Development and Controller Design of Wheeled-

Type Pipe Inspection Robot

Junghu Min, Yuhanes Dedy Setiawan, Hak Kyeong Kim

Pandu Sandi Pratama, Sang Bong Kim* Department of Mechatronics Engineering
Department of Mechanical Design Engineering Pukyong National University
Pukyong National University Busan, South Korea
Busan, South Korea line 4-e-mai
yuhanes_dedy@pknu.ac.kr l address if desired
*Corresponding author: kimsb@pknu.ac.kr

Abstract—A wheeled-type pipe inspection robot designed to Another type of pipe inspection robot is caterpillar type. In
work in 300 to 500 mm diameter pipe with multiple elbows is [7,8], active caterpillar pipe inspection robots were proposed.
introduced in this paper. The robot consists of two modules- Caterpillar type robots for indoor pipe inspection were
active module and passive module. Each module has three wheel developed in [9,10]. Each type of these robots has advantages
configurations with different mechanism to expand the wheels. A and disadvantages that makes a particular type robot works
proportional integral derivative controller is designed for the best only in a certain condition. In addition, some researches
robot to follow reference linear velocity and reference angular about special mechanism for robot moving in the pipeline
velocity. Simulations and experiments are conducted to verify were proposed [11,12]. Control and driving algorithms was
performances of the proposed controlled robot. The results
also researched and developed to make pipe inspection robot
demonstrate that the robot with the designed controller can work
move effectively [13,14].
well by following the reference velocity.
A wheeled-type pipe inspection robot designed to work in
Keywords—Pipe inspection robot; wheeled-type; controller 300 to 500 mm diameter pipe with multiple elbows is
design introduced in this paper. The robot consists of two modules-
active module and passive module. Each module has three
I. INTRODUCTION
wheel configurations with different mechanism to expand the
Pipeline has been one of the effective and efficient wheels. A PID controller is designed for the robot to follow
transportations to transfer essential fluid things, such as water, reference linear velocity and reference angular velocity.
oil and gas from one place to other places. Therefore, pipeline Simulations and experiments were conducted to verify
has been used for such a long time and become popular until performances of the proposed controlled robot. The results
now because its cost is relatively low compared to other demonstrate that the robot with the designed controller could
transportation such as transportation vehicles. However, since work well by following the reference velocity. The
pipeline is made by metal, pipeline does not have long contribution of this paper over the literatures is that this paper
operating time, due to deterioration, corrosion, fluid leaking proposes a new design of pipe inspection robot that has active
and plugging that commonly happen after few years of and passive modules which are connected with universal joint
installation. Thus, regular maintenance is essential to prevent to enable the robot move in flexible ways.
these problems. However, since most pipelines are located
underground, it is difficult for humans to perform maintenance
in pipeline directly. There have been some ways developed to
help humans perform maintenance in pipeline.
Pipe inspection robots have been utilized widely to
perform maintenance in pipeline. In [1], seven kinds of in-pipe
robots used worldwide were reported. Pig type which is
usually used in oil and gas pipeline is passively driven by
liquid or gas passing through the pipe [2]. Another type of the
robot is wheeled type robots, which are generally used for
commercial purpose. Wheeled in-pipe inspection vehicles for
φ 25, φ 50, φ 150 pipes were designed in [3,4]. Wheeled pipe
robot system with active steering capability was developed for
internal inspection of urban gas pipelines in [5]. And in [6] Fig. 1. The designed pipe inspection robot.
wheeled differential drive inpipe robot was developed for
navigation inside pipeline.

978-1-4799-3080-7/14/$31.00 2014
c IEEE 789
 II. CHARACTERISTICS OF ROBOT Fig. 4 depicts the structure design of the passive module.
Passive module has similar structure with the active module.
A. Structure Instead of thread shaft in the active module, compression
The designed robot consists of two modules, namely active spring is used for expansion and contraction of 4-bar linkage
and passive modules. The active term refers to the use of in the passive module. This is because pipe inspection robot
motor to expand and contract the wheels, whereas the passive only needs one active mechanism module to work in pipelines.
term refers to the use of spring. The active module is used for
giving pressure to the wall so that robot can grip the pipe wall
especially moving in elbows or vertical pipe. On the other
hand, the passive module is used to support the active module
in case that the active module could not move in some
conditions such as moving through elbow, and/or passing
through choked section in pipeline.
Fig. 2 shows the structure design of the active module in
SolidWorks. This module consists of three wheel
configurations, a thread shaft, three 4-bar linkage, two gears
and a high power 33.5 W expansion motor to expand and
contract the wheels. Each wheel configuration consists of a
12.8 W DC motor for driving motor and rubber wheel sets. 4-
bar linkages are used to assure the robot to have strong grip on
Fig. 4. Structure of passive module.
the pipe wall as depicted in Fig. 3. Rubber wheels make the
wheel give bigger wall press forces. Expansion and
contraction of the 4-bar linkage is done by rotating the thread
shaft which is actuated by expansion motor and thus making B. Comparison of the Wheeled-Type Pipe Inspection Robot
the shaft holder move forward and backward. Shaft holder, as and Other Type Robots
shown in Fig. 3, is used to transfer energy from thread shaft to Comparison of wheeled type with other type of pipe
the 4-bar linkage. inspection robot is shown in Table I.
TABLE I. COMPARISON OF PIPE ROBOTS

Type Caterpillar Mobile robot Wheeled type

Structure Wall press type Car like robot Wall press type
Principle Moving by Moving on the Moving by
of motion pressing and ground pressing and
walking on pipe walking on pipe
wall wall
Exterior Big Small Big
diameter
Advanta- Adaptable to Does not damage Moves fast as
ges changes of inner the inner surface wheels provide
pipe diameter of pipe due to highly efficient
small contact area propulsion
Disad- High friction force Unable to move in Wheel can be
vantages may damage the vertical path stuck in holes
inner surface pipe on the inner
pipe
Fig. 2. Structure of active module.
C. Specification of the Robot
Specification of the robot is shown in Table II. The total
length of the robot, including two robot modules and a
universal joint, is 685 mm. The total weight of the robot is 11
kg. The diameter range of pipe that this robot can work is 300
to 500 mm.

III. CONTROLLER DESIGN

To design the controller for the robot, kinematic modeling
from [1] is used. First of all, definition of xyz direction is
determined as shown in Fig. 5. XYZ denotes the global
Fig. 3. Structure of 4-bar linkage. coordinate frame and xyz represents the local coordinate frame

790 2014 International Conference on Advances in Computing,Communications and Informatics (ICACCI)

 TABLE II. SPECIFICATION OF THE ROBOT

Specification Value
vcz =
1
3
r
3
(
( v1 + v2 + v3 ) = θ1 + θ2 + θ3 ) (4)

Total weight 11 kg
The relationship between the input angular velocity vector
Length of active module 255 mm T
of wheels ș a = ª¬θ1 θ2 θ3 º¼ and the output velocity vector
Weight of active module 5.5 kg
T
Length of passive module 220 mm u = ª¬ω x ωy vcz º¼ is constructed as
Weight of passive module 4.5 kg
u = ª¬G ua º¼ ș a (5)
Total length of the robot 685 mm
Exterior diameter 300-500 mm
and the Jacobian is given as
Nominal speed 9 cm/s
Maximum speed 21 cm/s ª 3r 3r º
« 0 − »
Camera module length 135 mm « 3a 3a »
Serial communication device 15 m ªG a º = ∂u « 2r r r »
¬ ¼ ∂θ = « − 3a
u
(6)
a « 3a 3a »»
« r r r »
« 3 3 3 »¼
¬

The tracking error of the output velocity vector of the robot

is defined as follows:

ª e1 º ªωrx º ªω x º
« » « »
e = ««e2 »» = «ωry » − «ω y » = u r − u (7)
«¬ e3 »¼ «v » « v »
¬ rcz ¼ ¬ cz ¼

