC of Central South University of Forestry and Technology, Changsha 410004, China
2 College of Computer Science and Engineering, University of Jeddah, Jeddah 21959, Saudi Arabia
3 Department of Communication Engineering, College of Internet of Things, Hohai University, Changzhou Campus, Changzhou 213000, China
4 Faculty of Computing and Information Technology, King Abdulaziz University, Jeddah 21589, Saudi Arabia
5 Department of Computing and Mathematics, Manchester Metropolitan University, Manchester M13 9PL, U.K.

Corresponding author: Joseph Henry Anajemba ([email protected])

ABSTRACT This research presents an improved mobile inverted pendulum robot called Two-wheeled
Self-balancing robot (TWSBR) using a Proportional-Derivative Proportional-Integral (PD-PI) robust control
design based on 32-bit microcontroller in a sensed environment (SE). The robot keeps itself balance with
two wheels and a PD-PI controller based on the Kalman filter algorithm during the navigation process and is
able to stabilize while avoiding acute and dynamic obstacles in the sensed environment. The Proportional (P)
control is used to implement turn control for obstacle avoidance in SE with ultrasonic waves. Finally,
in a SE, the robot can communicate with any of the Internet of Things (IoT) devices (mobile phone or
Personal Computer) which have a Java-based transmission application installed and through Bluetooth
technology connectivity for wireless control. The simulation results prove the efficiency of the proposed
PD-PI controller in path planning, and balancing challenges of the TWSBR under several environmental
disturbances. This shows an improved control system as compared to the existing improved Adaptive Fuzzy
Controller.

INDEX TERMS TWSBR, 32-bit Microcontroller, robust control, PD-PI, sensed environment, IoT.

I. INTRODUCTION study in research such as how they can be implemented to

In recent years, robots which are an integration technology of not only adapt to specific environments and requirements of
multi-disciplinary theory and technology have been widely missions, but also reflect the highly flexible adaptability in
used in various fields as a new type of production tool. the dynamic change of complicated unknown environments.
This is because robots show great superiority in reducing Based on this context, self-balancing two-wheeled robots
labor intensity, improving labor productivity and reducing have emerged [2].
the dangers of operation. According to [1], robots can be As a new research, the self-balancing two-wheeled robot
divided into two categories: the first is the operating robot combines the characteristics of wheeled and autonomous
which can manipulate its mechanical arms for the movement mobile robots, and has advantages that compare with tradi-
of objects and the other is the mobile robot; a walking robot tional robot technology [3].
with mechanical legs or wheels. However, the mobile robot Firstly, its turning radius is zero because it can rotate
has become an important branch in the field of robotics due around the center of its body, which can make it flexible in a
to its wide prospect of application. Today, amongst differ- narrow place. Secondly, the CPU automatically gives positive
ent kinds of mobile robots, the robot with wheels is most and negative torque to achieve fast and stable braking, which
popular and has become a part of robot research because means its control is extremely convenient. Thirdly, with the
of its simple structure, low cost and high energy efficiency. continuous development of microelectronics, computer tech-
Wheeled mobile robots, are a focal point undergoing intense nology, control technology, power technology, drive tech-
nology and sensor technology in a SE, the self-balancing
The associate editor coordinating the review of this manuscript and two-wheeled robot has been provided a solid theoretical foun-
approving it for publication was Yin Zhang. dation and greatly decreases its development cost.

VOLUME 7, 2019 This work is licensed under a Creative Commons Attribution 3.0 License. For more information, see http://creativecommons.org/licenses/by/3.0/ 82337
 C. Iwendi et al.: Robust Navigational Control of a TWSBR in a Sensed Environment

Internet of Things (IoT) is considered a communication

between the physical and digital spheres. The digital sphere
communicates with the physical through a collection of sev-
eral sensors and actuators. These functionalities are applied
to either possibly change the state of an object or query its
state. Technically, IoT can be considered as a new kind of
phase where virtually all the devices and gadgets that we use
are interconnect and controlled, this includes Robots [4].
The IoT devices considered in this paper communicates
and are controlled through a Java-based program configured
to and interconnects the IoT devices for an efficient Robotic-
IoT transmission and control.
Although the PID controller can be well-designed, it still FIGURE 2. XOY modeling.
has a low robust ability compared with the robust controller
when the system encounters multiple obstacles such as tem-
perature, weather, power surge, and so on in the sensed Self-balancing, Sensed Environment and IoT were briefly
environment. The major drawback of a PID controller is in discussed; the section II makes a quick description of the
the feedback path. Notwithstanding the robustness of the con- related works with respect to self-balancing and robotic
troller, it does not perform efficiently in case of optimal con- control, section III describes the method and evaluation of
trol. Some other drawbacks associated with the controller’s derived mathematical operations and models. Simulations
linear system and derivative part is noise sensitivity. The PID showing the experimental implementations of the method,
is highly noise sensitive that any amount of noise in the and a Java code responsible for Robotic-IoT controls and
operating sensed environment can cause great alteration in communications were analyzed in section IV. The perfor-
the navigation output. However, in this paper, balancing of mance evaluation of the proposed technique compared with
the TWSBR is optimal. We established the dynamic model, a selected algorithm (Adaptive Fussy Control Algorithm) is
through Lagrangian Equation and considering Newton’s analyzed in section V. Conclusions and proposals for future
Second Law in three-dimensional coordinate system. By ana- works are presented VI and VII sections respectively.
lyzing the relationship between forces and motors’ voltage
through a series of mathematical derivations in X-O-Y and II. RELATED WORKS
X-O-Z coordinate system as shown in Figures 1 and 2. It is critical for this project that several related researches
Practical experiments as to show the speed, angular and turn as conducted on undertaking control and design of the robot
controls is performed and the result prove that with the PD-PI before be reviewed because some information regarding the
controller, the robot can navigate a sensed environment, with technologies and methodologies available were provided by
maximum noise resistance, and retained balance for a given many researchers around the world. Therefore, this section
period. presents a condensed summary of related works which link
with the self-balancing two-wheeled robot. An intelligent
system uses an adaptation fuzzy controller using Mamdani
algorithm modified by relation models for a two wheeled
self-balancing robot was developed by Mai et al [5]. The
proposed system that is considered differs, [6] uses the adap-
tive fuzzy controller for stability of the robot and allows a
more control to be implemented. Their proposed adaptation
fuzzy controller was not optimized based on the analysis
of dynamic characteristics as shown in [7] to improve the
performance of the robot. Karam et al in [8] modified the
Integral Sliding mode controller based on Neural Network
and Optimization. The work designed an intelligent non-
FIGURE 1. XOZ modeling. linear Modified Integral Sliding Mode Controller (MISMC)
based on simple Adaline neural network for balancing a two-
The result of the experiment on the avoidance mecha- wheeled self-balancing mobile robot and tried to improve the
nism shows the efficiency of the PD-PI controlled TWSBR performance of their robot in tracking the desired trajectory.
with Kalman Filter Algorithm with minimal Integral Square They used their simulation results to show the efficiency of
Error (ISE) as against the compared Improved Adaptive the proposed controller (MISMC with MCS) in handling the
Fussy Control (AFC) and other existing algorithms. tracking and balancing problems under uncertainties which
This paper is organized as follows: the first part I is the gave a high response speed. The authors in [9] proposed a
introduction and is the section where Robotic Technology, heuristic function algorithm to improve the efficiency of the

82338 VOLUME 7, 2019

C. Iwendi et al.: Robust Navigational Control of a TWSBR in a Sensed Environment

path-planning of two-wheeled self-balancing vehicle and a ADXL335 and angular velocity sensor ISZ-650 were used as
dynamic weighting method for the evaluation function. sensors and Kalman filtering and PID algorithm were used
The path planning algorithm has many advantages accord- as algorithms to implement the balance of the car with two
ing to the author, such as short time-consuming, higher wheels. This research provided some theories of algorithms
search efficiency, better smoothness and shorter path. for the two-wheel self-balanced technique. Some experimen-
A two-wheeled robot is an unstable system such that a control tal Data was provided for the two-wheel self-balanced robot
system is required to stabilize it [10]. Therefore, the authors from a paper named Two Wheel Self-Balanced Mobile Robot
in [11] presented a control design for two-wheeled robot Identification based on Experimental Data. Also, in [19],
(TWR) stabilization using linear quadratic regulator (LQR) two phases of system identification process were applied to
method and states estimation. The authors in [12] shows that implement dynamic equilibrium of the robot. [20] presented
wheeled inverted pendulum robots have advantages over the details on how to overcome the limitations of ‘Weiner-Hopf’
common automated guided vehicles; they have high maneu- Filter in solving problems of statistical nature which seri-
verability on flat surfaces, and they have a self-balancing ously curtailed practical usefulness. The process is named as
system such that height and weight of the payload do not have Kalman Filtering, which is powerful because it can estimate
effect on the base size. the past, present and future states. Several researchers have
The author in [13] investigates the trajectory tracking con- studied the application of Kalman Filtering in solving related
trol problem of wheeled mobile robots by firstly using ana- robotic balancing problems, around the world. The authors
lytic B-spline function commonly used to generate a smooth in [1], and [21] provide research on how to use the
feasible trajectory between the initial and the desired configu- Kalman Filtering to solve the problems of the self-balancing
rations. The reason is to allow the motion path to pass through two-wheeled robot control.
the desired intermediate points and satisfies the kinematic
constraints and curvature restrictions. The author was able III. METHOD AND EVALUATION
A. MODELING
to enhance the robot tracking performance with an assurance
In this section, the mathematical modeling according
that the error convergence, the robust adaptation laws for the
to [22] and [23], is established before analyzing the relation-
FCMAC and compensated controller were derived from the
ship between forces and motors’ voltage through a series of
stability analysis. The research in [14] is slightly different
mathematical derivations in X-O-Y and X-O-Z coordinate
from ours, however, it reveals the development of radio-
system.
frequency identification system and voice broadcast system
on a wheeled mobile robot. The main research purpose was 1) MATHEMATICAL MODELING
to establish a wheeled mobile robot on a working platform The Figure 1 and Figure 2 present the mathematical modeling
of the blind guide system and to assimilate the progressively in different coordinate system.
sophisticated radio frequency identification (RFID) system
and voice broadcast mechanism to read the concealed mes- 2) MATHEMATICAL DERIVATION
sages in the guide bricks or the Braille in public places. In the The main idea of mathematical derivations [24] is based on
system, a CMOS image sensor is used to detect two markings the Lagrangian Equation and Newton’s Second Law in three-
on the guide brick and to achieve the goal design of the guide dimensional coordinate system. Meanwhile, assume t = 0
blind robots. as a precondition. The mathematical derivations are shown as
The goal of [15] was to develop a controlled robot that can following. At first, we used cylindrical coordinate to establish
move with only two wheels. They elaborately discussed the mathematical relationship since wheels are cylindrical.
design and evaluation of a robotic chassis through the appli- (Xm , Ym , Z m ) = (Rβ cos ϕ, Rβ sin ϕ, R) (1)
cation of Lego Mindstorm NXT [16], and to be controlled
by the AVR ATMega16 microcontroller. Their experiment where
 
shows that a robot chassis must address stability and mechan- 1 R
(β, ϕ) = (βl + βr ) , − (βl − βr ) (2)
ical issues. [17] invented a famous balancing robot called 2 W
Segway, which can keep its balance with someone’s standing W W
on its platform. It uses brushless DC electric motors in the (Xl , Yl , Zl ) = (Xm − sin ϕ, Ym + cos ϕ, Zm ) (3)
2 2
wheels powered by lithium-ion batteries with encoders and W W
gyroscopic sensors to check the pitch in order to be upright. (Xr , Yr , Zr ) = (Xm + sin ϕ, Ym − cos ϕ, Zm ) (4)
2 2
JOE is a self-balancing robot based on the inverted pendulum (Xb , Yb , Zb ) = (Xm + L sin θ cos ϕ, Ym + L sin θ sin ϕ, Zm +
which was designed by the Industrial Electronics Laboratory L cos θ) The translational kinetic energy T1 , rotational energy
at the Swiss Federal Institute of Technology. It was con- T2 , and the potential energy U are derived as:
trolled by a Digital Signal Processor (DSP) IC and used the
feedback of a controller. When running, the maximum of 1 ·2 · ·
1 · · ·
T1 = (Xl + Yl2 + Zl2 ) + m(Xr2 , Yr2 , Zr2 )
speeds reaches 1.5m/s. The paper called ‘Two-wheel self- 2 2
· · ·
balanced car based on Kalman filtering and PID algorithm’ 1 2 2 2
+ M (Xb + Yb + Zb ) (5)
presented by [18] and [19] shows that a low-cost acceleration 2

