A Coordinated Gas Pipeline Robot Chain System Based On Wireless Relay Communication

System Based On Wireless Relay Communication

無線リレー通信技術に基づく協調的な ガス管ロボットチェインシステムの開発

February 2019

Waseda University

Graduate School of Creative Science and Engineering Department of Modern Mechanical Engineering,

Research on Intelligent Machines

Wen, ZHAO

チョウ ブン

With my deepest appreciation to those who made this dissertation possible.

At first, I would like to extend my most sincere gratitude to my Ph.D. supervisor Professor Dr. Shigeki Sugano–at Department of Modern Mechanical Engineering, School of Creative Science and Engineering of Waseda University– for his kindly guidance, encouragement, and the most important, for giving me the opportunities to join the distinguished research team. I also express my thankfulness to my reviewers Professor. Mitsuo Umezu, Professor. Jun Ohya and Professor. Hiroyasu Iwata for their contribution in improving this dissertation.

My supervisor Associate Professor Dr. Mitsuhiro Kamezaki for all discussions, constructive criticism, and inspiration that made this research original and focused. My team members, Kento Yoshida and Kaoru Yamaguchi for their hard work, efforts, and patience that made the practical evaluation of my research possible. To my colleges for challenging my ideas in our fruitful discussions. The lab secretaries Yoko Ono and Kyoko Arai for their support.

My friends Dr. Kui Chen, Dr. Udara from Sugano Lab accompany me throughout this wonderful three years.

The Japanese government-sponsored Monbukagakusho (MEXT) Scholarship, which made my study and research in Japan financially possible.

The Tokyo Gas Co. Ltd. which supplied the parts used in my experiments. Their cooperation make this research possible.

My beloved family for their love and support despite the long distance. To my father and mother for their constant encouragement to pursue the highest standards. To my friends in China for helping me focus my goals, and motivating me to continue working hard to achieve them.

Tokyo, 23 December 2018 Wen Zhao, Waseda

Pipeline networks especially the gas pipelines need to be inspected and maintained regularly since the leakage caused by destruction such as corrosion, deformation, and cracking is fatal to the safe operation and supply of gas. In recent year, various pipeline robots for pipe inspection and maintenance have been developed and promoted since they are more effective and low-cost compared with traditional destructive pipeline inspection methods. Besides, they are more precise and time-saving than conventional external non-destructive sensor network inspection technology.

In order to deal with inspection and maintenance missions safely and efficiently in the complex pipeline environment, it is important for the robots to realize the long distance real-time information interaction with external terminal and also the more intelligent operation for releasing more workload.

However, current pipeline inspection robots still meet several challenges or limitations in the pipe inspection task, which can mainly be described as two domains: environmental-factors and technical- factors. The environmental-factors are that the wireless signal attenuation limits effective in-pipe communication distance of wireless robot and further affects communication quality in the pipe. This result is due to the energy loss brought by the reflection and refraction of microwaves when are transmitted through the pipeline. Besides, same problems exist in the inspection method based on robot- sensor hybrid networks, such as WSN. Wireless signals are unable to pass through the thick soil layers or underground facilities. The technical-factors include firstly, the restrictions from communication cables for the wired robots due to the unexpected environment complexities (i.e., elbows, T-junctions) within the pipe. Secondly, the operation of robot rely heavily on manual operation with low efficiency and low precision. The failure of pipeline inspection even risk of further damage to robots and pipes may occur due to many human operation errors.

In order to deal with these factors, we have mainly done the researches on the two solutions:

operation method and communication method. In the former researches on the operation method, the most current pipeline robots which are heavily dependent on manual operation are unable to realize self- navigation in pipe. Other robots are capable of self-navigation based on multi-sensors detection and calculation. However, navigation performance will be significantly affected by the performance and accuracy of these sensors. The most important is that the self-navigation mode can reduce the workload and assist the robot to pass through some environment complexities more easily, whereas manual operation can help the operator to focus more on some key parts (i.e., pipe breakage, rust) when inspecting. The robot combining self-navigation and manual operation mode are urgently needed in the current robot-based pipeline inspection technology. For communication method technology, through

communication distance, poor communication quality, etc. Therefore a new method to optimize traditional communication method has been proposed in this research.

From the above questions and solutions we have described, we set the purpose of this research is to develop a more effective, automatic pipeline inspection system which can be operated with more reliable, stable, longer distance information interaction. Based on the above analysis, two new self- navigation methods are presented and implemented.

Firstly, an approach of pipeline robot’s self-navigation based on feature extraction and image processing techniques are described. Besides, the fuzzy logic control (FLC) algorithm is adopted. In this method, CCD is used to detect, locate and classify the region of interest (ROI) of the LED light reflection and pipe’s dark hole in elbow or T-junction. These ROIs were considered as the input variables in such FLC system. By reasoning on several rules about input-output logic relationship, the system can output the two important variables real-time yaw angle and velocity to support robot to realize self-navigation in the pipe. Compared with conventional studies on pipeline robot’s self-navigation method, the proposed approach could be more precise and faster with fewer external sensors. However, in this method, the detection, localization and classification of ROIs rely too much on the precision of CCD.

Besides, some other factors such as pipe diameter, different reflective properties on surface of pipe wall, metal rust and pipe dirt will influence the extraction and classification of ROIs. Finally, the interval classification of output variables of FLC algorithm is so intensive that require more precise motor control in real robot application. In the real robot experiment, it require further investigation on the application and implementation about such algorithm.

Secondly, we focus on another self-navigation approach based on ultrasonic detection and evaluation for elbow and T-junction of pipeline. In this robot design, we combine two operation modes together and adopt them in real pipeline environment test. The self-navigation method is based on ultrasonic evaluation and data processing from three ultrasonic sensors. The self-navigation mode can reduce the workload for operator however without manual operation we will lose some details of pipe inspection in some complex and unstructured pipeline environment. For these reasons, the combination of manual operation and self-navigation mode will increase the efficiency of inspection. Such method is also used in robot chain system (RCS) for cooperative and coordinated movement of system.

To realize the research objective, we need to investigate a reliable, stable, and long distance wireless communication method to overcome the challenges we discussed above. Finally we developed a pipeline inspection robot chain system (RCS) based on wireless relay communication (WRC) technology. The researches were carried out in the following three steps. The first is to do research on

system. The third is to develop a tracked pipeline robot chain system (RCS) based on wireless relay communication (WRC). Through several verification experiments, the results reveal that such system can effectively guarantee the reliable and long-distance wireless communication for pipeline inspection system.

Firstly, we started from wireless communication method to find a best solution to the problems mentioned above. For developing a wireless pipeline inspection system, it is preferable if we can know the transmission characteristics and model in pipe. Although we were sure about the wireless transmission model in the free-space, however, due to the complexity of pipe environment, the model was completely different from free-space transmission. There are multiple reflection and refraction of electromagnetic wave during transmission in the pipe. Through several experiments, we obtained the many transmission properties (i.e., transmission loss, (received signal strength indication (RSSI), maximum transmission distance, etc) in gas pipe. These parameters are crucial to determine which frequency or device are feasible to the further development. Besides, they are also important for the RSSI-based evaluation method for robot chain system.

Secondly, we proposed a concept of robot chain system based on wireless relay communication to overcome the communication challenges brought by the special pipeline environment mentioned previously. In order to ensure the feasibility, robustness and stability, we presented an approach of a reliable communication and localization method based on Received Signal Strength Indication (RSSI) theory for this robot chain. Without any GPS modules in the pipe, the robots are able to measure and calculate a certain distance from the adjacent one by detecting and estimating the received wireless signal strength. Besides, we described a “leader-follower” control method and localization and distance control of robot chain in pipe to support the RSSI-based communication. The leader robot is responsible for the task of whole pipe inspection with multi-sensors, and the follower robots act as the wireless

“signal relay node” for communication. Finally, such pipe robot chain cooperate and coordinate together, so that they are able to complete the inspection successfully. At last, we used virtual reality simulation to evaluate the performance of the proposed method. The simulation result indicated that the proposed method could realize the reliable wireless communication and localization for the robot chain system.

At last, through the above research on communication and operation methods, we proposed a prototype of a tracked robot chain system based on wireless relay communication to overcome the technical and environmental challenges as mentioned before. In this system, each robot serves as a “relay communication node”. Leakage information of pipes are transmitted through these “relay nodes”. To ensure the stability of relay communication between adjacent robots, we adopt RSSI (Received Signal Strength Indication)-based evaluation method for cooperative and coordinated movement of robot chain

system is capable of self-navigation and manual operation based on ultrasonic measurement module.

Finally, multiple experiments to evaluate relay communication quality, RSSI-based cooperative movement and test of RSSI and Link Quality Indication (LQI) were conducted. Results reveal that our proposed system could realize reliable, stable and long-distance relay wireless communication and basically increase the inspection range effectively with comprehensive performance.

As a conclusion, this proposed robot chain system had made a breakthrough not only in communication but also in operation method. With WALCP and RSSI-based evaluation technology, such system can realize more reliable, stable, long-distance wireless communication, which can assist the system to adapt to complicated environments of pipelines. With double operation modes, the system can release the workload of operator, reduce the risk of operation failure and finally increase the efficiency of inspection. This concept of research is not suitable for pipeline inspection robot, it also can be applied in special environments such as difficulties in using wired communication or with weak wireless signal.

C ONTENTS

1 I

... 1

1.1 Motivation ... 3

1.1.1 Environmental Restriction ... 3

1.1.2 Technical Bottleneck ... 3

1.1.3 The influence of these two factors ... 4

1.2 Objective ... 4

1.3 Thesis Structure ... 5

1.4 Chapter Summary... 6

2 B

R

... 7

2.1 Chapter Purpose ... 7

2.2 External Inspection System ... 8

2.2.1 Gas leakage detector based on gas leak ... 8

2.2.2 Gas leakage detector based on gas temperature ... 9

2.3 Internal Inspection system ... 10

2.3.1 Wheeled-type or tracked-type robot ... 10

2.3.2 Wall-press robot ... 11

2.3.3 Inchworm-type robot ... 12

2.3.4 Screw-type robot ... 12

2.4 Integrated Inspection System ... 12

2.5 Operation Method ... 13

2.5.1 Manual operation ... 13

2.5.2 Navigation based on passive self-adaption ... 14

2.5.3 Self-navigation based on vision ... 14

2.5.4 Self-navigation based on laser or infrared light ... 15

2.5.5 Discussion ... 17

2.6 Communication Method ... 17

2.6.1 Wired communication ... 17

2.6.2 Wireless communication ... 18

2.6.3 Discussion ... 19

3 S

-

A

... 21

3.1 Chapter Purpose ... 22

3.2 Robot Platform ... 23

3.2.1 Mechanical system ... 24

3.2.2 Drive and control system ... 24

3.2.3 Pipe inspection system ... 25

3.3 Self-navigation based on Feature Extraction & Image Processing ... 26

3.3.1 Landmark calculation and detection ... 26

3.3.2 Modeling of algorithm ... 29

3.3.3 Simulation ... 35

3.3.4 Results ... 36

3.4 Self-navigation based on Ultrasonic Detection ... 39

3.4.1 Architecture of robot ... 40

3.4.2 Self-navigation (automatic control) mode ... 41

3.4.3 Manual control mode... 41

3.4.4 Experimental design and results ... 42

3.5 Discussion ... 43

3.6 Chapter Summary ... 44

4 R

... 45

4.1 Chapter Purpose ... 46

4.2 RSSI Theory ... 46

4.2.1 Microwave transmission model in free-space ... 47

4.2.2 Microwave transmission model in pipe ... 50

4.3 Microwave Transmission Experiment ... 52

4.3.1 Measurement system ... 52

4.3.2 Transmission loss measurement ... 53

4.3.3 Experiments and results... 54

4.3.4 Discussion ... 58

4.4 Simulation based on RSSI Theory ... 59

4.4.1 RSSI-based reliable communication ... 60

4.4.2 Leader-follower-based robot chain ... 63

4.4.3 Simulation environment ... 67

4.4.4 Simulation results ... 68

4.5 RCS based on WRC ... 69

4.5.1 RSSI-based evaluation for robot chain system ... 71

4.5.4 Experiments and implementation results ... 79

4.6 Discussion ... 85

4.7 Chapter Summary... 86

5 C

& F

W

... 87

5.1 Research Achievements ... 88

5.2 Research Limitations ... 89

5.3 Future works ... 90

6 R

... 93

7 R

W

... 99

Patent ... 99

Conference Papers ... 99

Co-Authored Papers ... 100

8 A

... 101

CHAPTER 2

TABLE2-1SELF-NAVIGATION METHOD IN PIPELINE ... 40

TABLE2-2COMMUNICATION METHOD ... 43

CHAPTER 3

TABLE3-2FUZZY LOGIC RULE BASE ... 57

CHAPTER 4

TABLE4-2SCHEMATIC VIEW OF FRAME FORMAT ... 76

TABLE4-3ALGORITHM FOR RELAY COMMUNICATION ... 77

TABLE4-4ALGORITHM FOR RELAY COMMUNICATION ... 102

CHAPTER 5

CHAPTER 1

FIGURE 1-1DISTRIBUTION MAP OF GAS PIPELINE IN JAPAN ... 25

FIGURE 1-2THE GAS PIPELINE ACCIDENT ... 26

FIGURE 1-3OVERVIEW OF THIS RESEARCH AND THESIS ... 29

CHAPTER 2

FIGURE 2-2GAS LEAKAGE DETECTOR BASED ON GAS TEMPERATURE ... 9

FIGURE 2-3GAS PIPELINE ROBOT ... 10

FIGURE 2-4PIPELINE INSPECTION SYSTEM BASED ON WSN ... 13

FIGURE 2-5SELF-NAVIGATION BASED ON VISION ... 38

FIGURE 2-6SELF-NAVIGATION BASED ON LASER OR INFRARED LIGHT ... 39

CHAPTER 3

FIGURE 3-2 DIRECTION AND LOCATION OF LIGHT REFLECTION’S ROI IN ELBOW AND T- JUNCTION ... 50

FIGURE 3-4ROI DETECTION AND TRACKING OF ELBOW OR T-JUNCTION ... 28

FIGURE 3-5PARAMETERS OF LIGHT REFECTION AND YAW ANGLE OF PIPE ROBOT ... 29

FIGURE 3-6INPUT AND OUTPUT VARIABLES OF SYSTEM ... 30

FIGURE 3-7MEMBERSHIP OF INPUT VARIABLE ... 31

FIGURE 3-8MEMBERSHIP OF OUTPUT VARIABLE ... 56

FIGURE 3-9SIMULATION RESULT IN M-SHAPED PIPE ... 58

FIGURE 3-10THE OUTPUT VARIABLE OF RMSE, VELOCITY AND YAW ANGLE N M-SHAPED PIPE ... 59

FIGURE 3-11SIMULATION RESULT IN T-SHAPED PIPE ... 36

FIGURE 3-12 THE OUTPUT VARIABLE OF RMSE, VELOCITY AND YAW ANGLE N T-SHAPED PIPE ... 37

FIGURE 3-13ROBOT AND MOVEMENT IN THE PIPE ... 63

FIGURE 3-14M-SHAPED PIPELINE TEST FIELD ... 40

FIGURE 3-15OPERATION EFFECT IN TWO MODES ... 41

FIGURE 3-16THE FAILURE RATE IN TWO MODES... 66

CHAPTER 4

FIGURE 4-2MICROWAVE REFLECTION AND TRANSMISSION AT NORMAL OBLIQUE INCIDENCE BY PLANAR INTERFACE ... 72

FIGURE 4-3MICROWAVE MULTIPLE WAVE REFLECTION IN THE PIPE ... 73

FIGURE 4-4TRANSMITTER USED IN MEASUREMENT ... 75

FIGURE 4-6CALIBRATION OF MICROWAVE TRANSMITTER (A)5.8GHZ AND (B)2.4GHZ ... 78

FIGURE 4-7MEASUREMENT IN THE STEEL GAS PIPE ... 79

FIGURE 4-8TEST IN STRAIGHT STEEL GAS PIPE WITH 2.4GHZ AND 5.8GHZ ... 80

FIGURE 4-9TEST IN 5×200CM M-SHAPED STEEL GAS PIPE... 81

FIGURE 4-10TEST WITH ANTENNA GAIN ANDAND WITHOUT ANTENNA POWER GAIN ... 82

FIGURE 4-11LEADER-FOLLOWER-BASED ROBOT CHAIN FOR WIRELESS COMMUNICATION ... 84

FIGURE 4-12MODEL OF MICROWAVE REFLECTION AND REFRACTION IN GAS PIPE ... 85

FIGURE 4-13MOTION PLANNING OF A ROBOT CHAIN (FIVE 5 MEMBERS IN 15 M PIPE) ... 86

FIGURE 4-14CASUAL PROBLEM CAUSED BYRESULT FROM RSSI DETECTION FAILURE ... 87

FIGURE 4-15ROBOT CHAIN’S LOCALIZATION IN FOUR TYPICAL PIPELINE ... 88

FIGURE 4-16ROBOT CHAIN’S LOCALIZATION IN UNKNOWN PIPELINE ... 89

FIGURE 4-17SIMULATION RESULT WITH 〖RSSI〗_MEAN (DB)=88.34 ... 90

FIGURE 4-18 TRAVEL DISTANCE OF FIVE ROBOTS IN PIPE WITH 〖 RSSI 〗 _MEAN (DB)=88.34 ... 91

FIGURE 4-19SIMULATION RESULT WITH 〖RSSI〗_MEAN (DB)=71.42 ... 92

FIGURE 4-20TRAVEL DISTANCE OF NINE ROBOTS IN PIPE WITH 〖RSSI〗_MEAN (DB)=71.42 ... 93

FIGURE 4-21 SIMULATION RESULT IN THREE TYPES OF PIPE WITH 〖RSSI〗_MEAN (DB)=88.34 ... 94

FIGURE 4-22SMALL TRACKED PIPELINE ROBOT IN THE SYSTEM ... 95

FIGURE 4-23PROPOSED ROBOT CHAIN SYSTEM ... 97

FIGURE 4-24THE CONTROL FLOW CHART OF LEADER ROBOT ... 98

FIGURE 4-26RELAY COMMUNICATION TEST ... 103 FIGURE 4-27(A)(B)MOVEMENT TEST OF ROBOT CHAIN IN TRANSPARENT PIPE (C)MOVEMENT

TEST OF ROBOT CHAIN IN TRANSPARENT PIPE AND RSSI/LQI TEST IN PIPE .... 81 FIGURE 4-28TEST ON FRAME LOSS RATIO AND DATA ERROR RATIO ... 82 FIGURE 4-29RSSI OF THE FIRST LEADER AND THE FIRST FOLLOWER ROBOT ... 83 FIGURE 4-30(A)RSSI OF THE SECOND ROBOT (B)LQI OF THREE ROBOTS ... 84

CHAPTER 5

FIGURE 5-1 COMPREHENSIVE PERFORMANCE COMPARISON FROM FORMER PIPELINE ROBOTS

... 113 FIGURE 5-2EFFECT OF MATERIAL AND PIPE DIAMETER ON SYSTEM PERFORMANCE ... 90

RSSI → Received Signal Strength Indication RCS → Relay Communication System WRC → Wireless Relay Communication PWM → Pulse Width Modulation

WALCP → Wireless Application Layer Communication Protocol DTS → Distributed Temperature Sensing

WSN → Wireless Sensor Network SN → Sensor Node

RN → Relay Node BS → Base Station

IMU → Inertial Measurement Unit FLR → Frame Loss Ratio

DER → Data Error Ratio LQI → Link Quality Indication

1 I NTRODUCTION

The gas pipeline distribution network has a total length of roughly 200,000 km in the urban areas in Japan ad depicted in Figure 1-1. However, Japan is a country with frequent severe

natural disasters such as floods and earthquakes. With the natural disaster and the long-term Figure 1-1 Distribution map of gas pipeline in Japan [1]

usage of the pipeline, the leakage caused by pipe destruction such as corrosion, deformation, and cracking is fatal to the safe operation and supply of gas as shown in Figure 1-2 [1].

For the above reasons, it is necessary to know how long a segment of the pipeline network is still expected to offer the reliable service, and the most important is to find an effective way to realize the gas pipeline inspection. The inspection and maintenance of these pipelines are urgently required [1].

There are many different approaches which can be taken in order to deal with these problems. In recent years, various pipeline inspection and maintenance mobile systems have been developed and promoted since they are more effective and low-cost compared with traditional destructive pipeline inspection approaches. Destructive inspection is inefficient because the pipeline requires to be removed from the ground. Besides, the pipeline inspection system especially pipe robots are more precise and time-saving than conventional external non- destructive sensor network inspection technologies such as fiber optic distributed temperature sensing (DTS) system.

However, the inspection of the pipelines is still difficult since the gas pipes are generally buried underground and hard to access. Also the special complex pipeline environment such as rust, water, rupture, elbow, and T-junction will affect the performance of the detection system. Therefore development of a new type of inspection system is the focus of this research.

Figure 1-2 The gas pipeline accident [2]

1.1 Motivation

This section discuss the current challenges or limitations in pipeline inspection approaches. From former studies, the main challenge is how to effectively acquire information inside pipelines. Such challenge arises from two domains: technical bottleneck and environmental restriction. This research aims to propose an effective ways to deal with both of these challenges.

1.1.1 Environmental Restriction

Environmental restriction exists when the conventional pipeline inspection technology adopts microwave/ ultrasonic signal for information interaction. As for surface non-destructive pipeline damage inspection method and wireless pipeline robot inspection method mentioned before, the restriction mainly include:

 Wireless signal attenuation of wireless inspection system due to the fact that microwave can be refracted and absorbed by pipe wall

 Ultrasonic signal attenuation of surface detection equipment occur when signal passes through thick soil layers or underground facilities

1.1.2 Technical Bottleneck

The technical bottleneck is considered as another problem that seriously affects effective pipeline inspection technology. This bottleneck mainly exists in the wired or wireless pipeline robot inspection methods. It can be described in two aspects:

 Restrictions from communication cables due to the unexpected environmental complexities within the pipe, such as some special features: elbow, T-junction, Y-miter, etc.

 Failure of pipeline inspection due to human operation error, risk of pipeline’s deformation, leakage, cracking and even damage of inspection machine caused by human wrong operation, low efficiency and low precision brought by manual control, etc.

1.1.3 The influence of these two factors

The environmental restriction and technical bottleneck will influence the performance of the inspection system. Through analyzing these existing pipeline robots or systems, the main problems can be highlighted as below:

 Because of the communication cable’s weight and friction with the pipe, most wired pipe inspection robots own difficulties to travel in long distance as they will be inevitably restricted by the long communication cable.

 The pipeline inpection robots based on wireless communication technology can still be used in the large diameter pipes since the huge wireless radio transmission attenuation can happen in such small diameter pipes. Such signal attenuation can block the

communication between pipe robot and operation terminal.

 The pipeline inspection method based on sensor relay nodes has illustrated reliable leakage inspection. The signal could be obtained in real time with several sensor relay nodes, however it can only be applied to the shallow pipelines. Actually, most pipes are deeply buried in order to avoid underground facilities. So the wireless signals are unable to pass through the thick soil layers and underground facilities.

 Most pipe inspection robots are usually remotely operated by human. Pipe network is almost installed within the plants or underground, therefore self-navigation of a pipe inspection robot could reduce operational complexity.

1.2 Objective

The objective of this research is, to develop a more effective, intelligent inspection system which can be applied in deep buried pipeline at much longer inspection distance.

In order to realize this objective, the research task is divided into several parts:

 How to efficiently acquire reliable information inside pipelines in this research.

 How to realize the stable long distance communication even in small diameter pipes.

 How to enhance the pipe inspection efficiency and reduce the workload of operator by using intelligent inspection devices, such as technology of self-navigation.

1.3 Thesis Structure

Figure 1-3 Overview of this research and thesis

As it can be inferred from the objectives described above, the evaluation of the research’s hypothesis will be carried through the following aspects, and the relationship between them is illustrated in Figure 1-3.

1. Chapter 2-Background Research Survey: Previous researches about pipeline inspection systems are explained in detail. With such information, the requirements and the gap in this research are clearly defined.

2. Chapter 3-Self-navigation Method: To deal with first technical problem, two automatic control and self-navigation approaches based on ultrasonic detection and pipe feature extraction and evaluation are proposed in this section. Based on these methods, the pipeline robot can realize manual-control and self-navigation together.

3. Chapter 4-RCS based on wireless relay communication (WRC): In order to improve the second technical defect and environmental restriction, a robot chain system based on wireless relay communication (WRC) is proposed and implemented. Besides, to ensure the stability of such WRC between adjacent robots, we adopt received signal strength indication (RSSI)-based evaluation method for cooperative and coordinated movement of robot chain system. Moreover, wireless application layer communication protocol (WALCP) is used to increase the stable performance of WRC.

4. Chapter 5-conclustion and future works: In this chapter, we make a conclusion based on the performance between RCS and former inspection systems. And also list the limitations and inadequacies of this research. In the future works, we consider more environmental factors on this system and carry out more quantitative experiments on the improvement of RCS’s performance.

1.4 Chapter Summary

In this chapter, we introduced the motivation of this research which can be described as environmental restriction and technical bottleneck. Then we propose our solutions and objective for this research. To realize the objective, it is divided into several sub-objectives. At last, this chapter introduces the research structure and the corresponding chapters. The purpose of this chapter is to let the readers have an overall and comprehensive understanding of this thesis.

2 B ^ACKGROUND

R ^ESEARCH

In this chapter, we discuss about the recent researches on the different kinds of gas pipeline inspection technologies. The chapter begins by introducing the conventional external inspection systems. This systems are almost time consuming with low detection accuracy. Then we focus on the internal mobile inspection systems especially the different kinds of pipe robots.

Besides, we propose the other inspection system which combine the above two technologies.

As the most important part of this chapter, we further classify and discuss about the operation methods and communication methods from these different types of inspection systems.

2.1 Chapter Purpose

For designing a more effective, intelligent gas pipeline inspection system which can be applied in deep buried pipeline at much longer inspection distances, the investigation and analysis of conventional pipe inspection systems are necessary. In this chapter, the main objectives are listed below:

 Survey the current researches on the gas pipeline inspection method, especially the pipeline inspection robot, which include basic structure, function, operation and control methods, communication way, etc.

 Comparison between different pipeline inspection systems, comprehensive analyze the advantages and disadvantages of these systems.

 Analyse the current situation and propose our research solutions.

The following sections of this chapter are organized based on the list above. At the final part of this chapter, the chapter summary is proposed.

2.2 External Inspection System

In recent years, the external inspection technology is the most commonly used gas pipeline leakage detection method for Gas Co. The leak detection is mainly based on the gas leakage medium, such as physical parameters: gas concentration, gas temperature, air pressure, etc. The gas pipeline leakage detection technology has the defects of low sensitivity, long reaction time, low stability, and large influence by environmental factors, but the cost is relatively low.

2.2.1 Gas leakage detector based on gas leak

As Figure 2-1, this gas pipeline leak detector is usually operated by the operators. They walk along the pipeline laying route regularly to judge whether there is a gas leak by these detectors. For the mud surface, the gas sensitivity detector with adjustable concentration is directly detected on the ground, and the point where the maximum concentration is consistent with the pipeline positioning is regarded as the leak point. For the cement road in urban streets, the gas leaks along the cracks, voids and loose soil around the pipeline. It is not detected on the ground, but the ground crack is far away from the leak. In order to find out, this situation requires drilling and leaking and then install the sensor network.

(a) (b) (c)

Figure 2-1 Gas leakage detector based on gas leak (a) XP-3110 combustible gas detector (b) MBFUMA gas detector (c) SL-908 gas detector [3]

2.2.2 Gas leakage detector based on gas temperature

When a leak happens form a pipeline, the gas can contain a temperature signature which can differ from the surrounding environment. In some cases, the temperature difference can be substantial. By detecting the temperature change of the surroundings, the distributed temperature sensor network can not only detect the presence of the pipe leak, but also pinpoint the location of the leak.

As illustrated in Figure 2-2, this is a typical system based on gas temperature measurements using fibre-optic distributed temperature sensing (DTS) technology. This system is used to monitor the temperature thermal response of the system due to gas expansion as a result of a leak occurrence. The optic fibre temperature sensors are installed as close together as possible to the pipeline to accurately detect a leak. They can take temperature measurements every 1-5m along a fibre-optic cable with coverage of 60 km per unit. By utilising a DTS-based gas leak detection system, we can realize detection and precise localization of a leak.

Figure 2-2 Gas leakage detector based on gas temperature [4]

2.3 Internal Inspection system

Since the complex pipeline installation environment can block human to intervene directly, it is hard to do quality inspection and fault diagnosis in the pipeline. So, the detection and maintenance of pipeline have become challenge in industrial development. Traditional pipeline external inspection methods have considerable drawbacks of low accuracy. As the most important part of internal inspection technology, the applications of pipeline robots for the inspection and maintenance are considered as one of the most attractive solutions available.

Pipeline robot, as effective detection equipment, can do detection work in the narrow place beyond the human reach. As Figure 2-3, according to robots’ structure and locomotion pattern, they can be mainly classified into five types: wall-press type, legged-/walking-type, wheeled- /tracked-type, inchworm-type, screw-type, etc.

2.3.1 Wheeled-type or tracked-type robot

The wheeled-type or tracked-type robot can be considered as the most conventional and industrialized mechanism for pipeline inspection. These robots are commonly used to move along the bottom of pipe with wheels or tracks. As illustrated in Figure 2-3 (a), wheeled robots such as KARO and MAKRO usually use these movement mechanisms. The robot MAKRO owns several articulated joints between the different wheeled modules. The robots are optimised for moving in the pipe due to the cylindrical shape and snake-like structure. For this type of robot it can also applied to elbows and T-junctions in pipes. However, since the debris

(a) (b) (c)

(d) (e) (f)

Figure 2-3 Gas pipeline robot [5]

caused by corrosion will make the bottom surface so rough that hard for this wheeled robots to go forward.

The robot KANTARO (Figure 2-3(d)) is the other typical sewer robot. A technology of passive adaptation of robot wheels to the bends has been developed so that such robot can move into the straight part of pipes and passes through wide variety elbows with water flow. Because of such passive mechanism, the robot can only move with the water flow, it is unable to navigate actively with human operation.

As another typical Explorer-II (Figure 2-3 (c)) is an autonomous snake-like wheeled type pipe inspection robot built by Carnegie Mellon University). This robot is specially used for detecting underground gas pipelines. The robot uses wireless communication method. It can travel in the maximum distance of 100m in straight pipeline. The robot can carry colour camera with a “Fisheye” lens on the front. The robot can move through straight pipe, sharp bends, Y- junctions, and elbows by using a combination of an on-board driving arms and steering joints.

These functions are served through a real-time external communication through a wireless 802.11b by UDP traffic. The whole movement and navigation of robot is realized by the operator.

2.3.2 Wall-press robot

One better solution to overcome these weaknesses of the wheeled-/tracked-type robots is regarded to adopt the wheel/track mechanism to support against the pipe wall. The special wall- pressing mechanism enables the robots to travel in the pipeline with the friction power by pressing the pipe wall. This mechanism is usually composed of a small ball screw, a strong spring, and the linkage parts. The wall pressing mechanism could adjust the initial position of front or rear wheels with the ball screw. Moreover, every wheel could adjust to the various pipe environment by adopting the strong spring which is attached between each robot wheel and the ball screw slider. This mechanism could create the wall press force through controlling the compressed range of springs. Such as the typical multi-functional robot for in pipe inspection (MRINSPECT) series, as depicted in Figure 2-3 (f). The researcher have designed the robots for the inspection mission of the urban gas pipe network. The robot MRINSPECT-III and -IV could freely travel through the basic environment of pipeline with horizontal or vertical parts.

Furthermore they could travel through the elbows and T-junctions by adopting the special steering modules. Typically, the three dimensional steering ability provides outstanding

mobility in self-navigation which could be a good characteristic in pipe network. Even if the robots own the steering ability, their steering approaches are very different from others. The robot MRINSPECT-III is the system whose active articulated support joints is like a snake in the nature. The robot MRINSPECT-IV is the other differential-drive shape pipe inspection mobile system which could carry on the steering by modulating the velocity of the driving wheels.

2.3.3 Inchworm-type robot

Based on bionics knowledge, inchworm-type pipeline robots are designed by imitating the deformation and peristalsis movement mechanism as earthworm, caterpillars or other insects. The mechanism mainly consist of two clamp parts and an extension parts. The robots move along a pipe by repeatedly clamping the rear parts against the pipe wall, extending the front parts forward. This mechanism can enable the robots to go upward or downward in the vertical pipe without slipping. The other advantage is that they can adjust to the diameter of pipes. Several robots use compressed air instead of electric motors as the drive power. This particular use of compressed air make sense due to the sparks generated by motors could be fatal to the gas pipe. There are also another drawbacks that the movement of robots is slow.

2.3.4 Screw-type robot

By rotating a set of angled wheels corresponding to the direction of movement, the screw- type robot can be moved like “spiral” movement style. As shown in Figure 2-3 (b), these types of robots are ideal for vertical motion application since three or more point contacts prevent them from falling in vertical pipes. The wheels are generally sprung loaded so that the robots can adjust to different diameters. However, these types of robots cannot work particularly well when the robot pass T-junction of pipe or internal damages.

2.4 Integrated Inspection System

Since internal pipe inspection approaches have advantage such as precise detection and accurate localization. While external inspection methods are more convenient with low cost.

Combining various methods, some hybrid inspection methods are proposed. The most typical one is a method using robotic wireless sensor networks (WSNs) as depicted in Figure 2-4. Such WSNs can provide an effective method for pipeline inspection. The networks usually consist

of one underground sensor node (SN) and several aboveground relay nodes (RNs). The robots serve as SN for information acquisition and communication. The robots transmit the information wirelessly to the aboveground relay nodes. The RNs collect the information and send to base station (BS). Such hybrid system are able to perform accurate and real-time

inspection, especially in some complicated pipeline environment such as T-junctions and elbows.

2.5 Operation Method

Since the mobile pipe robots must finish the task of pipe inspection according to the human’s requirements, the suitable operation methods are important to the robots with high efficiency. There are two operation methods: manual operation and self-navigation method based on self-adaption, vision, and laser/infrared light.

2.5.1 Manual operation

In the former research, most of the pipeline robots are operated by the operator. The robots enter the pipe and operator control this robot through the wireless or cable communication.

Since such method enables the robots to go to the desired leakage position precisely by the human’s command, this approach is more precise than conventional systems. Through user

Figure 2-4 Pipeline inspection system based on WSN [6]

interface or monitor, the operator can watch the situation of the robot in real time. When encountering some problems, the operator can take some control measures to avoid some uncertain accidents. As illustrated in Figure 2-3 (a), SCHRODER is the typical robot with only human operation.

2.5.2 Navigation based on passive self-adaption

This type of pipeline robots are usually driven by the drive motor, actuator, air flow or water flow. KANTARO (Figure 2-3 (d)) is the best example which a passive moving mechanism called “NASIR Mechanism”. A technology of passive adaptation of robot wheels to the bends in the pipe was developed so that this robot can move into the straight pipes and passes wide variety of the pipe bends with water flow. When the header of robot arrive at elbow, and the robot will passively bend or change the direction of movement according to the curvature of the pipe, so that the robot can pass through elbows smoothly. Recently, a great deal of researches are focused on how to improve the mechanical structure of the robot, such as increase flexibility, increase degree of freedom. However this type of robot can just pass through the straight pipe or elbows, cannot choose the pathway in T-junctions or branches.

2.5.3 Self-navigation based on vision

Such type of robot navigation method is usually based on the detection and measurement with CCD camera. Firstly, the robot could find the in-pipe landmarks such as: the elbows, T- junctions, Y-miters with image processing or pattern matching. And then, when the robot reaches the elbows or T-junctions of the pipe, it could adjust the movement direction based on the feedback signal from the several positioning sensors such as gyroscope module or the other distance calculation sensor modules. The self-navigation system based on the landmark

Figure 2-5 Self-navigation based on vision [7]

detection by processing shadow images from the CCD has been applied in the robot MRINSPECT-V. The illuminator light module of the robot which includes at least 64 high flux LEDs has been installed around the CCD. In order to detect and measure the shadow image of the elbows and T-branches created by the light, the processing of the pattern matching image has been implemented. A algorithm process for the shadow feature recognition is shown in Figure 2-5.Firstly, before reaching the landmarks of pipe, the robot could measure the driving distance by adopting an odometer. Secondly, when the robot starts steering at the landmarks, the robot could stop calculate the driving distance. Since the robot could detect the precise direction of the travelling path by adopting landmark recognition module, it could update these information to the map database in the real-time. Meanwhile, this robot starts to sense the change of the position by adopting the two-axis gyro sensors. Then, after completely processing from the landmarks of pipe, the robot could calculate the driving distance again and again. By repeating the above steps, the robot could finally realize the self-navigation in the whole pipes.

2.5.4 Self-navigation based on laser or infrared light

The robot MRINSPECT-V is equipped with a CCD, an illuminator and a line laser projector. The line laser projector in Figure 2-6 projects the line-shape laser beam on the internal surface while rotating along the centre of the camera. At first, in order to detect the elbows or branches in straight pipe precisely. The laser beam is projected on a shadow area, the projected line is viewed as two separate lines from the view point of the camera. If the line laser projector rotates along the viewing direction of CCD, the accumulated pattern of projected lines would

Figure 2-6 Self-navigation based on laser or infrared light [8]

be different for each landmark. By using this accumulated pattern, the robot is able to distinguish each landmark and straight pipeline segment. In Figure 2-6, the feature images of one laser beam is projected on the straight part, elbow and T-branch of the pipe. Each image

could be distinguished from the other features. In order to make the robot to recognise the landmarks by adopting such image, we require to encode such shape image to some specific numbers. By analysis, we could obtain only three types of shapes of the landmarks: one single line, one long separated line and a short separated line. By using these numbering method, the

TABLE 2-1 Self-navigation method in pipeline

feature image could become only a ring-shape array of integers from 1 to 3. Such array could be well matched to an ideal array of landmarks after filtering processing.

2.5.5 Discussion

Based on the above introduction of former operation method for pipe inspection system, we can obtain Table 2-1. This Table contains all of the above mobile detection systems.

Although the passive self-adaption mechanism is not complicated, it can only applied in very simple pipe environment. The manual operation rely heavily on the experiences of operator. In addition, the self-navigation based on laser/infrared mechanism can enable the robot to realize the navigation in complex however it also occur with some inaccuracy during the data calculation. Therefore, we want to combine the manual operation and self-navigation approach and applied in our new robot inspection system in this research.

2.6 Communication Method

As the other important technical difficulties in the pipe inspection technology, we also focus on this topic with large work in this research. There are many communication methods which can be applied in the pipeline inspection technology. These methods can be mainly divided into two categories: wireless and wired communication. These two methods are currently being studied widely.

2.6.1 Wired communication

For internal inspection method, wired communication is usually realized by the copper cable. Such as ethernet, serial, CAN bus cable, SPI bus cable, even USB cable. However, the usage of cable can limit the inspection range of pipe due to the fact that the long and heavy cable will reduce the robot speed, sometimes, even stop the smooth motion of robot because some complex pipeline features such as elbows and T-junctions can generate a lot of frictions to easily restrict the inspection robot.

For external inspection method based on gas temperature measurements using fibre-optic distributed temperature sensing (DTS) technology also has some drawbacks, such as complexity of network layout and impossibility of monitoring for unknown pipeline. