19.robust railway crack detection scheme (rrcds)

Robust Railway Crack Detection Scheme (RRCDS)
Using LED-LDR Assembly
Selvamraju Somalraju

Vigneshwar Murali

Gourav Saha

Dr. V. Vaidehi

Dept. of Instrumentation
Engg.
Madras Inst. Of Tech., Anna
Univ.
Chennai, India

Engg.
Univ.
Chennai, India
vinima1991@hotmail.com

Engg.
Univ.
Chennai, India
sahahotmail@hotmail.com

Dept. of Information Tech.
Univ.
Chennai, India
vaidehi@annauniv.edu

ssr_selvamraju@yahoo.com

Abstract—In India, most of the commercial transport is being
carried out by the railway network and therefore, any problems
in the same has the capacity to induce major damage to the
economy-notwithstanding the societal impact of loss of life or
limb. This paper proposes a cost effective yet robust solution to
the problem of railway crack detection utilizing a method that is
unique in the sense that while it is simple, the idea is completely
novel and hitherto untested. The paper discusses the technical
and design aspects in detail and also provides the proposed
robust crack detection algorithm. The paper also presents the
details of the implementation results of the RRCDS utilizing
simple components inclusive of a GPS module, GSM Modem and
LED-LDR based crack detector assembly. The proposed scheme
has been modeled for robust implementation in the Indian
scenario.

of our nation. Though rail transport in India growing at a rapid
pace, the associated safety infrastructure facilities have not kept
up with the aforementioned proliferation. Our facilities are
inadequate compared to the international standards and as a
result, there have been frequent derailments that have resulted
in severe loss of valuable human lives and property as well. To
demonstrate the gravity of the problem, official statistics say
that there have been 11 accidents in 2011 till the month of July
alone, which leaves much to be desired. On further analysis of
the factors that cause these rail accidents, recent statistics
reveal that approximately 60% of all the rail accidents have
derailments as their cause, of which about 90% are due to
cracks on the rails either due to natural causes (like excessive
expansion due to heat) or due to antisocial elements. Hence
these cracks in railway lines have been a perennial problem,
which has to be addressed with utmost attention due to the
frequency of rail usage in India. These cracks and other
problems with the rails generally go unnoticed due to improper
maintenance and the currently irregular and manual track line
monitoring that is being carried out. The high frequency of
trains and the unreliability of manual labour have put forth a
need for an automated system to monitor the presence of crack
on the railway lines.

Keywords- Railway Cracks, Arduino, LED-LDR assembly,
GSM, GPS, Robot.

I.

INTRODUCTION

In today’s world, transport is a key necessity because in its
absence it would be impossible for products to be consumed in
areas which are not in the immediate vicinity of the production
centers. Throughout history, transport has been a necessity for
the expansion of trade. Economic prosperity can be achieved
by increasing the rationality and capacity of transport systems.
The proper operation and maintenance of transport
infrastructure has a great impact on the economy. Transport,
being one of the biggest drainers of energy, its sustainability
and safety are issues of paramount importance. In India, rail
transport occupies a prominent position in quenching the everburgeoning needs of a rapidly growing economy. However, in
terms of the reliability and safety parameters, global standards
have not yet been truly reached.

Owing to the crucial repercussions of this problem, this
paper presents an implementation of an efficient and cost
effective solution suitable for large scale application.
With the advent of powerful digital signal processors,
Image Processing techniques [1] have been explored to
formulate solutions to the problem of railway crack detection.
Though it provides good accuracy, this method uses
techniques like image segmentation, morphology and edge
detection all of which take a lot of processing power and an
extreme amount of time rendering the robot slow and thereby
unsuitable. Recent research has investigated the use of
microwave horn antennas for crack detection [2]. This
technique was found to produce very accurate results in lab
based testing. But, unfortunately it requires spectrum
analyzers which are both costly and also can’t be placed onboard a moving robot because of their delicacy. Eddy current
based methods ([3], [4] and [5]) are used to tide over
limitations associated with ultrasonics and microwave

The Indian railway network today has a track length of
113,617 kilometres (70,598 mi).over a route of 63,974
kilometres (39,752 mi) and 7,083 stations. It is the fourth
largest railway network in the world exceeded only by those of
the United States, Russia and China. The rail network traverses
every length and breadth of India and is known carry over 30
million passengers and 2.8 million tons of freight daily. Despite
boasting of such impressive statistics, the Indian rail network is
still on the growth trajectory trying to fuel the economic needs

ISBN: 978-1-4673-1601-9/12/$31.00 ©2012 IEEE

477

ICRTIT-2012

FIGURE 1. HIGH LEVEL CONCEPTUAL DESIGN

techniques. However they have the problem of very slow
overall speed which reduces the usability of the same. A vast
majority of the work done in the field of crack detection uses
the infrared sensing technique ([6], [7] and [8]). It is a well
understood technique so much so that it was initially thought
to be the best solution to the problem of crack detection, but
later it was found to be prone to external disturbances and
hence came to be considered inaccurate. Techniques that
employ ultrasonics ([9], [10] and [11]) tide over some of the
problems mentioned earlier, but they can only inspect the core
of the track; that is, it cannot check for surface and nearsurface cracking where most faults are usually located. Several
other miscellaneous techniques like observation and analysis
of wave propagation via model impacts and piezo actuation
[12] have also been developed.

offer several advantages, when their applicability for large
scale implementation in the current Indian scenario was
considered, they were found to lack robustness and practicality
in a number of aspects.
First, in the Indian rails, typically there are small gaps in the
rail tracks to provide for thermal expansion during the summer.
This design is provided so as to ensure that the track does not
twist or crack due to the heat. When the existing technique of
crack detection was implemented, it was found that the system
was giving false positive signals; that is, it was counting the
thermal gaps as cracks.
Another issue faced during practical implementation is the
presence of railway bifurcations. If the mechanical design of
the robot is unsuitable, then it will have a tendency to either get
stuck in these bifurcations or in worst case even fall out of the
tracks.

The problem inherent in all these techniques is that the
cost incurred is high. Hence this paper proposes a cheap, novel
yet simple scheme with sufficient ruggedness suitable to the
Indian scenario that uses an LED-LDR arrangement to detect
the crack in railway lines, which proves to be cost effective as
compared to the existing methods ([13], [14] and [15]). The
important role played by transport in the development of an
economy has been studied [16]. In addition, statistics of the
number of rail accidents and their corresponding causes have
also been studied [17].

During the designing of prototype for actual on-field
implementation, the problem of presence of debris on the
outsides of the tracks was encountered. Though this problem
seemed trivial, the effects of dirt on our robot wheels could
have been substantial.
In addition, as the proposed design utilized a LED-LDR
based design, the ambient light intensity variations imposed
extreme challenges to our design concept.

This paper is organized as follows: Section II discusses the
design issues; Section III discusses the proposed RRCD
Scheme using an LED-LDR arrangement. Section IV
elaborates on the electrical design and Section V explains the
mechanical design. Section VI gives the implementation of
pseudo code, Section VII provides the results of
implementation and Section VIII provides the conclusion.
II.

III.

PROPOSED RRCD SCHEME

In the process of designing the prototype, Chennai’s
Suburban Railway System, South Line, which runs between
Tambaram to Chennai Beach spanning a total of 22Km was
considered as the testing and usage area. This railway line
doesn’t operate between 12:30 am to 4:30 am. This gives us a
four hour window during which the robot has to traverse the
railway line looking for cracks. Figure 1. illustrates the overall
conceptual design of the proposed scheme.

DESIGN ISSUES INHERENT TO INDIAN SCENARIO

A literature survey on the existing techniques for crack
detection revealed a number of sophisticated and accurate
crack detection technologies. Whilst techniques based on
ultrasonic imaging, IR method and electromagnetic detection

To
ensure
robustness,
repeatability and
easy
implementation, the principle idea has been kept very simple.

478

ICRTIT-2012

FIGURE 2. TECHNICAL DESIGN

The core of the proposed crack detection scheme consists of a
Light Emitting Diode (LED)-Light Dependent Resistor (LDR)
assembly that functions as the rail crack detector. The principle
involved in crack detection is the concept of LDR. In the
proposed design, the LED will be attached to one side of the
rails and the LDR to the opposite side. During normal
operation, when there are no cracks, the LED light does not fall
on the LDR and hence the LDR resistance is high.
Subsequently, when the LED light falls on the LDR, the
resistance of the LDR gets reduced and the amount of
reduction will be approximately proportional to the intensity of
the incident light. As a consequence, when light from the LED
deviates from its path due to the presence of a crack or a break,
a sudden decrease in the resistance value of the LDR ensues.
This change in resistance indicates the presence of a crack or
some other similar structural defect in the rails. In order to
detect the current location of the device in case of detection of
a crack, a GPS receiver whose function is to receive the current
latitude and longitude data is used. To communicate the
received information, a GSM modem has been utilized. The
function of the GSM module being used is to send the current
latitude and longitude data to the relevant authority as an SMS.
The aforementioned functionality has been achieved by
interfacing the GSM module, GPS module and LED-LDR
arrangement with a microcontroller. The robot is driven by four
DC motors.