VOLUME 7, 2019 82339

 C. Iwendi et al.: Robust Navigational Control of a TWSBR in a Sensed Environment

1 · 2 1 · 2 1 · 2 1 · 2
T2 = Jw βl + Jw βr + Jϕ θ + Jϕ θ
2 2 2 2
· ·
1 · 1 ·
+ Jm n2 (α) l − θ 2 + Jm n2 (α) r − θ 2 (6)
2 2
U = mgZl + mgZr + mgZb (7)
Considering the combination of kinetic and potential energy,
L = T − V is considered the Lagrangian Equation.
The equation is given through the following the Lagrangian FIGURE 3. (a)Describes the stable state for the two-wheeled
equation. self-balancing robot. (b) Indicates the robot leans to the right. (c)
Demonstrates the robot leans to the left.
L = T1 + T2 − U (8)
According to the Newton’s Second Law, F = ma. Combining nKt Kb
the Lagrangian equation and Newton’s Second Law, the equa- µ= + fm (20)
Rm
tion of motion is derived as: λ · ·
Fβ = (Vl + Vr ) − (µ + fm ) β + µ θ (21)
d ∂L ∂L
   
2
− = Fx (9) · ·
dt ∂x ∂x Fθ = −λ(Vl + Vr ) + 2µ(β − θ ) (22)
R W ·
The motion equations about this modeling can then be Fβ = − λ(Vl − Vr ) − (µ + fw ) ϕ (23)
deduced as: W R
d ∂L ∂L
   
− = Fβ (10) B. ROBOTIC CONTROLS
dt ∂β ∂β 1) CONTROL ALGORITHM
d ∂L ∂L
   
− = Fθ (11) The control algorithm of the self-balancing –two-wheeled
dt ∂θ ∂θ robot can be based on PID control according to [25]–[27].
d ∂L ∂L However, some problems are shown in these papers. There-
   
− = Fϕ (12) fore, the control algorithms of this project are improved
dt ∂ϕ ∂ϕ

1 R
 to be PD-PI control. Adopting adaptive and internal model
(Fβ , Fθ , Fϕ ) = (Fl +Fr ), Fθ ,− (Fl −Fr ) control (IMC) system, [28] focused their study on practically
2 W
implementing of remote control. Hence, the adaptive auto-
(13)
· · · adjustable controller is dependent on the pole’s placement
Fl = nKt Il + fm (θ − β ) − fw βl (14) control technique and the hierarchical identification strategy.
l [29] developed and implemented a neural network inte-
· · ·
Fr = nKt Il + fm (θ − β ) − fw βr (15) grated modfied DAYANI technique for path and navigational
r control of a TWSBR in a cluttered environment. The authors
· ·
Fθ = (nKt Il + nKt Ir + fm (θ − β ) instigated a five-layered back-propagation neural network to
l find out the intensity of various weight factors considering
· · seven navigational parameters as obtained from the modfied
+ fm (θ − β )) (16)
r DAYANI method.
where It/r is the current of left or right DC motor. Hence the
voltage of DC motor is based on PWM, which means current 2) UPRIGHT CONTROL (PD CONTROL)
cannot be used to deduce the voltage. Therefore, the DC Upright Control of self-balancing two-wheeled robot makes
motor equation shown as (17) is evaluated to build up the two wheels balancing on the ground with the body of the robot
relationship between current and voltage. using negative feedback. Having two wheels, the body of the
robot does nothing but fall down in a direction that the wheels
· · ·
L m Il/r = Vl/r + Kb θ − αl/r −Rm Il/r (17) are running.

where is Lm the motor inductance, assume that it is approxi- 3) PITCH ANGLE CONTROL
mately as zero. The current It/r is given as Using the mathematical modeling to analyze the pitch angle
· · control as described in the Figure 4 below:
Vl/r +Kb (θ ) − αl/r
It/r = Rm (18) a(t) : Acceleration of Running Wheel
x(t) : Angular Velocity Caused by External Force
In conclusion, the relationship between forces and DC motor Deriving from the modeling, a motion equation of the robot
voltage can be calculated as: is written as:
nKt
λ= (19) L d dtθ2(t) = g sin θ (t) − a(t) cos θ (t) + Lx(t)
2
(24)
Rm

82340 VOLUME 7, 2019

C. Iwendi et al.: Robust Navigational Control of a TWSBR in a Sensed Environment

C. SPEED CONTROL (PI CONTROL)

Because of the angle bias on angular control, the robot
is accelerating in the direction of pitch to keep it balanc-
ing on horizon, which leads to the problem of speed con-
trol. The speed control is complex because it cannot adjust
directly the revolving speed of DC motors to implement the
balance [30], [31].

1) SPEED CONTROL
FIGURE 4. Pitch angle control modellin.
To give a suitable velocity from motors, a method which is
the speed control is drawn as Figure 5 below.
When the Angle θ is approaching to 0, the motion equation is
rewritten as:

L d dtθ2(t) = gθ (t) − a(t) + Lx(t)

2
(25)

Considering the static condition a(t) = 0, simplify the fore-

going second-order differential equation:

L d dtθ(t)
2
2 = gθ (t) + Lx(t) (26)
FIGURE 5. Block diagram of speed control.
Converting the differential equation from time-domain to
s-domain through Laplace Transform (LT), and then calculate The biases of speed control are directly adding into the
its transform function; controlled real velocity variables of DC motors through PI
2(s) 1 controller.
H (S) = = 2 g (27) In order to test the PI Control, it will combine with PD
X (s) s − L control as the control system to run the robot on a straight
Assume s2− Lg = 0 the poles line by (33).
q of the transform function H (S) X
are given as Sp = ± Lg . Because of the positive pole, V = Kp × θ + Kd × θ − Kp × e(k) + Ki e(k) (30)
the system of this robot is unstable.
Rewriting the transform function, we obtained; 2) TURN CONTROL (P CONTROL)
When the upright and speed controls are done, the robot can
2(s) 1 keep balance with direction. However, the robot needs to
H (s) = = k2 k1 −g
(28)
X (S) s2 + Ls+ L implement the object avoidance mechanism so that it needs
turn control to change different directions.
The poles of this system are given as: The turn control focuses on the different voltages in the
q left and right motor because different voltages will generate
−k2 ± k22 − 4L(k1 − g) the speed differences that can used to control the robot’s turn.
Sp = (29)
2L P Controller is used in order to implement these differences
The system is stable in the condition that all poles are neg- in voltage. The block diagram for the robot’s turn control is
ative. Therefore, the roots in (29) should satisfy k1 > g, shown in Figure 6.
k2 > 0.
D. KALMAN FILTER ALGORITHM
In the angle control, k1 and k2 are respectively referred to
as the proportional and differential parameters. Moreover, the The gyroscope and accelerometer of MPU6050 provide
differential parameter is equivalent to damping force, which measures of instantaneous angular and accelerated velocity
can restrain effectively the vibration of body of this robot. change but they also generate drifts when MPU6050 is run-
ning. Besides, the output of signals from MPU6050 is often
4) ANGULAR VELOCITY CONTROL
corrupted with noise. It is a fact that the drifts and noise have
an influence on control of the robot. Therefore, the robot
Angular velocity is measured by gyro of MPU6050. In addi-
needs a signal-level fusion technique to filter drifts and noise
tion, since motion of the robot has no influence on the angular
in signals. In the robot, the Kalman Filtering [32], [33] as the
velocity, the noise in the signal of angular velocity is too
filter algorithm is implemented to overcome these problems.
tiny to ignore at the time of designing circuits of the robot.
Kalman Filtering is described by a Linear Stochastic Dif-
Furthermore, the integral of angular velocity signal is to
ferential Equation like the (31).
calculate the pitch angle of the robot, which can make the
angle signal more stable. X (k) = AX (k − 1) + BU (k) + W (k) (31)

VOLUME 7, 2019 82341

 C. Iwendi et al.: Robust Navigational Control of a TWSBR in a Sensed Environment

TABLE 1. Physical parameters of mathematical model.

FIGURE 6. Block diagram of turn control.

The system measurement is added by:

Z (k) = HX (k) + V (k) (32)
Assume that the robot’s acceleration a(t) = 0, and then define
→
the vector of the state x and covariance Pk as follow
 
→ position
x
velocity
X X !
pp pv
Pk = X X (33)
vp vv
Suppose that xEk−1 and xEk as the last and current state,
the updating state equation can be given according to equa-
tions of Kalman Filtering,
pk = pk−1 + σ tvk−1
vk = vk−1 (34)
Rewrite the vector of the state xE
1 σt
 
→
xk = xEk−1 = Fk xEk−1 (35)
0 1
By establishing the covariance matrix to update dependency
between different states we derived:
X X
Cov(x) = Cov(Ax) = A AT (36)
Combining the vector matrix with covariance matrix
xEk = Fk xEk−1
Pk = Fk Pk−1 FkT (37)
Assuming the robot has acceleration a(t), it gives
1 p
Pk = Pk−1 + σ tvk−1 + a(t)σ t 2 a2 + b2
2
vk = vk−1 + a(t)σ t (38) In Figure 7, the TWSBR with ultrasonic equipment is
Finally, Use a matrix to stand for updating (39) transmitting blue ultrasonic waves to detect roadblocks. And
the red waves are reflected by a roadblock. The robot is
σt
 2
→ turning to avoid the objective in this simulation of MATLAB
xEk = Fk xEk−1 + 2 a(t) = Fk xEk−1 + Bk µk (39)
σt Robot Toolbox.
A Genetic algorithm (GA) based on Dynamic path plan-
E. AVOIDANCE MECHANISM ning algorithm (DPPA) proposed in [32] is analyzed and
The avoidance mechanism according to [34], [a] is implemented in the path planning and obstacle avoidance of
implemented for this robot due to its dynamic working the TWSBR. The algorithm confirms the flexibility of the
environment. The introduction of Ultrasonography makes robot in reaching its targeted destination from any arbitrary
this mechanism possible in this robot because of ‘Ultrasonic starting position. The implementation shows an improvement
Waves’ character that an ultrasonic wave is transmitted before in the TWSBR feasible path generation, acute and dynamic
meeting an objective, and then it will be reflected in the same obstacles avoidance, and shortest path selection from starting
path. point to the desired destination.

82342 VOLUME 7, 2019

C. Iwendi et al.: Robust Navigational Control of a TWSBR in a Sensed Environment

FIGURE 7. Simulation of avoiding mechanism.

FIGURE 9. Top-layer design.

FIGURE 8. Avoidance system.

F. WORKING PRINCIPLE
Based on this character, a system for the avoidance mecha-
nism is designed like Figure 8, which includes transmitting
probe, receiving probe and amplifier.
In order to decrease the cost and size of this robot, a simple
ultrasonic product having two probes is used as the avoid-
ing sensor. Meanwhile, ADC of STM32 microcontroller is
adopted to confirm the Transit Time.

IV. SYSTEM DESIGN AND PERFORMANC

A. MECHANICAL STRUCTURE
Figure 9 show the mechanical structure of the top layer
designed TWSBR. The sub-layer of the circuit is to place DC
motors as the drive module for this robot. In order to decrease
the size and space of this robot, a plastic baseboard and two
FIGURE 10. Software main flow char.
metallic racks in are employed to immobilize the left and right
DC motor as two-wheeled system. When they are combined,
the sub-layer design will be done. The aim of the middle-layer The most important task in the system is the settings of
design is to lay the 12V battery over the plastic baseboard as a Baud Rate which can be define as 12800 Baud per second
place of the power supply. Because the robot has a top-layer in this situation. Additionally, the length of frame which is
to put the main PCB, copper cylinders are used to support a constant while the microcontroller is communicating with
the plastic baseboard of top-layer. The goal of the top-layer the application. Furthermore, the delay time in the commu-
design is to install the main PCB with all sub-modules and nication between the two frames need to be less than 1ms.
circuits of this robot. In addition, a transparent plastic board Otherwise, forward frame will be missing.
over main PCB plays a defending role for this robot. Finally, there is no limit for refreshing the User Inter-
face (UI) on the application, which means that the refreshing
B. SYSTEM PERFORMANCE rate of UI depends directly on the Graphics Processing
Analyzing the system of the TWSBR, the main flowchart Unit (GPU) and Central Processing Unit (CPU) of the com-
in Figure 10 is established in order to present the main idea for puter and the reliability of the communication chain. While
programming the robot, and it can be provided by visualized the PC application is running to communicate between the
theoretical basis for what main functions should be included robot and the PC, the window of this application will show
in the software of this robot. the communicating signals like Figure 11.

VOLUME 7, 2019 82343

 C. Iwendi et al.: Robust Navigational Control of a TWSBR in a Sensed Environment

FIGURE 11. Running PC application.

In Figure 10, the numbers with different colors presents

peaks of signals. In order to observe the communicating data, FIGURE 12. PD control SIMULINK result.

they are amplified to be 100 times as much as initial values.

C. PERFORMANCE OF BALANCING MEDIAN

The balancing median is significant because it has an influ-
ence on the selection of parameters of PI-PD-P controller.
To obtain the median, the self-balancing two-wheeled robot
need be laid on the horizontal ground and then given an
initializing speed for recording the angle which can make
this robot balancing. While the robot was debugging on
ground, the balancing median was showing on the window
of the application, which was around 0 degree. Therefore,
the median is defined as 0, hence Bias = Angle-0.

D. PERFORMANCE OF POLARITY OF
FIGURE 13. PD-PI control SIMULINK result.
PROPORTIONAL PARAMETER
To confirm the polarity of proportional parameter, the proba- E. CONFIRMING AMPLITUDE OF
ble theoretical range of proportional parameter is evaluated PROPORTIONAL PARAMETER
at first. In this project, standing for the 100 percent Duty Based on the theoretical range of proportional parameter
Ratio, the setting of PWM’s parameter is 7200. Therefore, the between 0 and 720, different proportional parameters can
maximum theoretical proportional parameter can be assumed be tested into codes to discover the suitable parameter, with
as 720, which leads to the full turn of the robot at positive and increasing of proportional parameters until the robot vibrates
negative 10 degree. According to the physical situation, it is dramatically in low frequency.
too small to control the robot accurately. However, the theo- Some proportional parameters as following are checked
retical absolute range of proportional parameter is seemingly into codes:
confirmed between 0 and 720. Based on the range of propor- 1) When Kp = 200, the robot has the state trend of upright
tional parameter, the polarity of proportional parameter can with slow response.
be confirmed via assuming positive or negative proportional 2) When Kp = 350, the robot upright with faster response
parameter into codes to check out the operating situation of than first situation, the state of response is so slow that
the robot. On the one hand, choose the negative proportional the robot cannot reach a balancing state at a moment
parameter into codes of the project before turning on the not keep balanced on a whole time.
robot, and then operating circumstance of the robot is that 3) When Kp = 500, the response is significantly fast
it is falling with the acceleration of DC motors, which is with dramatically vibration in low frequency, which
an undesirable situation because it illustrates the result of means the proportional parameter is suitable and the
positive feedback. On the other hand, refresh the proportional D controller need be added into the control system to
parameter to be positive, which lead to the upright state of restrain the vibration in low frequency.
the robot. The upright circumstance means there is a negative
feedback between DC motor and the body of the robot, which F. CONFIRMING POLARITY OF
is an expecting result. According two different situations, DIFFERENTIAL PARAMETER
the polarity of the proportional parameter can be confirmed The initialization of data from output of MPU6050 an assem-
as positive one. blage whose maximum is under 10000 and the 100 percent

82344 VOLUME 7, 2019

C. Iwendi et al.: Robust Navigational Control of a TWSBR in a Sensed Environment

FIGURE 15. (a) PD-PI controlled TWSBR robot w/ Kalman filter algorithm
against the adaptive fussy control – (Control Signal). (b) PD-PI controlled
TWSBR robot w/ Kalman filter algorithm against the adaptive fussy
control – (Error Signal).

are debugged into codes. With adding of the differential

parameter, the robot is vibrating at high frequency, which can
decrease the influence of vibration of low frequency to make
FIGURE 14. Response of the TWSBR using the PD-PI controller (a. for the the robot balancing in static.
pitch angle, b. for the angular velocity, and c. for velocity). Some differential parameters as following are checked into
codes:
Duty Ratio 7200, the theoretical absolute range of the differ- 1) When Kd = 0.5, the vibration of low frequency is
ential parameter can be computed from 0 to 2. canceling.
Compared with the range of the differential parameter, 2) When Kd = 1, the vibration of low frequency is gone,
the positive and negative parameters are selecting into codes. and the robot is balancing on the ground.
Foremost, the differential parameter is supposed to a negative 3) When Kd = 1.7, the robot is vibrating on high
one, which causes the rotating body of robot with the opposite frequency.
turning DC motors, which leads to the unbalancing robot.
Secondly, turn the differential parameter to the positive one, V. EXPERIMENTS AND RESULTS
which implement the effect of following that the body of the The theory of the PD Control is verified by SIMULINK
robot is rotating with synchronous DC motors. This result of MATLAB to make sure it is correct. The result shown
means that the close-loop of angular velocity is implementing in Figure 12 indicates that the robot can be balanced with two
into the control system and is conclude that the polarity of the wheels as time goes because the dip angles are going to be 0◦ .
differential parameter is positive. When a disturbance is simulated for the robot on
SIMULINK’s window, the line chart in Figure 6 presents that
G. CONFIRMING AMPLITUDE OF the robot will expedite in the direction of the disturbance
DIFFERENTIAL PARAMETER to reach the saturation. To confirm the working principle is
According to the theoretical absolute range of the differen- correct, a simulation of Robotic toolbox is done as shown
tial parameter from 0 to 2, some parameters in this range in Figure 13.

VOLUME 7, 2019 82345

 C. Iwendi et al.: Robust Navigational Control of a TWSBR in a Sensed Environment

angle, angular velocity, and the velocity of the tested TWSBR

respectively. Analyzing the simulation, it is observed that
both the pitch angle, the angular velocity, and the velocity of
the TWSBR can be controlled as close as to 0 [rad], the robot
position attains equilibrium point and remains constant
after 4.3 [s] for the pitch angle, at 1.5 [s] the angular velocity
attained equilibrium and remain constant, while the velocity
reached equilibrium at 2.4 [s] and remained stable.
The simulation results in Figures 15a and 15b show that the
TWSBR stabilized at the set 0o even under pressure and exter-
nal force disturbance and follows the intended input signal
very fast with zero error. The experiments show error signal
indicating the performance of the TWSBR in the sensed envi-
ronment. The avoidance mechanism of obstacles, uncertain-
ties, and noise signal control. The proposed PD-PI controller
based on Kalman filter algorithm is compared against the
improved Adaptive fuzzy controller (AFC) in [5]. The results
show that the performance of PD-PI controlled TWSB robot
with Kalman Filter Algorithm is more efficient with minimal
Integral Square Error (ISE) as against the compared improved
AFC. It is also observed that the TWSB robot with PD-PI
control follow the planned path in the sensed environment
with an optimal obstacle avoidance, and more accurately
with minimal overshoot and zero steady state error. A block
diagram explaining for the PD-PI controller is represented
in Figure 16.

