Industrial Robot: An International Journal

Development of an autonomous in-pipe robot for offshore pipeline maintenance

Zhongwei Wang, Qixin Cao, Nan Luan, Lei Zhang,
Article information:
To cite this document:
Zhongwei Wang, Qixin Cao, Nan Luan, Lei Zhang, (2010) "Development of an autonomous in‐pipe robot for
offshore pipeline maintenance", Industrial Robot: An International Journal, Vol. 37 Issue: 2, pp.177-184, https://
doi.org/10.1108/01439911011018957
Permanent link to this document:
https://doi.org/10.1108/01439911011018957
Downloaded on: 15 July 2017, At: 00:15 (PT)
References: this document contains references to 15 other documents.
To copy this document: permissions@emeraldinsight.com
Downloaded by Universiti Teknikal Malaysia Melaka At 00:15 15 July 2017 (PT)

The fulltext of this document has been downloaded 963 times since 2010*
Users who downloaded this article also downloaded:
(2007),"In-pipe robot for inspection and sampling tasks", Industrial Robot: An International Journal, Vol. 34 Iss 1 pp. 39-45 <a
href="https://doi.org/10.1108/01439910710718432">https://doi.org/10.1108/01439910710718432</a>
(2010),"Development of an industrial pipeline inspection robot", Industrial Robot: An International Journal, Vol. 37 Iss 3 pp.
309-322 <a href="https://doi.org/10.1108/01439911011037721">https://doi.org/10.1108/01439911011037721</a>

Access to this document was granted through an Emerald subscription provided by emerald-srm:534729 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service
information about how to choose which publication to write for and submission guidelines are available for all. Please visit
www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of
more than 290 journals and over 2,350 books and book series volumes, as well as providing an extensive range of online
products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics
(COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation.

*Related content and download information correct at time of download.

 Research article

Development of an autonomous in-pipe

robot for offshore pipeline maintenance
Zhongwei Wang
Research Institute of Robotics, Shanghai Jiao Tong University, Shanghai, China, and
Qixin Cao, Nan Luan and Lei Zhang
Institute of Welding Engineering, School of Materials Science and Engineering, Shanghai Jiao Tong University, Shanghai, China

Abstract
Purpose – The purpose of this paper is to develop a novel autonomous in-pipe robot to perform the preventive point reparation for long-distance
Downloaded by Universiti Teknikal Malaysia Melaka At 00:15 15 July 2017 (PT)

offshore oil pipelines.

Design/methodology/approach – The autonomous in-pipe robot performs online ultrasonic inspection for pipe wall thickness, and the original
inspection data are stored in large capacity hard disk. Through the offline data analysis by the data analysts and the software tool, the pipeline health
status is known. If server defects lie there, the in-pipe robot is introduced into the pipeline once more to indicate the defect’s location to the
maintenance ship.
Findings – The laboratory tests and the field tests prove the feasibility and validity of the developed autonomous in-pipe robot. Furthermore, the
application of intelligent control techniques ensures the mission completion by the autonomous in-pipe robot, which worked in the awful pipeline
environment.
Practical implications – The developed autonomous in-pipe robot helps eliminate lost production costs and pipeline downtime caused by leakages
and guarantees the safe run of offshore oil pipelines.
Originality/value – For the application of the autonomous in-pipe robot, there are no special requirements for maintained pipelines themselves, so it
is applicable to the point reparation for most long-distance welded offshore pipelines.

Keywords Pipelines, Maintenance, Robotics, Ultrasonic devices, Inspection, Intelligent sensors

Paper type Research paper

1. Introduction Lin and Zhang, 2006). The system works based on detecting
an abrupt break-induced pressure wave. The pressure wave
Recently, offshore oil pipelines are the most energy-efficient, travels in both directions away from the break point and is
safe, environmentally friendly, and economic way to transport reflected at the pipeline boundaries. Using the pressure wave,
crude oil over long distance from an oil well/oil terminal in the data measured at one location along the pipeline, the time of
sea to an refinery/oil storage device on land. However, for the the initial and reflected transient waves induced by the break
long-term erosion of transported materials and environmental determines the location of the break. The pipeline leak
effects, flaws can appear and grow on the walls of pipelines. detection system performs a continuous monitoring for
As a result, there exist potential pipeline leakages. Any failure pipeline operation and helps to find pipeline leakage in time
of these offshore pipelines will not only affect productivity to minimize the losses, but it cannot prevent and avoid
negatively, but also cause tremendous environmental hazards. pipeline leak. Besides, localization error of the leakage point is
So, this issue has motivated the development of effective usually several hundred meters, and it brings tremendous
methods for leak detection and pipeline maintenance. difficulty to implement the offshore pipelines maintenance
In the reviewed work, the pipeline leak detection systems were (Feng and Zhang, 2005; Lin and Zhang, 2006). To
developed and successfully applied to detecting and locating implement the preventive pipeline maintenance and then
the breaks in transport pipelines ( Journal of Offshore Technology, completely avoid leakage losses caused by metal corrosion, the
2004; Henrique et al., 2005; Feng and Zhang, 2005; pipeline inspection devices, commonly referred to the
intelligent pig, were developed to implement the regular
The current issue and full text archive of this journal is available at pipeline inspection (Okamoto et al., 1999; Reber et al., 2002;
www.emeraldinsight.com/0143-991X.htm Yu et al., 2005; Lopez and Sadovnychiy, 2007). The devices
run through all the length of the pipelines to detect potential

Industrial Robot: An International Journal

37/2 (2010) 177– 184 This work was supported by National “863” High Technology Research
q Emerald Group Publishing Limited [ISSN 0143-991X] and Development Program of China (2001AA602021) and partially by
[DOI 10.1108/01439911011018957] Natural Science Foundation of China (50705054).

177
 Development of an autonomous in-pipe robot Industrial Robot: An International Journal
Zhongwei Wang, Qixin Cao, Nan Luan and Lei Zhang Volume 37 · Number 2 · 2010 · 177 –184

leaks and their location coordinates. According to the also used as the odometer wheels to measure the traveled
detection results, the point reparations for those severe distance of the in-pipe robot inside the pipeline. Inside the
defects are performed, which is a relatively cost-saving and sealed cabins are electric crawler controller, power supply
easily implemented maintenance method. To adopt the which includes four cabins, intelligent controller, extra low
maintenance method, the necessary precondition is that the frequency (ELF) emitter, data-processing module, and
detected pipelines must be equipped with magnetic markers ultrasonic inspection module.
every certain distance and their coordinate information is The electric connection among all units of the in-pipe robot
known. Correspondingly, the pipeline inspection device is is shown in Figure 3. The nine sealed cabins are
equipped with special magnetometer and strap-down inertial interconnected by two power cables for 24 V DC supply
navigation system (SINS). When performing pipeline and two signal cables for controller area network (CAN) bus.
inspection, the magnetometer can reliably detect magnetic In addition, electric crawler controller supplies the driving
markers along the pipelines and their known coordinate data current for the electric crawler and the ultrasonic inspection
are used for the correction of the SINS measurement. module transmits sensor data to the data-processing module
Consequently, the pipeline inspection device achieves the with multi-cable connection. For the waterproof protection,
pipeline inspection and also acquires the precise location the electric connection wires are all enclosed in sealed
coordinates of those detected corrosion points (Yu et al., 2005; hosepipes. Furthermore, the detail about the in-pipe robot
Lopez and Sadovnychiy, 2007). However, most offshore modules will be introduced in the following text.
pipelines are not equipped with magnetic markers because
Downloaded by Universiti Teknikal Malaysia Melaka At 00:15 15 July 2017 (PT)

when laying pipelines, the business of maintenance in 2.1 Electric crawler and electric crawler controller
the future was not adequately taken into consideration. The electric crawler and the electric crawler controller are
Therefore, the maintenance method cannot apply to most traction devices for the in-pipe robot moving through the
offshore pipelines. inspected pipeline. The electric crawler, shown in Figure 4,
Considering that most offshore oil pipelines are welded from has six wheeled driving arms fixed circumferentially with 608
fixed-length steel pipe segments and the girth welds along the apart on the outside of body frame. The driving wheels are
pipeline can roughly indicate the locations relative to the located on the end of driving arms and there are motors built
starting point of the pipeline, A novel approach for precisely in the driving arms. The motors control the wheels,
locating the corrosion inside the pipelines is proposed, which respectively, to form the driving mode of six independent
first locates the problematic pipe segment by eddy current driving wheels. The electric crawler features with compact
sensor detecting girth welds, and then determines the distance structure, great transmission efficiency and big tractive forces,
from the forward girth weld of the problematic pipe segment to and provides the moving speed from 50 to 240 mm/s for the
the corrosion point by multi-odometer. Based on the in-pipe robot. The electric crawler controller receives control
localization method, an autonomous in-pipe robot is instructions such as start-stop, moving direction, moving
developed to detect the potential leaks in long-distance speed, etc. from intelligent controller and supplies the
offshore pipelines and implement point reparation for them. corresponding driving current for the electric crawler.
This paper presents in detail the design and implementation
issues related to the autonomous in-pipe robot. Also, intelligent 2.2 Power supply
control techniques are applied to ensuring the mission The power supply module is fitted with Li batteries and
completion by the autonomous in-pipe robot, since it works in provides 24 V, the maximum 30 A and over 40 h DC for the
the awful pipeline environment (Figure 1). in-pipe robot system. The Li batteries are rechargeable for 20
times approximately and have an available period of two years.
2. General description of the autonomous in-pipe The power supply module monitors the battery energy
utilization and the residual capacity, and feedbacks the battery
robot working status to intelligent controller as the basis for its
The autonomous in-pipe robot is shown in Figure 2 and an decision making.
overview of the design specifications is provided in Table I.
The in-pipe robot’s mechanical structure is composed of an 2.3 Intelligent controller
electric crawler and nine cylindrical sealed cabins, and they The intelligent controller is the control center of the
are all connected together via hook joints, which allow the in- autonomous in-pipe robot. We adopted an embedded pc104
pipe robot to negotiate 3D bends. The sealed cabins are industrial computer as the core to meet the need of complex
supported by tensioner wheels, which not only keep them control calculation. To facilitate compiling program code and
centered in the pipe but also adapt to pipe diameter debugging them, the development of the intelligent control
deformation in a certain extent. Three tensioner wheels are software is based on an embedded Linux system, which runs
on the pc104 computer. Based on CAN bus, the intelligent
Figure 1 The schematic of precisely locating the corrosion inside the controller sends control instructions to other modules and
receives feedback messages from them, too. The feedbacks
offshore pipelines
include the control instruction confirmation, the modules’
Locateing the problematic pipe segment by Determining the distance The problematic working status and the sensor data. Before performing
eddy current sensor detecting girth welds by multi-odometer pipe segment
the pipeline maintenance, the mission objective and
intelligent control algorithms are download into the
intelligent controller, and then based on the receiving sensor
data, it makes decisions to achieve the mission. Besides, when
The starting point The girth weld along The corrosion point to encountering an emergency, such as power supply failure and
the pipeline be maintained electric crawler malfunction, the instruction of emitting

178
 Development of an autonomous in-pipe robot Industrial Robot: An International Journal
Zhongwei Wang, Qixin Cao, Nan Luan and Lei Zhang Volume 37 · Number 2 · 2010 · 177 –184

Figure 2 Outside view of the autonomous in-pipe robot

Experimental pipeline Power supply Data-processing Ultrasonic

system module inspection module

Electric crawler Electric crawler controller Hook joint Intelligent controller ELF emitter
Downloaded by Universiti Teknikal Malaysia Melaka At 00:15 15 July 2017 (PT)

The characteristic meets the requirements in this

Table I The main design specifications of the autonomous in-pipe
application, so the in-pipe robot is equipped with an ELF
robot
emitter to indicate its location to the exterior – maintenance
The main design specifications Data ship. The effective communication distance between the ELF
emitter and the ELF signal receiver located on the
Number of bodies 10
maintenance ship is 11 m, approximately. The sealed cabin
Length Approx. 5,670 mm
for the ELF emitter, shown in Figure 1, also contains batteries
Diameter 320 mm
as emergency power for emitting emergency signal in case of
Weight Approx. 300 kg power supply failure.
Moving speed range From 50 to 240 mm/s
Pipeline inspection distance Approx. 20 km at the speed
2.5 Ultrasonic inspection module and data-processing
range of 150 mm/s
module
Minimum bend radius 3D
The ultrasonic inspection module contains 64 channel
Permitted pipeline deformation 2 percent of the pipeline diameter
ultrasonic sensors to detect pipe wall thickness of the
pipeline. As shown in Figure 5, the ultrasonic sensors are
divided into two circles with 32 sensors for each circle, and
emergency signal is send to the ELF emitter module to get the two circles of ultrasonic sensors are regularly staggered to
rescue for the in-pipe robot. ensure full circumferential coverage of the pipe. The
ultrasonic sensors work in a pulse-echo mode with a
2.4 ELF emitter selected center frequency. The acoustic wave, emitted by
The early offshore pipeline inspection devices used isotopes the ultrasonic sensors, travels in the liquid medium and is
for tracking and locating the equipments. The restricted use reflected at reaching the inner- and outer-wall of the pipe.
and combined operational limitations of isotopes led to the The time of flight between the echoes from the inner- and
development of a new tracing method – the ELF signal (ELF outer-wall is related to the pipe thickness and can be used to
wave) for its better penetration capability (da Silva et al., evaluate the metal corrosion of the pipeline. The center
2004; Chen et al., 2006). Experiment showed the ELF signal frequency of the ultrasonic sensor relates to its detection
has the characteristic of penetrating through metal pipe, resolution and detection distance. The greater the ultrasonic
subset sediments and seawater (Chen et al., 2006). sensor frequency is, the shorter the wavelength is and

Figure 3 Schematic diagram for electric connection among the in-pipe robot units
Multi-cable
Intelligent Data-processing Ultrasonic
Electric crawler Power supply ELF emitter
controller module inspection module

Multi-cable

Electric crawler
controller

2 cables Can bus

2 cables 24V direct current

179
 Development of an autonomous in-pipe robot Industrial Robot: An International Journal
Zhongwei Wang, Qixin Cao, Nan Luan and Lei Zhang Volume 37 · Number 2 · 2010 · 177 –184

Figure 4 Mechanical model of the electric crawler extract their features during the pipeline inspection, and only
the defects’ feature data and their location data are stored for
The driving arm The driving wheel performing the subsequent defect reparation (Okamoto et al.,
1999; Reber et al., 2002). For this project, based on the
digital signal processor þ field-programmable gate array
techniques, the high-speed data acquisition system is
developed to AD-convert and compress the total inspection
data, and then they are stored in large capacity hard disk.
Later, the stored original inspection data can be used for the
offline pipeline evaluation to investigate the pipeline’s
whole health status rather than the local defects.
Besides, since the professional data analysts can participate
in the offline evaluation work, so it is more precise and
reliable than the automatic online evaluation mentioned
above.

3. Operation process for the offshore pipeline

maintenance
Downloaded by Universiti Teknikal Malaysia Melaka At 00:15 15 July 2017 (PT)

Figure 5 Actual photograph of the inspection sensor module The entire operation process for performing the point
Ultrasonic sensor
reparation in offshore pipelines with the in-pipe robot
includes four steps: preparations for the in-pipe inspection,
online pipeline inspection, offline data analysis and point
reparation for the defect, and the details of the operation
process are shown in Figure 7.
As the preparation work for the in-pipe inspection, the
cleaning operations for pipeline internal surface are
performed with cleaning pigs, which is necessary to obtain
good quality inspection data during the online pipeline
inspection; then the geometry inspection is performed to
measure the pipeline’s cross-section by applying a caliper tool,
which ensures the safe passage through the long-distance
pipeline by the in-pipe robot. For the online pipeline
therefore the better is the detection resolution. However, the inspection, the autonomous in-pipe robot passes through
acoustic attenuation in the liquid medium exponentially the whole pipeline at the speed of 150 mm/s. The pipe wall
increases with the frequency increase. So, as a compromise, thickness is detected by the ultrasonic sensors, and the
5 MHz ultrasonic sensor frequency is selected as the solution location is synchronously got by the eddy current sensor
for this application. Then, the ultrasonic sensor equipped on detecting girth welds along the pipeline and the multi-
the inspection sensor module has a beam width of 1.8 mm odometer detecting the precise position between adjacent
and its detection distance is 40 mm. girth welds. The original inspection data are stored for the
The ultrasonic inspection module also contains 1 eddy subsequent offline data analysis. After the in-pipe inspection,
current sensor to detect the girth welds along the pipeline, the inspection data are retrieved from the in-pipe robot for the
whose induction coil, as shown in Figure 6, is located on the offline data analysis. The software tool is developed to
cross-section of the sealed cabin. Based on the sensitivity to automatically analyze the ultrasonic data to obtain defect
different electromagnetism characteristics between girth weld candidates, which reduces the number of indications to be
material and steel pipe material, the eddy current sensor is manually checked. As shown in Figure 8, the software tool not
utilized to detect girth welds. only visualizes multi-ultrasonic sensor data in 2D image with
The data-processing module compresses and stores the the pseudo color, which uses a dominant color for defect
sensor data from the ultrasonic inspection module. Usually, areas, but it also displays A-scan view of any channel for
the pipeline inspection devices online evaluate the defects and accessing the relevant portions of the defect. For better

Figure 6 The layout of the eddy current sensor’s induction coil in the sealed cabin

The cross section of The eddy current sensor is

detected pipeline enclosed in the sealed cabin
for the ultrasonic inspection
module

The cross section of

the sealed cabin

180
 Development of an autonomous in-pipe robot Industrial Robot: An International Journal
Zhongwei Wang, Qixin Cao, Nan Luan and Lei Zhang Volume 37 · Number 2 · 2010 · 177 –184

Figure 7 Schematic of the operation process for offshore pipeline maintenance

64 channel
The pipe wall thickness
ultrasonic sensors
The total pipeline
Pipeline location information inspection
Cleaning pipeline Geometry The offshore
data being
internal surface inspection pipeline Girth welds Eddy current sensor stored
in hard disk
The position between
Multi-odometer
adjacent girth welds
Preparations for the inspection Online pipeline inspection

The software tool Eddy current sensor Girth welds

The report recording
ULF Maintenance
the severe defects and The position between
Multi-odometer emitter ship
their locations
Downloaded by Universiti Teknikal Malaysia Melaka At 00:15 15 July 2017 (PT)

adjacent girth welds

The data analysts
Locate the reparation point
Offline data analysis Point reparation for the defect

Figure 8 The screenshot of the software tool for inspection data analysis

A-Scan view of any The defects

channel feature display

Functional areas &

status display

Magnified partial
C-Scan image

C-Scan view of
pipe wall thickness

accuracy, data analysts can check the feature of any defect Thus, the preventive point reparation for offshore pipelines is
candidate, such as its length, width, depth, and severity grade, achieved.
which helps to the identification of the defects. When the
offline data analysis is finished, the report of recording the 4. The intelligent control techniques applied to
severe defects and their location is got. To implement point the in-pipe robot
reparation for the severe defect, their location data
are downloaded into the intelligent controller and then the The autonomous in-pipe robot runs in the awful working
in-pipe robot is introduced into the pipeline again. By the environment of oil pipelines, where the detailed environment
information is unknown even though the geometry inspection
eddy current sensor detecting the girth welds passed by and
is performed in advance as described in section 3. After being
counting them, the autonomous in-pipe robot first locates the
introduced into the pipeline, the in-pipe robot runs
problematic pipe segment, and then by multi-odometer, it autonomously and is no longer under the control of us.
determines the distance from the forward girth weld of the Some problems were exposed in the laboratory tests and field
problematic pipe segment to the reparation point. When tests, which were induced by the awful pipeline environment
arriving at the reparation point, the in-pipe robot stops to and severely affected the in-pipe robot’s usability. So it is
emit the ELF signal and the maintenance ship will receive it necessary for the autonomous in-pipe robot to be provided
and therefore locate the reparation point. The reparation for with the adaptability and robustness to the interferential
the defect is done by the maintenance ship, and working environment, and the following intelligent control
meanwhile, the in-pipe robot is taken out from the pipeline. techniques are applied to achieve the purpose.

181
 Development of an autonomous in-pipe robot Industrial Robot: An International Journal
Zhongwei Wang, Qixin Cao, Nan Luan and Lei Zhang Volume 37 · Number 2 · 2010 · 177 –184

4.1 The intelligent localization technique .

when the in-pipe robot moves L, that is, it has not yet
During performing the point reparation for the defect, to passed the next girth weld, the measured value by the
locate the problematic pipe segment depends on detecting the multi-odometer must be in the interval (aL, L); and
girth welds passed by. Therefore, whether the girth welds can .
when the in-pipe robot moves L/a, that is, it has already
be reliably detected determines the success or failure of the passed the next, the measured value must be in the
point reparation for the offshore pipeline. Plenty of interval (L, L/a).
underwater tests for detecting the girth weld by the eddy
current sensor were dong, as shown in Figure 9, that when the Therefore, when the measured value has passed the interval
sealed cabin of containing the eddy current sensor passed a ðaL; LÞ < ðL; L=aÞ ¼ ðaL; L=aÞ; one girth weld is traversed by
girth weld, the gathered signal was converted into one pulse the in-pipe robot for certain, and if no girth weld is detected,
by converting circuit. And the test results indicate when the then we infer one girth weld has been missed and the fault-
moving speed of the eddy current sensor, relative to the girth tolerance is performed to compensate the missed. With the
weld, is more than the threshold – 50 mm/s the girth welds aforesaid method, we can perform fault-tolerance processing
can be inerrably detected. However, the jounce movement of for successive multiple undetected girth welds.
the in-pipe robot and the abrupt drop of its moving speed, Generally, odometers are applied to the localization of
caused by the hindrance of welding beadings at girth welds, automatic inspection devices inside the pipelines. Owing to
may induce the instantaneous speed to be less than the the cumulative error effect, the longer the pipeline is, the
threshold. Therefore, the detection ability of the eddy current lower the localization accuracy is. In this application, the girth
Downloaded by Universiti Teknikal Malaysia Melaka At 00:15 15 July 2017 (PT)

sensor is depressed and some girth welds may be undetected. welds along the pipeline are used as calibration messages for
Consequently, fault-tolerance processing for those undetected every L distance to insure high-localization accuracy by the
girth welds is performed to ensure correctly locating the odometers. That is, at detecting the girth weld, the register of
problematic pipe segment by the in-pipe robot. recording the movement distance resets and anew records. So
The fault-tolerance processing is based on the fact that after the odometers are applied to the localization between adjacent
the in-pipe robot has moved a certain distance, there is one girth welds in most time and the fault-tolerance processing for
girth weld to be detected, so if not detecting girth weld, the missed girth weld sometimes. For the point reparation of
the judgment of one girth weld missed can be made and offshore pipelines, the more precise to locate the defect is, the
the counter of recording the girth weld passed by will be easier the maintenance work is. To locate the defect precisely,
purposively modified to compensate the missed one. Here, we
three odometers are applied to obtaining the movement
use production rule to represent the aforementioned fault-
distance and the multi-sensor data fusion is adopted to merge
tolerance idea and the proposed rule is given as the following:
the redundant movement distance data into more precise and
IF , Measured value by multi – odometer having passed reliable one. The odometers are easily disturbed in the awful
the interval ðaL; L=aÞ . pipeline environment, so the redundant movement distance
data may be uncertain or not in agreement. Problems will
AND , Without detecting girth weld . arise in data fusion if any of sensor data are uncertain or
THEN , One undetected girth weld occurs; the girth contains biggish inaccuracy. Here, based on the consistent
weld counter plus 1 . data fusion algorithm (Luo et al., 1988; Mintz, 1990), the
dubious sensor data are first eliminated before data mergence,
In the production rule, L is the length of steel pipe segments so the rest are consistent or in agreement. Then the precise
that offshore pipelines are made of and a (a [ (0, 1)) is the localization inside the pipelines is achieved by merging the
reliability coefficient of the multi-odometer. The reliability consistent data with the maximum likelihood estimate
coefficient means when the in-pipe robot moves L, the algorithm (Luo et al., 1988). Figure 10 is the schematic of
measured value is at least aL. Consequently, the following data fusion for the multi-odometer and the eddy current
inferences are made for the case that one girth weld is sensor to achieve intelligent precise localization of the in-pipe
detected and the next will be missed: robot inside the pipelines.

Figure 9 The underwater test of detecting girth welds

The girth weld

182
 Development of an autonomous in-pipe robot Industrial Robot: An International Journal
Zhongwei Wang, Qixin Cao, Nan Luan and Lei Zhang Volume 37 · Number 2 · 2010 · 177 –184

Figure 10 The schematic of multi-sensor data fusion for the intelligent precise localization

Consistent data Maximum likelihood Locating the defect in the

Movement distance Multi-odometer
fusion estimate problematic pipe segment

Calibration message

eddy current Determining the problematic

Girth welds Production rule
sensor pipe segment

4.2 Reactive self-rescue control for the autonomous pipeline wall thickness was less than 0.5 mm and the defects
in-pipe robot of corrosion area being more than 10 £ 10 mm2 could be
The specially designed electric crawler, introduced in reliably detected. Figure 11 shows the artificial defects on the
Section 2, can provide sufficient tractive forces for the experimental pipeline and their C-scan images for feature
in-pipe robot system; however, there still occurred the cases analysis.
that it encountered the impassability inside the pipeline for The field tests, shown in Figure 12, were performed in a
unclear reason. When the emergency happens, the 3.5 km-length beach sea oil pipeline at Shengli Oilfield in
Downloaded by Universiti Teknikal Malaysia Melaka At 00:15 15 July 2017 (PT)

straightforward rescue method is to inform the maintenance Shandong Province, China. The test result was that by the
ship by emitting ELF signals, and then the rescue activities is intelligent localization technique, the in-pipe robot could
performed to locate the in-pipe robot, cut-off the pipeline and reliably locate the repaired defect with the maximal
get it back. The rescue method can availably avoid the localization error being less than 1 percent of the pipe
equipment loss, but the rescue activity is troublesome and segment. That is the maximal localization error approximates
meanwhile, it has to terminate the pipeline maintenance task. 20 cm when the maintained pipeline is made of 20 m length
It expends many resources to perform the maintenance work pipe segments. The location precision adequately meets the
with the in-pipe robot, and therefore the mission failure also engineering requirement for implementing point reparation.
means sever economic losses. Besides, the reactive self-rescue control technique helped to
In the application field of autonomous mobile robots, the passage through the pipeline by the in-pipe robot. In
researchers advocate that they should be provided with short, those tests showed the developed in-pipe robot was
autonomous learning ability, which is to learn from self- adequate to the preventive point reparation for the long-
exploration and observation without the need of examples or distance offshore pipeline.
a teacher (Gachet et al., 1994; Zalama et al., 2002). Thus, the
autonomous mobile robot will have certain adaptability for
unstructured environment and unknown emergency. For this 6. Conclusion and future work
project, based on Q-learning algorithm (Watkins and Dayan, The development of an autonomous in-pipe robot for
1992), the motion control strategy of guiding the in-pipe offshore oil pipeline maintenance was described in this
robot through the impassability is proposed. When the paper. The test results showed the storage of original
emergency occurs, instead of immediately emitting the ELF inspection data during the online inspection and the offline
signals for help, the autonomous in-pipe robot stops moving data analysis by data analysts combined with the software tool
forward and falls back a certain distance, then it moves
forward again at different speed to try to overcome the
impassability. The trial process is repetitively performed. Figure 11 The artificial defects outside a pipe segment and their
When one trial achieves good result that is, the in-pipe robot C-scan images
moves forward farther, the action is encouraged then, and the
repetition of such action tend to problem solving. The self-
rescue control skill of the autonomous in-pipe robot is got on
line through the interaction between it and the pipeline
environment, which helps to avoid the loss for the rescue
activity and the mission failure. When the self-rescue strategy
does not work, it is necessary to implement the mentioned
rescue activity for the in-pipe robot.