IV.

ELECTRICAL DESIGN

A. Microcontroller
An Arduino Uno board which has ATMega328
microcontroller forms the brain of the scheme. This board has
been chosen for two important reasons other than the fact that
it is cost effective. First, the Arduino integrated development
environment (IDE) is an open-source project which highly
simplifies the coding and debugging process. Secondly it has
all the required pins to interface the required peripherals. It has
6 analog input pins, 14 digital I/O pins (of which 6 provides
PWM output) and one UART. The detailed description about
how various components have been interfaced with Arduino is
also discussed hereafter.
B. GPS module
EM-406 GPS receiver has been used as the GPS module. It
follows NMEA convention. With a baudrate of 9600 bps, 1Hz
update rate and 1 sec hot start time, the properties of the said
module was found to ideally match the requirements. It is
interfaced with Arduino using the UART.
C. GSM module
The SIM 300 GSM module has been chosen to achieve the
SMS functionality. Since the Arduino Uno board has only one
UART, it was necessary to program 2 of the digital pins (pins
2 and 3 in our case) of Arduino to act like a virtual UART so
as to interface the GSM with the Arduino. The overall
electrical design of the RRCDS has been shown in Figure 2.

Before the start of the railway line scan the robot has been
programmed to self-calibrate the LED-LDR arrangement. It is
necessary because the LDR has a natural tendency to show a
drifting effect because of which, its resistance under the same
lighting condition may vary with time. After calibration, the
robot waits for a predetermined period of time so that the onboard GPS module starts reading the correct geographic
coordinate. This is necessary because any GPS module will
take some time to synchronize with the satellites.

D. DC Motors
To traverse a distance of 22 Km in 4 hrs, an average speed
of 1.5 meters/sec is needed. The proposed design uses 4 DC
motors (Torque Rating: 10Kg and Speed Rating: 500 rpm)

479

ICRTIT-2012

stones and other debris are comparatively less on the inner
side tracks. If the bigger wheels are placed outside it may
brush against the debris causing it to destabilize or in worst
case get stuck or even fall.
2) The LED-LDR assembly shouldn’t go below the rim of
the rail otherwise it may get damaged due to the scattered
debris.
FIGURE 3. DEAD BAND CONCEPT

3) The distance between the front wheel and the LEDLDR assembly is a crucial design aspect. The front wheel of
the robot should be kept sufficiently behind the LED-LDR
assembly so that the robot has sufficient distance to stop after
a crack is detected. In our case it is 12 cm.

interfaced with the Arduino using H-Bridges. With a wheel
diameter of 5.2 cm and the total mass of around 5 Kg the
approximate speed of the robot is around 0.5 metres/sec.
Hence it has been calculated that three such robots would be
required to scan the whole Southern Chennai Suburban
Railway System.

VI.

After the robot is powered ON it executes the following
algorithm:
1) Set LowThreshold = 200, HighThreshold = 800.
2) Calibrate LDR:
a) Switch OFF left LED
b) Set LOWleft = Average of left LDR signal
c) Switch OFF right LED.
d) Set LOWright = Average of right LDR
signal.
e) Switch ON left LED.
f) Set LOWleft = Average of left LDR signal.
g) Switch ON right LED.
h) Set LOWright = Average of right LDR
signal.
3) Turn ON GSM. Set GSM parameters.
4) Turn ON GPS.
5) REPEAT
a) Read Latitude from NMEA string
b) Read Longitude from NMEA string
UNTIL (Latitude and Longitude not equal Zero)
6) Turn ON motors.
7) Read left LDR and right LDR signal.
8) Map left and right LDR signals between 0 and 1000
using following formulas
(ana log Re ad (LDRleft ) − LOWleft ) × 1000.0
INTENSITYleft =
(HIGHleft − LOWleft )

E. LED-LDR Assembly
The common 5V LED and cadmium sulphide LDR was
found to be sufficient. The LED is powered using one of the
digital pin of the Arduino. The LDR and a 45kΩ resistor form
a potential divider arrangement. The output of the potential
divider is given to one of the analog input channel of the
Arduino. The LDR is calibrated every time the robot is used.
To compensate for the ambient light we use the concept of
deadband. Figure 3. clearly illustrates it.
V.

IMPLEMENTED ALGORITHM

MECHANICAL DESIGN

The mechanical design of the robot is clearly illustrated in
Figure 4. The robot runs on both the railway tracks. This
increases its stability preventing it from falling when it moves
over a railway bifurcation. In addition, the robot has been
designed to be symmetrical. It consists of two wooden
frameworks each supporting 2 motors, 1 battery and one LEDLDR assembly. Each battery was found to weigh a little over
300 grams giving additional weight on the wheels which also
ensured stability of the robot when it moved over railway
bifurcations. These two wooden frameworks were connected
by two cylindrical aluminum rods (3/4 inch diameter and 0.25
mm thickness). The length of the aluminum is so chosen that
the four wheels of the robot rest exactly on a typical broad
gauge railway track. In India, the distance between two rails in
a broad gauge railway is 1.676 meter. The circuit box
containing mainly the Arduino Uno Board, the GPS and the
GSM module is exactly centered on the aluminum rod. Two
bunches of 10 wires (2 each for the LED, the LDR, the two
motor and the battery) each enters the circuit box from its left
and right side. The proper packaging of these many wires is a
crucial in design of the robot.

INTENSITYright =

(ana log Re ad (LDRright ) − LOWright ) × 1000.0
(HIGHright − LOWright )

analogRead(LDRleft) and analogRead(LDRright)
Signal
from LDRs
INTENSITYleft and INTENSITYright
Mapped values
9) If
INTENSITYleft
<
LowThreshold
and
INTENSITYright < LowThreshold then,
a) Motors are powered ON.

There are few more design criteria which were taken into
account:
1) The wheels of the robot will be similar to the wheels of
the a train, i.e. a big wheel welded/joined with a smaller
wheel. The smaller wheel runs on the track while the bigger
one prevents the robot from falling. It is must that the bigger
wheel is on the inner side of the railway track (as shown in
figure 4). It is because in the general Indian scenario the

480

ICRTIT-2012

FIGURE 4. MECHANICAL DESING OF THE ROBOT

the GSM modem (with an Airtel SimCard) to a mobile
phone indicating the co-ordinates of the artificial crack. The
crack was detected exactly at a distance of 200 metres from
the start of the course of the robot, which is exactly the
distance at which the crack was created. These tests were
also conducted in uniform and un-uniform lighting
conditions and no false output was detected in either case.
The crack detection was also tested for different distances of
the created crack with no false output detected.

10) If
INTENSITYleft
<
HighThreshold
and
INTENSITYright < HighThreshold then,
a) Motor is powered OFF.
b) Read robots coordinate using GPS.
c) Send robots coordinate to a mobile as SMS
using GSM
11) Jump to step 7.
VII. IMPLEMENTATION AND RESULTS

Insufficiant GSM and GPS signal, has however been a
problem during simulated tests, while this problem was not
observed during the field tests due to the open area of the
track lines without much obstruction to signals.

As a part of this project, the proposed novel crack detection
scheme has been tested by placing the robot on an actual rail
track. The testing was carried out at two different lighting
conditions: daytime (4.30PM) and night time (1.00 AM)
with irregular lighting along the railway lines, between
Chrompet and Pallavaram railway stations (Chennai, India),
covering 1 Kilometer along the track. The field trials gave
negative results for the presence of crack in the length of the
tested track-length due to the absence of cracks in the tested
area. These results were tested over and again and no false
outputs were obtained as the LED-LDR arrangement was
recalibrated before each startup.
However, calibration failure (identified by false output
before the testing stage starts) was found to have occurred in
the ratio of 1:25 which could be rectified by resetting the
entire setup.
In order to test the functionality of the crack detection
system as well as the GPS and GSM modules, a mechanical
arrangement to simulate an actual crack was created and the
system was found to accurately detect the presence of it and
the GSM module successfully transmitted the current coordinates obtained from GPS. The accuracy of the GPS
system was tested by comparing the obtained co-ordinate
locations using google maps. Figure 5 shows an SMS sent by