VII. CONCLUSION
The research experimented a PD-PI navigational control
for a TWSBR. Kalman filter algorithm is considered to
determine the stability of the robot in a sensed envi-
ronment. This research is successful since the objects of
self-balancing two-wheeled robots were implemented, and
a maximum obstacle avoidance achieved. The robot can
make itself balancing with wheels through PD-PI control and
turned by P control. In addition, the Kalman Filtering is used
to eliminate the drift of MPU6050’s gyro allowing for an
accurate estimate of the tilt angle. Avoidance mechanism is
implemented by using the ultrasonic waves to detect obstacles
around the robot. Communication between the robot and the
IoT devices in the environment is established through a JAVA
code embedded in the controller and manipulated through
Bluetooth connection.
Finally, the proposed PD-PI control technique introduces
FIGURE 16. PD-PI controller block diagram.
theoretically the relationship of different controls including
upright control (PD Control), speed control (PI Control) and
VI. PERFORMANCE EVALUATION turn control (P Control) to make the robot self-balanced with
Using MATLAB simulations, we conduct several experi- two wheels. In the built Kalman algorithm, parameters of
ments to evaluate the performance of the proposed PD-PI PD-PI Control need to pay attention to the polarity and ampli-
controller. The point of balance is set at 0 [rad], as we simulate tude which have an influence on the balance of the robot.
the mathematical model of two-wheeled self-balancing robot However, the maximum proportional and differential param-
using the Kalman Filter Algorithm. The response curves eters are 500 and 1.7 respectively. Although the performance
of the controller in the balance point applied is shown in of the robot in maximum is vibrating slightly. Therefore,
Figures 14 (a., b., and c.). These simulation results with to decrease the slight vibration, the maximum proportional
PD-PI control as shown in Figure 14 represents the pitch and differential parameters multiply an influencing constant

82346 VOLUME 7, 2019

C. Iwendi et al.: Robust Navigational Control of a TWSBR in a Sensed Environment

whose value is 0.6. In addition, the suitable proportional [14] T.-F. Wu, P.-S. Tsai, N.-T. Hu, and J.-Y. Chen, ‘‘Intelligent wheeled mobile
and differential parameters become 300 and 1 respectively robots for blind navigation application,’’ Eng. Computations, vol. 34, no. 2,
pp. 214–238, 2017. doi: 10.1108/EC-08-2015-0256.
without any vibration in the robot. These parameters are [15] P.-S. Tsai, L.-S. Wang, and F.-R. Chang, ‘‘Modeling and hierarchical track-
programmed into the codes and the robot keeps upright during ing control of tri-wheeled mobile robots,’’ IEEE Trans. Robot., vol. 22,
navigation. no. 5, pp. 1055–1062, Oct. 2006.
[16] Q. Yong, L. Yanlong, Z. Xizhe, and L. Ji, ‘‘Balance control of two-wheeled
self-balancing mobile robot based on TS fuzzy model,’’ in Proc. 6th Int.
VIII. FUTURE RESEARCH Forum Strategic Technol., vol. 1, Aug. 2011, pp. 406–409.
[17] A. Salerno and J. Angeles, ‘‘A new family of two-wheeled mobile
More research should be conducted to exploit more com- robots: Modeling and controllability,’’ IEEE Trans. Robot., vol. 23, no. 1,
plex algorithm for avoidance mechanism based on ultra- pp. 169–173, Feb. 2007.
sonic waves and extend the connectivity for Bluetooth to [18] K. Liu, M. Bai, and Y. Ni, ‘‘Two-wheel self-balanced car based on Kalman
filtering and PID algorithm,’’ in Proc. IEEE 18th Int. Conf. Ind. Eng. Eng.
Wireless-Fidelity. Manage., vol. 1, Sep. 2011, pp. 281–285.
The parameters of PD-PI control is not perfect because the [19] A. Srilalitha, C. Paramasivan, and P. Dinakaran, ‘‘Design and evaluation
robot shakes a little when the velocity is increased. Therefore, of two wheeled balancing robot using lego mindstorms nxt,’’ Int. J. Mul-
tidisciplinary Res. Mod. Educ., vol. 1, no. 1, 2015. [Online]. Available:
future research is recommended for improving the control http://rdmodernresearch.org/wp-content/uploads/2015/08/3.pdf
algorithm to implement more self-balancing with two wheels [20] Y. Hitaka, T. Yoshitake, and M. Yokomichi, ‘‘Obstacle avoidance of snake
as an inverted pendulum robot. robot by switching control constraint,’’ Artif. Life Robot., vol. 17, no. 2,
pp. 180–185, 2012.
[21] J. H. Anajemba, Y. Tang, J. A. Ansere, and C. Iwendi, ‘‘Perfor-
REFERENCES mance analysis of D2D energy efficient IoT networks with relay-
[1] H. S. Zad, A. Ulasyar, A. Zohaib, and S. S. Hussain, ‘‘Optimal controller assisted underlaying technique,’’ in Proc. 44th Annu. Conf. IEEE Ind.
design for self-balancing two-wheeled robot system,’’ in Proc. Int. Conf. Electron. Soc., Washington, DC, USA, Oct. 2018, pp. 3864–3869.
Frontiers Inf. Technol. (FIT), Islamabad, Pakistan, Dec. 2016, pp. 11–16. doi: 10.1109/IECON.2018.8591373.
[2] R. P. M. Chan, K. A. Stol, and C. R. Halkyard, ‘‘Review of modelling [22] B. Mahler and J. Haase, ‘‘Mathematical model and control strategy of a
and control of two-wheeled robots,’’ Annu. Rev. Control, vol. 37, no. 1, two-wheeled self-balancing robot,’’ in Proc. 39th Annu. Conf. Ind. Elec-
pp. 89–103, Apr. 2013. tron. Soc., Nov. 2013, pp. 4198–4203.
[3] N. Uddin, T. A. Nugroho, and W. A. Pramudito, ‘‘Stabilizing Two- [23] X. Bin, G. Lei, W. Shimin, S. Yuan, and Z. Ying, ‘‘Dynamics modeling
wheeled robot using linear quadratic regulator and states estimation,’’ and system parameter identification experiment of a kind of two-wheeled
in Proc. 2nd Int. Conf. Inf. Technol. Inf. Syst. Elect. Eng. (ICITISEE), robot,’’ in Proc. IEEE Int. Conf. Inf. Automat., Aug. 2015, pp. 404–408.
Yogyakarta, Indonesia, Nov. 2017, pp. 229–234. doi: 10.1109/ICI- [24] R. Hongge, W. Zhilong, L. Fujin, and H. Meijie, ‘‘The balance control
TISEE.2017.8285501. of two-wheeled robot based on bionic learning algorithm,’’ in Proc. 26th
Chin. Control Decis. Conf., May/Jun. 2014, pp. 4166–4170.
[4] O. Vermesan, P. Friess, P. Guillemin, S. Gusmeroli, H. Sundmaeker,
[25] J. J. R. Pasaye, J. A. B. Valencia, and F. J. Pérez, ‘‘Tilt measurement based
A. Bassi, I. S. Jubert, M. Mazura, M. Harrison, M. Eisenhauer, and
on an accelerometer, a gyro and a Kalman filter to control a self-balancing
P. Doody, ‘‘Internet of Things strategic research roadmap,’’ in Inter-
vehicle,’’ in Proc. IEEE Int. Autumn Meeting Power Electron. Comput.
net of Things: Global Technological and Societal Trends, vol. 1,
(ROPEC), Nov. 2013, pp. 1–5.
O. Vermesan, P. Friess, P. Guillemin, S. Gusmeroli, H. Sundmaeker, and
[26] A. M. Mohtasib and M. H. Shawar, ‘‘Self-balancing two-wheel electric
A. Bassi, Eds. Gistrup, Denmark: River Publishers, 2011, pp. 9–52.
vehicle (STEVE),’’ in Proc. 9th Int. Symp. Mechatronics Appl. (ISMA),
[5] T. A. Mai, D. N. Anisimov, T. S. Dang, and V. N. Dinh, ‘‘Development of
Apr. 2013, pp. 1–8.
a microcontroller-based adaptive fuzzy controller for a two-wheeled self-
[27] C. Sun, T. Lu, and K. Yuan, ‘‘Balance control of two-wheeled self-
balancing robot,’’ Microsyst. Technol., vol. 24, no. 9, pp. 3677–3687, 2018.
balancing robot based on linear quadratic regulator and neural network,’’
doi: 10.1007/s00542-018-3825-2.
in Proc. 4th Int. Conf. Intell. Control Inf. Process. (ICICIP), Jun. 2013,
[6] D. N. Anisimov and A. M. Tkhe, ‘‘Dynamic properties of the fuzzy control pp. 862–867.
systems based on the relational models,’’ăMekhatronika, Avtomatizatsiya, [28] W. An and Y. Li, ‘‘Simulation and control of a two-wheeled self-balancing
Upravlenie, vol. 18, no. 5, pp. 298–307, 2017. doi: 10.17587/mau.18.298- robot,’’ in Proc. IEEE Int. Conf. Robot. Biomimetics (ROBIO), Shenzhen,
307. China, Dec. 2013, pp. 456–461.
[7] A. M. Almeshal, K. M. Goher, and M. O. Tokhi, ‘‘Dynamic modelling [29] A. Chhotray and D. R. Parhi, ‘‘Navigational control analysis of
and stabilization of a new configuration of two-wheeled machines,’’ Robot. two-wheeled self-balancing robot in an unknown terrain using back-
Auton. Syst., vol. 61, pp. 443–472, May 2013. propagation neural network integrated modified DAYANI approach,’’
[8] E. Karam and N. Mjeed, ‘‘Modified integral sliding mode controller Robotica, pp. 1–17, 2019. doi: 10.1017/S0263574718001558.
design based neural network and optimization algorithms for two wheeled [30] F. Zhang, S. Li, S. Yuan, E. Sun, and L. Zhao, ‘‘Algorithms analysis of
self balancing robot,’’ Int. J. Mod. Educ. Comput. Sci., vol. 10, no. 8, mobile robot SLAM based on Kalman and particle filter,’’ in Proc. 9th Int.
pp. 11–21, 2018. doi: 10.5815/ijmecs.2018.08.02. Conf. Modelling, Identificat. Control (ICMIC), Jul. 2017, pp. 1050–1055.
[9] Q. Wang, W. Huang, B. Liu, and Y. Zhang, ‘‘An improved A∗ algorithm [31] X. Ruan and W. Li, ‘‘Ultrasonic sensor based two-wheeled self-balancing
for path-planning of two-wheeled self-balancing vehicle,’’ in Proc. 13th robot obstacle avoidance control system,’’ in Proc. IEEE Int. Conf. Mecha-
IEEE Conf. Ind. Electron. Appl. (ICIEA), Wuhan, China, May/Jun. 2018, tronics Automat., Aug. 2014, pp. 896–900.
pp. 841–846. [32] Y. Olfa, B. Saida, and A. Kamel, ‘‘Implementation of remote control for
[10] A. Weiss, E. Fadida, and U. B. Hanan, ‘‘Optimizing step climbing by two time delay systems based on adaptive and internal model approaches,’’
connected wheeled inverted pendulum robots,’’ Procedia Manuf., vol. 21, in Proc. 18th Int. Conf. Sci. Techn. Autom. Control Comput. Eng. (STA),
pp. 236–242, Jan. 2018. Monastir, Tunisia, Dec. 2017, pp. 80–85.
[11] M. Boukens, A. Boukabou, and M. Chadli, ‘‘Robust adaptive neural [33] S. C. Yun, S. Parasuraman, and V. Ganapathy, ‘‘Dynamic path planning
network-based trajectory tracking control approach for nonholonomic algorithm in mobile robot navigation,’’ in Proc. IEEE Symp. Ind. Electron.
electrically driven mobile robots,’’ Robot. Auton. Syst., vol. 92, pp. 30–40, Appl., Langkawi, Malaysia, Sep. 2011, pp. 364–369.
Jun. 2017. [34] Z. Yang, Y. Ding, H. Ding, and Y. Su, ‘‘Improvement of bluetooth tech-
[12] S. Blažič, ‘‘Two approaches for nonlinear control of wheeled mobile nology scheduling algorithm for wireless personal area network,’’ in Proc.
robots,’’ in Proc. 13th IEEE Int. Conf. Control Automat. (ICCA), Jul. 2017, IEEE 2nd Inf. Technol. Netw. Electron. Automat. Control Conf. (ITNEC),
pp. 946–951. Chengdu, China, Dec. 2017, pp. 408–411.
[13] T.-F. Wu, ‘‘Tracking control of wheeled mobile robots using fuzzy CMAC [35] W. H. Hayt, J. E. Kemmerly, and S. M. Durbin, Engineering Cir-
neural networks,’’ J. Internet Technol., vol. 19, no. 6, pp. 1853–1869, cuit Analysis, 8th ed. New York, NY, USA: McGraw-Hill, 2012,
Nov. 2018. pp. 321–371.

VOLUME 7, 2019 82347

 C. Iwendi et al.: Robust Navigational Control of a TWSBR in a Sensed Environment

CELESTINE IWENDI (M’01–SM’17) received AHMED S. ALFAKEEH received the bachelor’s

the second master’s degree in communication degree in computer science and education, in 2005,
hardware and microsystem engineering from from King Abdulaziz University, and the M.Sc.
Uppsala University, Sweden, 2008, ranked under degree in IT from De Montfort University, in
100 in the world University ranking and the 2010, the Ph.D. degree from Software Technology
Ph.D. degree in electronics from the University Research Laboratory, a research institute estab-
of Aberdeen, U.K., in 2013. He is an Associate lished within De Montfort University, Leicester,
Professor from Sweden. He is highly motivated UK, in 2015. He is an Assistant Professor with
researcher with a wireless sensor network security Information System Department, Faculty of Com-
book, and over 100 publications. He is currently puting and Information Technology, King Abdu-
a Senior Lecturer with the BCC of Central South University of Forestry laziz University, Jeddah, Saudi Arabia. He is currently a Vice-Dean with the
and Technology, Changsha, China, and has strong teaching emphasis on Research and Consultation Institute for Development. His research projects
communication, hands-on experience, willing-to-learn, and 18 years tech- involve feature interactions detection and resolution in context-aware sys-
nical expertise and currently teaches engineering team project, circuit the- tems, smart homes systems, software engineering, the IoT, and data science.
ory, data networks and distributed systems, and control systems. He has
developed operational, maintenance, and testing procedures for electronic
products, components, equipment, and systems; provided technical support
and instruction to staff and customers.
He is a wireless sensor network Chief Evangelist, researcher, and designer. ZHIYONG ZHANG (M’17) received the B.Sc.
A Board Member of IEEE Sweden Section (2017 to date) and a Fellow of degree in electronic engineering from Bangor Uni-
The Higher Education Academy, U.K. to add to his teaching and professional versity and Central South University of Forestry
experiences. He is the Editor, International Journal of Engineering and and Technology. He is an enthusiastic learner and
Allied Disciplines 2015, Newsletter Editor, IEEE Sweden section-2016– believer of the benefits of WSNs and control sys-
2018, Editor-in-Chief, Wireless Sensor Network Magazine, 2009, Committee tems. He has presented and published papers in
Member, International Advisory Panel, International Conference on Marine, IEEE conferences. He has a project of a drawing
Ocean, and Environmental Sciences and Technologies (MAROCENET) robot which was accepted by Hunan (China) Gov-
2014–2016, Editor-in-Chief, Journal of Wireless Sensor Network, 2009, ernment and got the scholarship to finish it.
Advisory Board, International Journal of Innovative Computer Science and
Engineering (IJICSE) 2013. He is the Co-Chair of the special session on
‘‘Wireless Sensor Networks: Hardware/Software Design aspects for Indus-
try’’ at the Prestigious International Conference of Industrial Technology
ICIT. His research focuses on wireless sensor networks, Security of Things
(SoT), communication controls, the Internet of Things (IoT), electromag-
ALI KASHIF BASHIR (SM’16) received the B.S.
netic machines, 5G networks and low power communication protocols.
degree from the University of Management and
Technology, Pakistan, the M.S. degree from Ajou
University, South Korea, and the Ph.D. degree
MOHAMMED A. ALQARNI received the bache- in computer science and engineering from Korea
lor’s degree in computer science from King Khalid University, South Korea. He is a Senior Lec-
University, Saudi Arabia, in 2008, and the M.Sc. turer with the School of Computing, Mathematics,
degree in computational sciences from Laurentian and Digital Technology, Manchester Metropolitan
University, Sudbury, Canada, in 2012, the Ph.D. University, U.K. He is a Distinguished Speaker
degree in computer science from McMaster Uni- of ACM. His past assignments include: Associate
versity, Hamilton, Canada, in 2016. He is cur- Professor of Information and Communication Technologies, Faculty of Sci-
rently an Assistant Professor and the Dean with ence and Technology, University of the Faroe Islands, Denmark; Osaka
the College of Computer Science and Engineering, University, Japan; Nara National College of Technology, Japan; the National
the University of Jeddah, Saudi Arabia. He enjoys Fusion Research Institute, South Korea; Southern Power Company Ltd.,
research in a wide variety of topics including the IoT. South Korea, and the Seoul Metropolitan Government, South Korea.
He is author of over 80 peer-reviewed articles. He is supervising/co-
supervising several graduate (M.S. and Ph.D.) students. His research interests
JOSEPH HENRY ANAJEMBA received the bach- include the Internet of Things, wireless networks, distributed systems, net-
elor’s degree in computer science from the Fed- work/cyber security, cloud/network function virtualization, etc. He is serving
eral Polytechnic Oko, Nigeria, the M.Sc. degree in as the Editor-in-Chief of the IEEE FUTURE DIRECTIONS NEWSLETTER.
information communication technology from the He is the Editor of several journals and also has served/serving as Guest
National Open University of Nigeria, in 2016. He Editor on several special issues in journals of IEEE, Elsevier, and Springer.
is currently pursuing the Ph.D. degree in infor- He has served as the Chair (program, publicity, and track) and the Chair
mation and communication engineering with the on several conferences and workshops. He has delivered several invited and
College of Internet of Things, Department of Com- keynote talks, and reviewed the technology leading articles for journals like
munication engineering, Hohai University, China. IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, the IEEE Communication
He has credited several articles and conference Magazine, the IEEE COMMUNICATION LETTERS, IEEE Internet of Things, and
papers. His current research interests include cellular wireless communica- the IEICE journals, and conferences, such as the IEEE Infocom, the IEEE
tions, antenna and V2V technology, and 5G cellular networks and security, ICC, the IEEE Globecom, and the IEEE Cloud of Things.
and several other IoT related areas.

82348 VOLUME 7, 2019