5. Experimental results
The in-pipe robot was tested both in the laboratory and in the Figure 12 The in-pipe robot being introduced into the pipeline for field
field. For the laboratory test, a 110 m length experimental tests
pipeline system was constructed which was similar to the
actual offshore transmission pipeline, and some artificial
defects were made outside a pipe segment to determine the
effectiveness of the ultrasonic inspection. The laboratory tests
were implemented, using water as couplant medium. Figure 2
shows the in-pipe robot being introduced in the experimental
pipeline. The test result was that the measurement error of

183
 Development of an autonomous in-pipe robot Industrial Robot: An International Journal
Zhongwei Wang, Qixin Cao, Nan Luan and Lei Zhang Volume 37 · Number 2 · 2010 · 177 –184

could achieve the reliable evaluation for pipelines’ health pipelines using a fuzzy system”, Journal of Petroleum Science
status. To implement the point reparation for the severe and Engineering, Vol. 49 Nos 3/4, pp. 223-38.
defect, the intelligent localization technique was proposed to Journal of Offshore Technology (2004), “Pipeline monitoring
achieve reliable and precise localization for the defect, which and control”, Journal of Offshore Technology, Vol. 12 No. 5,
was proven to meet the engineering requirement adequately. pp. 15-17.
Besides, the reactive self-rescue control technique was Lin, W.G. and Zhang, X. (2006), “A novel approach for
adopted for the in-pipe robot to overcome the impassability dynamic pressure transducer based pipeline leak detection”,
in the awful pipeline environment, which effectively avoided Proceedings of the 2006 World Congress on Intelligent Control
the loss for the rescue activity and the mission failure. For the and Automation, Dalian, China, June, pp. 5299-303.
application of the in-pipe robot, there are no special Lopez, J.M. and Sadovnychiy, S. (2007), “Small PIG for
requirements for maintained pipelines themselves, so it inspection pipeline”, Proceedings of the 2007 Electronics,
adapts to the point reparation of most welded offshore Robotics and Automotive Mechanics Conference, Cuernavaca,
pipelines. Nevertheless, more tests in longer offshore oil Morelos, Mexico, September, pp. 585-90.
pipelines will be conduced to verify the usability and reliability Luo, R.C., Lin, M.H. and Scherp, R.S. (1988), “Dynamic
of the in-pipe robot before its practical application. And we multi-sensor data fusion system for intelligent robots”, IEEE
will develop similar in-pipe robot for larger diameter offshore Journal of Robotics and Automation, Vol. 4 No. 4, pp. 238-41.
pipeline maintenance. Mintz, M. (1990), “Comments on ‘dynamic multi-Sensor
data fusion system for intelligent robots’”, IEEE Journal of
Robotics and Automation, Vol. 6 No. 1, pp. 104-6.
Downloaded by Universiti Teknikal Malaysia Melaka At 00:15 15 July 2017 (PT)

References Okamoto, J. Jr, Adamowski, J.O., Tsuzuki, J.C., Buiochi, M.G.

and Camerini, F. (1999), “Autonomous system for oil
Chen, H.J., Zhang, X.H. and Li, J.Y. (2006), “Ultra low
pipelines inspection”, Mechatronics, Vol. 9 No. 7, pp. 731-43.
frequency electromagnetic wave localization and application
Reber, K., Beller, M., Willems, H. and Barbian, O.A. (2002),
to pipeline robot”, Proceedings of the IEEE International “A new generation of ultrasonic in-line inspection tools for
Conference on Intelligent Robots and Systems, Beijing, China, detecting, sizing and locating metal loss and cracks in
October, pp. 1201-5. transmission pipelines”, Proceedings of the 2002 IEEE
da Silva, J.A.P., de Jesus Ribeiro Afonso, O., Dos Santos, J.A.C., Ultrasonics Symposium, Munich, Germany, October, pp. 665-71.
Matsuura, M., de Castro Antunes, R., de Freitas Santa Cruz, S. Watkins, C. and Dayan, P. (1992), “Q-learning”, Machine
and Hashiguchi, D.I. (2004), “A smart location system”, Learning, Vol. 8 Nos 3/4, pp. 279-92.
Proceedings of the Biennial International Pipeline Conference, Yu, J., Lee, J.G., Park, C.G. and Han, H.S. (2005), “An off-
Calgary, Canada, October, pp. 975-81. line navigation of a geometry PIG using a modified
Feng, J. and Zhang, H.G. (2005), “Diagnosis and localization nonlinear fixed-interval smoothing filter”, Control
of pipeline leak based on fuzzy decision-making method”, Engineering Practice, Vol. 13 No. 11, pp. 1403-11.
Acta Automatica Sinica, Vol. 31 No. 3, pp. 484-90. Zalama, E., Gomez, J., Paul, M. and Peran, J.R. (2002),
Gachet, D., Salichs, M.A., Moreno, L. and Pimentel, J.R. “Adaptive behavior navigation of a mobile robot”, IEEE
(1994), “Learning emergent tasks for an autonomous Transactions on Systems, Man, and Cybernetics Part A:
mobile robot”, Proceedings of the IEEE International Systems and Humans, Vol. 32 No. 1, pp. 160-9.
Conference on Intelligent Robots and Systems, Munich,
Germany, September, pp. 290-7.
Corresponding author
Henrique, V.S., Morooka, C.K., Guilherme, I.R., Tiago, C.F.
and Mendes, J.P. (2005), “Leak detection in petroleum Zhongwei Wang can be contacted at: wangzw@sjtu.edu.cn

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com

Or visit our web site for further details: www.emeraldinsight.com/reprints

184
 This article has been cited by:

1. Li Yang, Xiaobing Ma, Rui Peng, Qingqing Zhai, Yu Zhao. 2017. A preventive maintenance policy based on dependent two-
stage deterioration and external shocks. Reliability Engineering & System Safety 160, 201-211. [CrossRef]
2. Everson Brum Siqueira, Rodrigo Zelir Azzolin, Silvia Silva da Costa Botelho, Vinícius Menezes de OliveiraInside Pipe Inspection:
A Review Considering the Locomotion Systems 449-458. [CrossRef]
3. Michael Knowles, David Baglee, Pankaj SharmaApplication of Open Source Hardware to the Development of Autonomous
Maintenance Support Systems 333-347. [CrossRef]
4. Hong Zhao, Yi-Xin Zhao, Zhi-Hui Ye. 2016. Interactions between the fluid and an isolation tool in a pipe: laboratory experiments
and numerical simulation. Petroleum Science 13:4, 746-759. [CrossRef]
5. Everson Siqueira, Silvia Botelho, Rodrigo Azzolin, Vinicius OliveiraA review about robotic inspection considering the locomotion
systems and odometry 571-576. [CrossRef]
6. David Rollinson, Howie Choset. 2016. Pipe Network Locomotion with a Snake Robot. Journal of Field Robotics 33:3, 322-336.
[CrossRef]
7. Yi Lu, Jingjing Yu, Chunping Sui, Jianda HanDesign of in-pipe 3SPR/3RPS parallel manipulator and its kinestatics analysis
42-47. [CrossRef]
Downloaded by Universiti Teknikal Malaysia Melaka At 00:15 15 July 2017 (PT)

8. Muhammad Mysorewala, Muhammad Sabih, Lahouari Cheded, Mohammad Tariq Nasir, Muhammad Ismail. 2015. A Novel
Energy-Aware Approach for Locating Leaks in Water Pipeline Using a Wireless Sensor Network and Noisy Pressure Sensor
Data. International Journal of Distributed Sensor Networks 2015, 1-10. [CrossRef]
9. Jinwei Qiao, Jianzhong Shang. 2013. Application of axiomatic design method in in-pipe robot design. Robotics and Computer-
Integrated Manufacturing 29:4, 49-57. [CrossRef]
10. Florian Enner, David Rollinson, Howie ChosetMotion estimation of snake robots in straight pipes 5168-5173. [CrossRef]
11. Muhammad Imran, Mohamed A. Alnuem, Waleed Alsalih, Mohamed YounisA novel wireless sensor and actor network
framework for autonomous monitoring and maintenance of lifeline infrastructures 6484-6488. [CrossRef]
12. Jungwan Park, Dongjun Hyun, Woong-Hee Cho, Tae-Hyun Kim, Hyun-Seok Yang. 2011. Normal-Force Control for an In-
Pipe Robot According to the Inclination of Pipelines. IEEE Transactions on Industrial Electronics 58:12, 5304-5310. [CrossRef]