T
where u r = ª¬ωrx ωry vrcz º¼ is reference velocity vector of
Fig. 5. The xyz definition in pipe inspection robot.
the robot.
at the center of the pipeline inspection robot. iˆ , ĵ , and k̂ are
the unit vectors of the local coordinate frame. The x-axis is The tracking error of the input angular velocity vector of
always in parallel with P1 regardless of how the robot moves, the wheel is defined as:
where P is normal force of the pipe wall to the wheel.
ª eθ 1 º ªθ1r º ªθ1 º
By assuming that linear velocities of all wheels v1, v2, and « » « »
v3 exist at the same time, the total rotational velocity vector of eθ = ««eθ 2 »» = «θ2 r » − «θ2 » = ș ar − ș a (8)
the robot is obtained by: «¬ eθ 3 »¼ «θ3r » «θ3 »
¬ ¼ ¬ ¼
ω = ω x iˆ + ω y ˆj (1)
where where ș ar is the input angular velocity reference.
3r  3r  The relation between eș and e is given by
ωx = θ2 − θ3 (2)
( )
3a 3a −1
and eș = Gua e (9)
2r  r r
ωy = − θ1 + θ2 + θ3 (3) The DC motor system is shown in Fig. 6. The system state
3a 3a 3a
variable vector is chosen as
T
r is radius of the wheels and a is the distance from the center Į = ª¬i a ș a º¼ (10)
of the robot to the center of the wheel. ω x and ω y are angular
velocity vectors of the robot in x and y axes, respectively.
θ1 , θ2 , and θ3 are input angular velocities of the three wheels. where i a = [i1 i2 i3 ] is current vector on motor and i1 , i2 ,
T

The linear velocity at the center of the robot is obtained by and i3 are currents on motor 1, motor 2, and motor 3,
taking the average of linear velocities of the wheels as respectively.

2014 International Conference on Advances in Computing,Communications and Informatics (ICACCI) 791

 Fig. 6. DC motor system.

The system equation of DC motor is expressed by

Į = AĮ + Bv m (11)

ª R Kφ º Fig. 8. Pipeline for simulation and experiment.

«− − » ª1º
L L »
A=« , B = «L»
« Kφ b » « »
« − » ¬0¼
¬ J J ¼

where v m is motor voltage vector as an input, b is viscous

friction, J is moment of inertia for the motor load, Kφ is
armature or emf constant, L is motor inductance, and R is
motor resistance.
Then, a PID controller for the pipe inspection robot (11) to
track reference linear velocity and reference angular velocity
is designed as follows:

ª vm1 º
de
v m = ««vm 2 »» = K p eθ + K I eθ dt + K D θ
³ (12)
dt
«¬ vm3 »¼ Fig. 9. Elbow and pipe specification (mm).

where K p , K I , and K D are 3x3 diagonal matrices, and v m is TABLE III. PARAMETER AND INITIAL VALUES
motor input voltage vector of the wheel motor. Parameter Value Unit
L 0.1 H
Block diagram of the proposed PID controller is shown in
Fig. 7. Kφ 0.3 -
J 0.1 Kg.m2
b 0.01 -
e eθ vm ș u R 2 ȍ
u r (G ua ) −1 G ua
r 47.45 mm
a 102.55 mm
ia [ 0 0 0 ]T A
Fig. 7. Block diagram of the proposed controller.
u [ 0 0 0 ]T [rad/s rad/s mm/s]T
ș [ 0 0 0 ]T rad/s
a
IV. SIMULATION RESULTS
Simulations were done to verify the proposed controller and its inverse matrix is
performances. A pipeline used for simulation and experiment ª 0 −2.16 0.021º
is shown in Fig. 8 and its dimension is shown in Fig. 9. The −1
parameter values used in the simulation are listed in Table III.
ªG ua º
¬ ¼ = «« 1.87 1.08 0.021»»
«¬ −1.87 1.08 0.021»¼
By subtituting r = 47.45 mm and a = 102.55 mm to (6), the
Jacobian matrix becomes
Likewise, by using motor parameters in Table III, A and B
ª 0 0.267 −0.267 º
in (11) can be calculated.
ªG ua º = « −0.309 0.154 0.154 »
¬ ¼ « »
¬« 15.82 15.82 15.82 ¼»

792 2014 International Conference on Advances in Computing,Communications and Informatics (ICACCI)

 ª −20 −3 º ª10 º
A=« », B=« »
¬ 3 −0.1¼ ¬0¼

In the simulation, the robot moves five paths, i.e. move

straight for 950 mm, move in elbows for 703.63 mm, move
straight for 700 mm, move in elbows for 703.63 mm, and
finally move straight for 700 mm. Nominal linear velocity
( vcz ) = 90mm/s is used as reference in the simulation. When
the robot moves in elbows ( relbow = 447.95 mm), the angular Fig. 11. Angular velocity with respect to y axis.
velocity ω x of the robot is
vcz 90 mm s
ωx = = = 0.2 rad s
relbow 447.95mm

The linear velocity = 90 mm/s and angular velocity = 0.2

rad/s are taken as references for simulation. Figs. 10-12 show
the simulation results. The results show that the proposed
controller can enable the robot to follow the reference
velocities very well. Fig. 10 shows the simulation result of
angular velocity of the robot with respect to x axis. This figure
shows that the robot was able to follow the reference angular Fig. 12. Linear velocity.
velocity very well in straight and elbow pipelines. The big
rapid change in the beginning of simulation result is usually
caused by nonlinearity of DC motors.
Fig. 11 shows the simulation result of the angular velocity
of the robot with respect to y axis. Since the robot did not
move in vertical direction, the reference value in vertical
direction was zero. Fig. 11 indicates that the proposed PID
controller was good enough since the robot could follow the
reference very well in straight and elbow pipelines. The small
rapid changes that appear in the figure occurred when the
Fig. 13. Wheel velocities.
robot turned in the elbow. Then, simulation result for the robot
linear velocity can be seen in Fig. 12. This figure shows that
the robot could follow the reference well although it took
about 18 seconds for the robot to reach the reference linear
velocity. Then, wheel angular velocities are shown in Fig. 13.
It can be seen that when the robot moved in straight pipeline,
the three wheels had the same speed, whereas when the robot
turned in the elbow pipeline, the wheels had different speed to
enable the robot to turn in the elbow. Figs. 14 and 15 show
simulation results of motor current and motor voltage as
control input, respectively.
Fig. 14. Motor current.

Fig. 10. Angular velocity with respect to x axis.

Fig. 15. Motor voltage.

2014 International Conference on Advances in Computing,Communications and Informatics (ICACCI) 793

 V. IMPLEMENTATION
A. Hardware Architecture
In the implementation, active and passive modules are
connected with a universal joint to give flexibility for active
and passive modules. Wireless camera, VIJE IP-2000 PTW, as
shown in Fig. 16, is used to enable users monitor the robot in Fig. 18. Flow chart for the robot system.
the pipe. The motion of the robot is made by driving motors in
the wheel configurations. The purpose of using collaboration
of active and passive modules is to support each other in
moving in pipelines as shown in Fig. 17. If the active module
cannot move in a certain position, the passive module will
push the active module such that this robot can move.
Likewise, if the passive module cannot move, the active
module will pull the passive module. It is very useful
especially when the robot cannot move in elbows since
sometimes wheels cannot touch and grip the elbow wall.
Fig. 19. Components of pipe inspection robot.

Fig. 16. VIJE IP-2000 PTW wireless camera.

Fig. 20. Motor control module.

Fig. 17. Collaborative structure between two modules
The robot working principle is relatively straightforward
since a computer with a Graphic User Interface (GUI) is
provided for users to control the robot manually and obtain
vision information inside the pipe. Fig. 18 shows a flow chart
of how the robot works. Control order which is given by users
in GUI is sent to Atmega 8. Then, Atmega 8 translates it into
Pulse Width Modulation (PWM). After that, DC Motor is
actuated by this signal and finally the robot is driven by the
DC motor. Information data inside the pipe is obtained from
two sensors, i.e. camera and encoder. By using the encoder,
position of the robot can be examined. Fig. 19 shows the
picture of the whole pipeline inspection robot system that
mainly consists of a robot device and a control system. The
control PC is used to display GUI and vision information
obtained from camera. The motor control module consists of
multipoint control unit (MCU) printed circuit board (PCB)
(Atmega 8), a motor driver PCB and a sensor processor (AVR)
as shown in Fig. 20.
B. Control Process
For control process, C# and AVR ATmega are used for
GUI and motor driver control, respectively. For monitoring
using the camera, built-in software VIJE Savitmicro is used as
shown in Fig. 21 which also shows the display of C# GUI.
There are three parts in the GUI, i.e. camera control panel
where users control the camera directions, camera vision
information window that displays images taken by camera,
and robot control panel to control the robot motion. In the
robot control panel, there are three buttons, namely backward,
stop, and forward. These buttons simply control the robot to
move forward and backward. The PWM value that is sent to
motors is displayed at the three left bottom boxes, whereas the
PWM value that is read by encoders is shown at the right
bottom boxes.
VI. EXPERIMENTAL RESULTS
Experiments were conducted to verify performances of the
designed robot and the proposed PID controller. The
experiment was conducted by making the robot following the
reference velocities that was made previously in the
simulation. The result shows that the proposed PID controller
could make the robot go through the pipe successfully by
following the reference velocities in horizontal straight and
elbow pipelines.

794 2014 International Conference on Advances in Computing,Communications and Informatics (ICACCI)


Fig. 21. Graphic User Interface.
Fig. 22. Experimental results.
VII. CONCLUSION
Wheeled-type pipe inspection robot was developed to
move inside 300 - 500 mm diameter pipe with multiple
elbows. This robot consisted of two modules, i.e. active and
passive modules. The active term referred to the use of motor
to expand and contract the wheels, whereas the passive term
referred to the use of spring. A PID controller was applied to
make the robot able to follow the reference linear velocity and
reference angular velocity. Simulations and experiments were
2014 International Conference on Advances in Computing,Communications and Informatics (ICACCI)
conducted to verify performances of the designed robot and
the proposed controller. The results showed that, with the
proposed controller, the designed robot could work very well
in horizontal straight and elbow pipelines.
ACKNOWLEDGMENT
This research was financially supported by the Ministry of
Trade, Industry and Energy (MOTIV), Korea Institute for
Advancement of Technology (KIAT) and Dongnam Institute
for Regional Program Evaluation through the Leading Industry
Development for Economic Region.
REFERENCES
[1] S.G. Roh and H.R. Choi, “ Differential-drive in-pipe robot for moving
inside urban gas pipelines,” in IEEE Transactions on Robotics, vol. 21,
no. 1, Febraury 2005, pp. 1-17.
[2] J. Okamoto, Jr., J.C. Adamowski, M. S. G. Tsuzuki, F. Buiochi, and C.S.
Camerini, “Autonomous system for oil pipelines inspection,”
Mechatronics, vol. 9, 1999, pp. 731-743.
[3] S. Hirose, H. Ohno, T. Mitsui, and K. Suyama, “Design of in-pipe
inspection vehicles for φ 25, φ 50, φ 150 pipes,” in Proc. IEEE Int.
Conf. Robotics, Automation, 1999, pp. 2309-2314.
[4] K. Suzumori, T. Miyagawa, M. Kimura, and Y. Hasegawa, “Micro
inspection robot for 1-in pipes,” IEEE/ASME Trans. Mechatronics, vol.
4, Sep. 1999, pp. 286-292.
[5] H. Choi and S. Ryew, “Robotic system with active steering capability
for internal inspection of urban gas pipelines,” Mechatronics, vol. 26, no.
1, 2002, pp. 105-112.
[6] S. Roh and H. Choi, “Strategy for navigation inside pipelines with
differential-drive inpipe robot,” in Proc. IEEE Int. Conf. Robotics,
Automation, 2002, pp. 2575-2580.
[7] J.W. Park, T.H. Kim, and H.S. Yang, “Development of an actively
adaptable in-pipe robot,” in Proc. of the 2009 IEEE International
Conference on Mechatronics, 2009.
[8] D.W. Kim, C.H. Park, H.K. Kim, and S.B. Kim, “Force adjustment of an
active pipe inspection robot,” in Proc. ICROS-SICE International Joint
Conf. 2009, pp. 3792-3797.
[9] P. Harish and V. Venkateswarlu, “Design and motion planning of indoor
pipeline inspection robot,” International Journal of Innovative
Technology and Exploring Engineering, vol. 3, iss. 7, December 2013.
[10] Y.S. Kwon and B.J. Yi, “Design and motion planning of a two-module
collaborative indoor pipeline inspection robot,” in IEEE Transactions on
Robotics, vol. 28, no. 3, June 2012, pp. 681 – 696.
[11] W. Neubauer, “A spider-like robot that climbs vertically in ducts or
pipes, “ in Proc. IEEE/RSJ Int. Conf. Intelligent Robots Systems, 1994,
pp. 1178-1185.
[12] W.S. Jeon, J.W. Park, I.H. Kim, Y.K. Kang, and H.S. Yang,
“Development of high mobility in-pipe inspection robot,” in Proc.
IEEE/SICE Int. Symposium on System Integration, 2011, pp. 479-484.
[13] H.P. Huang, J.L. Yan, and T.H. Cheng, “Development and fuzzy control
of a pipe inspection robot,” IEEE Transactions on Industrial Electronics,
vol. 57, no. 3, March 2010, pp. 1088-1095.
[14] S.G. Roh, D.W. Kim, J.S. Lee, H.P. Moon, and H.R. Choi, “In-pipe
robot based on selective drive mechanism,” International Journal of
Control, Automation, and Systems, vol. 7(1), 2009, pp. 105-112.
795