FIGURE 5. COORDINATES OF DETECTED CRACK RECEIVED AS SMS

481

ICRTIT-2012

[14] http://www.fra.dot.gov/downloads/Research/ord0416.pdf,
“A
comparison between different crack detection techniques”
[15] http://en.wikipedia.org/wiki/Rail_inspection, “Basic tehniques of rail
crack detection”
[16] http://en.wikipedia.org/wiki/Transport, “Importance of Transport”
[17] http://en.wikipedia.org/wiki/Indian_Railways, “Statistics about Indian
Railway accidents”

VIII. CONCLUSION
In this paper, we have presented the rationale, design of
our robust LED-LDR based railway crack detection scheme.
The authors hope that their idea can be implemented in large
scale in the long run to facilitate better safety standards for
rail tracks and provide effective testing infrastructure for
achieving better results in the future.
IX.

ACKNOWLEDGEMENT

The authors wish to thank the IEEE Bangalore Section for
having provided funding for this project. The All India
Young Engineers’ Humanitarian Challenge 2011 is also
being thanked for providing the required platform to carry
out the project.
REFERENCES
[1]

[2]

[3]
[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

Qiao Jian-hua; Li Lin-sheng; Zhang Jing-gang; “Design of Rail
Surface Crack-detecting System Based on Linear CCD Sensor”, IEEE
Int. Conf. on Networking, Sensing and Control, 2008
K. Vijayakumar, S.R. Wylie, J. D. Cullen, C.C. Wright, A.I. AIShamma’a, “Non invasive rail track detection system using
Microwave sensor”, Journal of App. Phy., 2009
Tranverse crack detection in rail head using low frequency eddy
currents, Patent US6768298, www.google.com/patents/US6768298
M. Cacciola, G. Megali, D. Pellicanµo, S. Calcagno, M. Versaci, and
F. C. Morabito, "Rotating Electromagnetic Field for Crack Detection
in Railway Tracks", PIERS ONLINE, Vol. 6, NO. 3, 2010
Wojnarowski, Robert John Welles, II, Kenneth Brakeley Kornrumpf,
William Paul, "Electromagnetic system for railroad track crack
detection and traction enhancement", Patent US6262573,
www.patentstorm.us/patents/6262573/description.html
Richard J. Greene, John R. Yates and Eann A. Patterson, "Crack
detection in rail using infrared methods", Opt. Eng. 46, 051013, May
2007
R.J. Greene, J.R. Yates,E.A. Patterson, "Rail Crack Detection: An
Infrared Approach to In-service Track Monitoring", SEM Annual
Conference & Exposition on Experimental and Applied Mechanics,
2006
Hartman, G.A., Infrared Damage Detection System (IDDS) for realtime, small-scale damage monitoring, Proc. SEM Ann. Conf. on Exptl
Mech., Charlotte, North Carolina (2003)
Stuart B Palmer, Steve Dixon, Rachel S Edwards and Xiaoming Jian,
"Transverse and longitudinal crack detection in the head of rail tracks
using Rayleigh wave-like wideband guided ultrasonic wave", Centre
for Materials Science and Engineering The University of
Edinburgh,www.cmse.ed.ac.uk/AdvMat45/Rail-crack-detection.pdf
Thomas Heckel, Hans-Martin Thomas, Marc Kreutzbruck and Sven
Ruhe, "High Speed Non-destructive Rail Testing with Advanced
Ultrasound and Eddy-Current Testing Techniques", NDTIP
Proceedings, Prague, 2009
Lanza di Scalea, F., Rizzo, P., Coccia, S., Bartoli, I., Fateh, M., Viola,
E. and Pascale, G., “Non-contact ultrasonic inspection of rails and
signal processing for automatic defect detection and classification,
Insight – NDT and condition monitoring”, Special Issue on NDT of
Rails 47(6) 346-353 (2005)
Spencer Ackers, Ronald Evans, Timothy Johnson, Harold Kess,
Jonathan White, Douglas E Adams, Pam Brown, "Crack detection in
a wheel end spindle using wave propagation via modal impacts and
piezo actuation", Health Monitoring and Smart Nondestructive
Evaluation of Structural and Biological systems V, SPIE (2006)
http://www.tc.gc.ca/media/documents/railsafety/technologies.pdf,
“Railway Safety Technologies”

482

ICRTIT-2012

19.robust railway crack detection scheme (rrcds)

More Related Content

What's hot (20)

Viewers also liked (7)

Similar to 19.robust railway crack detection scheme (rrcds) (20)

Recently uploaded (20)

19.robust railway crack detection scheme (rrcds)