Autonomous Climbing Robot for Corrosion Inspection

Intelligent Manufacturing
Özgür Acar
Yildiz Technical University
Control and Automation Engineering
Turkey, Istanbul

Abstract—This paper describes a solution to a mobile climbing
robot with electromagnets, designed for inspecting fuel tank
surfaces made of metal sheets. A mechanical design has been
developed which can rotate intantaneous center of rotation.There
is information for used components and the ultrasonic test for
corrosion inspection. The inspection system has been developed
based on client/server architecture. The robot runs a server
application and a remote PC is client. Autonomous tracking
and mapping system powered by IMU sensor fusion and pid dc
motor position control with distrubances.
keywords: wall climbing robots,tank inspection, ultrasonic test,
sensor fusion

Wall-climbing robots, which have been developed in the
last few decades, are mainly employed for the tasks which are
hazardous and/or costly when performed by humans due to the
harsh environment. These tasks include remote maintenance of
large storage tanks, inspection of large concrete structures such
as bridge pylons, cooling towers or dams and inspection in narrow
spaces. One essential element of wall-climbing robot is the
adhesion mechanism, which decides the adhesion capability of
the device on the vertical wall or ceiling. [6] The application of

Fig. 1: Wind turbine corrosion test with manpower

mobile robots in high places doing work such as cleaning outer
walls of high-rise buildings, construction work, painting large
vessels and inspecting storage tanks in nuclear power plants is
required because they are currently performed predominantly
by human operators and are extremely dangerous. For this
reason, as a specific research field of mobile robotics, a number
of climbing robots capable of climbing vertical surfaces have
been researched and developed all over the world. Most climbing
robots developed at the present can be classified into two
main functions: locomotion and adhesion. With an adhesive
mechanism, climbing robots can attach to the wall by using
suction force, magnetic force, micro-spines for interlocking
and van der Waals force. The mechanism using magnetic force
is only available when the climbing environment is composed
of a ferromagnetic surface.[4] In the ship building industry,

Fig. 2: Robot concept and working surface, a)Grit blasting
robot design b)Ship hull curvature

many procedures are performed on ship hulls including the
removal of rust, the stripping of coatings and corrosion, the
cleaning of weld seams, non-destructive testing and inspecting,
and so on. Traditionally, these operations are done manually
by staff on aerial working platforms: the tasks are intense,
tedious, and potentially dangerous. A ship hull could be over
30 m high, with a surface area exceeding 20,000 m2, and could
include a variety of surface features such as: obstacles, high
curvature, weld seams, and so on.[2] The robot described here
was designed for inspecting gas tanks that are made out of thin
metal sheets and are installed in huge ships. From time to time,
they have to be analyzed for leaks, especially at the welds. For
doing this, helium is injected in the surrounding structure from
outside. A sensor that can detect this helium then is moved to
all places inside the tank to find the position of the leak. Until
now, this sensor was carried by a balloon that was operated
manually, using some ropes. As this method was very slow and
imprecise, a better inspection system, preferably consisting of
a climbing robot on magnetic wheels, had to be developed. As

Fig. 3: Surface cleaning robots already in application.
a)HydroCAT b)VAL-250 c)M250

the environment cannot support much force, the main goal was
to make this robot as light as possible. To simplify the control
and increase the reliability, another method was using only few
actuators. To ensure a correct functionality, the most critical
risks were analyzed. This analysis does not only incorporate
the possibility of some components within the robot breaking
down. It also accounts for the risks of plastically deforming
the environment, falling or slipping. [3]
A. Mechanical Design
The aim is to develop a climbing robot which can take an
ultrasonic sensor to every part of the oil tank’s surface and
will deal with the constraints of a real inspection work space.
B. Components
 Motors: There are 4 DC motors(with reduction and
encoder) for forward-downward movements, 2 Servo
motors for 180 Degree right-left movements of wheels,
and DC motor(with reduction and encoder) for linear
actuator’s movement.
 Electromagnets: are using for vertical movements. Vertical
movements needs friction forces, and electromagnets
provide that.
 Ultrasonic Distance Sensors: They are looking for obstacles
around of robot. Robot can avoid from obstacles by
ultrasonic distance sensors.
 Capacitive Sensors: They are looking for emptiness in
work-space of robot. If there is emptiness,robot stops in
this direction, and robot understands the wall is finished.
 IMU(inertial mesurement unit): This sensor is for correction.
If there is any skid or sliding in movement, this
sensor gives this states to robot.
 Wheels: The rotation mechanism is a critial problem
in mobile robots. The design in this paper solved this
problem. The wheels configuration allows it to rotate
around an instantaneous center of rotation
 Linear actuator: There is a linear actuator for movement
of NDT ultrasonic, nozzle, and camera.

Nozzles: NDT Ultrasonic sensor needs zero hole between
surface and probe. So, that holes are filled with water.
Nozzles sprays water in 120 Degree angle. In this prototype,
B-Scan data presentation is using.
 Camera: Camera is needed to show corrosion to client
users in ground.
 NDT Ultrasonic Sensor: this sensor gives value for corrosion
in sheet metal.
Ultrasonic Testing Ultrasonic Testing (UT) uses high frequency
sound waves ( typically in the range between 0.5 and
15 MHz) to conduct examinations and make measurements.
Besides its wide use in engineering applications (such as
flaw detection/evaluation, dimensional measurements, material
characterization, etc.), ultrasonics are also used in the medical
field (such as sonography, therapeutic ultrasound, etc.). In
general, ultrasonic testing is based on the capture and quantification
of either the reflected waves (pulse-echo) or the transmitted
waves ( through-transmission ). Each of the two types is
used in certain applications, but generally, pulse echo systems
are more useful since they require one-sided access to the
object being inspected. Basic Principle: A typical pulse-echo

UT inspection system consists of several functional units,such
as the pulser/receiver, transducer, and a display device. A
pulser/receiver is an electronic device that can produce high
voltage electrical pulses. Driven by the pulser, the transducer
generates high frequency ultrasonic energy. The sound energy
is introduced and propagates through the materials in the form
of waves. When there is a discontinuity (such as a crack) in the
wave path, part of the energy will be reflected back from the
flaw surface. The reflected wave signal is transformed into an
electrical signal by the transducer and is displayed on a screen.
Knowing the velocity of the waves, travel time can be directly
related to the distance that the signal traveled. From the signal,
information about the reflector location, size, orientation and
other features can sometimes be gained.[7] Data Presentation
Ultrasonic data can be collected and displayed in a number of
different formats. The three most common formats are known
in the NDT world as A-scan, B-scan and C-scan presentations.
Each presentation mode provides a different way of looking
at and evaluating the region of material being inspected.
Modern computerized ultrasonic scanning systems can display
data in all three presentation forms simultaneously. A-Scan
Presentation The A-scan presentation displays the amount of
received ultrasonic energy as a function of time. The relative
amount of received energy is plotted along the vertical axis

Fig. 7: Corrosion test by Siui NDT sensor probe and device

and the elapsed time ( which may be related to the traveled
distance within the material) is displayed along the horizontal
axis. Most instruments with an A-scan display allow the signal
to be displayed as a rectified signal, or as either the positive or
negative half of the signal. In the A-scan presentation, relative
discontinuity size can be estimated by comparing the signal
amplitude obtained from an unknown reflector to that from
a known reflector. Reflector depth can be determined by the
position of the signal on the horizontal time axis.

Fig. 8: Data representation types

B-Scan Presentation The B-scan presentation is a type of
presentation that is possible for automated linear scanning
systems where it shows a profile ( cross-sectional ) view of
the test specimen. In the B-scan, the time-of-flight ( travel
time) of the sound waves is displayed along the vertical axis
and the linear position of the transducer is displayed along the
horizontal axis. From the B-scan, the depth of the reflector and
its approximate linear dimensions in the scan direction can be
determined. The B-scan is typically produced by establishing
a trigger gate on the A-scan. Whenever the signal intensity is
great enough to trigger the gate, a point is produced on the
B-scan. The gate is triggered by the
sound reflected from the backwall of the specimen and by
smaller reflectors within the material. In the B-scan image
shown previously, line A is produced as the transducer is
scanned over the reduced thickness portion of the specimen.
When the transducer moves to the right of this section, the
backwall line BW is produced. When the transducer is over
flaws B and C, lines that are similar in length to the flaws and

at similar depths within the material are drawn on the B-scan.
It should be noted that a limitation to this display technique
is that reflectors may be masked by larger reflectors near the
C-Scan Presentation The C-scan presentation is a type of
presentation that is possible for automated two-dimensional
scanning systems that provides a plan-type view of the location
and size of test specimen features. The plane of the image is
parallel to the scan pattern of the transducer. C-scan presentations
are typically produced with an automated data acquisition
system, such as a computer controlled immersion scanning
system. Typically, a data collection gate is established on the
A-scan and the amplitude or the time-of-flight of the signal is
recorded at regular intervals as the transducer is scanned over
the test piece.[7]
The robotic inspection system developed is based on
client/server architecture. As Figure 3 shows, the client application
is related to the climbing robot tasks, while, the server
application runs on a remote PC. The two applications are
communicated by a wireless network using TCP/IP protocol.

the server application running at the robot CPU are: to
check the status of the robot’s sensor measures, position
estimation, trajectory generation, actuator control and
client communication.
 Client Application: The client application is focused
on configuring the inspection task, and monitoring the
robot’s status and store/visualization of the thickness
measures for a tank wall. Operator can set the application
to work in autonomous or manual joystick mode. User
Interface gives the tank details required by the dataset
(tank identification, plan, weld intersections to use as
landmarks), configuration of the robot trajectory and areas
to inspect in autonomous mode.[1]
A fundamental task for an autonomous mobile robot is
that of localization, i.e., determining its location in a known
environment. Absolute localization relies on landmarks, maps,
beacons, or satellite signals to determine the robot’s global
position and orientation. Dead-reckoning (open-loop estimation)
is commonly used for the estimation of position during
path execution. Deadreckoning is often used when wheel
encoders are available for drive wheel position measurement.
However, errors in kinematic model parameters, wheel slip, or
an uneven surface may cause poor position estimates to occur.
A worse scenario is one in which poor estimates would cause
a collision, thus clogging the robot’s operation. It is therefore
important to minimize errors in estimated position during the
path execution phase. Mobile robots use additional sensors to
deal with the localization problem. The localization strategies
are also based on the use of redundant information from the
sensorial system which would lead to data fusion methods.
Sensor fusion should give better results than a single sensor
given data error. In comparison to other wall climbing robot
inspection environments, oil tanks show special features such
as welding joins that could be used as landmarks. When the
robot passes over a weld, the robot position could therefore
be updated. To compute the absolute robot position in the
tank, we assume that: The oil tank landmarks are known from
the dataset The initial robot position is set up by the operator
The robot movements are considered in a vertical planar space
(x, y). The coordinate y therefore corresponds with the robot
height with regard to the floor, while the x coordinate is the
horizontal robot position with regard to the tank coordinate
origin. The absolute robot position is computed by developing
a sensor fusion process. The data fusion process takes into
consideration the position computed by means of the encoders
(odometry) and inclinometer, IMU measures and the well
known landmarks. [1]

Fig. 10: Static forces with movement directions

Control system has 3 disturbances: Gravity Force (When
robot goes up, gravity force is against force. When robot goes

Fig. 11: Block diagram of dc motor possition control

down, gravity force is assistant force.), Electromagnetic force,
this force is changing with the climbing angle.), Coloumb
friction force(it is about motor). Position control is using PID
controller and sensor fusion for correction of position.
This paper presents an autonomous climbing robot literature
reviews and new design ideas for the non-destructive inspection
of fuel tanks. The remote tank inspection system has been
expressed. And sensor fusion strategy to estimate the absolute
robot position have been explained. With all these climbing
robot can work autonomously on vertical tank’s sheet metal
surface. To sum up, there is examined autonomous climbing
robot for tanks inspection literature reviews and new design
This research was supported by Yildiz Technical University
Departmant of Control and Automation Engineering. Thanks
to Claudia Fernanda YAS¸AR who provided insight and expertise
that greatly assisted the research of this paper.
[1] Ra´ul Fern´andez, Elizabeth Gonzalez, Vicente Feliu Dep. de Ingenierıa
El´ectrica, Electr´onica y Autom´atica, E.T.S. Ingenieros Industriales Ciudad
Real, UCLM, SPAIN – Antonio Gonz´alez Rodr´ıguez Dep. de Mec´anica
Aplicada e Ingenier´ıa de Proyectos, , E.T.S Ingenieros Industriales Ciudad
Real, UCLM, SPAIN – A Wall Climbing Robot for Tank Inspection. An
Autonomous Prototype – 978-1-4244-5226-2/10/$26.00 IEEE
[2] Zheng-Yi Xu, Ke Zhang, Xiao-Peng Zhu and Hao Shi – Design and
Optimization of Magnetic Wheel for Wall Climbing Robot – Springer
International Publishing Switzerland 2015 T.-J. Tarn et al. (eds.), Robotic
Welding, Intelligence and Automation, Advances in Intelligent Systems
and Computing 363, DOI 10.1007/978-3-319-18997-0 54
[3] Wolfgang Fischer1, Fabien Tˆache2, and Roland Siegwart3 – Magnetic
Wall Climbing Robot for Thin Surfaces with Specific Obstacles – C.
Laugier and R. Siegwart (Eds.): Field and Service Robotics, STAR 42, pp.
551–561, 2008.
c Springer-Verlag Berlin Heidelberg
[4] Hwang Kim, Dongmok Kim, Hojoon Yang, Kyouhee Lee, Kunchan Seo,
Doyoung Chang and Jongwon Kim – Development of a wall-climbing
robot using a tracked wheel mechanism – Springer – Journal of Mechanical
Science and Technology 22 (2008) 1490 1498
[5] Francisco Ochoa-Cardenas and Tony J. Dodd – Design of a Continuously
Varying Electro-Permanent Magnet Adhesion Mechanism for Climbing
Robots – Springer International Publishing Switzerland 2015 C. Dixon
and K. Tuyls (Eds.): TAROS 2015, LNAI 9287, pp. 192–197, 2015. DOI:
10.1007/978-3-319-22416-9 23
[6] Yuan Chang and Xiaoqi Chen – Design of a Scalable Wall Climbing
Robot for Inter-plane Traversing – Springer International Publishing
Switzerland 2015 T.-J. Tarn et al. (eds.), Robotic Welding, Intelligence
and Automation, Advances in Intelligent Systems and Computing 363,
DOI 10.1007/978-3-319-18997-0-12
[7] Instructor: Dr. Ala Hijazi – Introduction to Non-Destructive Testing
[9] 210.html

Bir cevap yazın

E-posta hesabınız yayımlanmayacak.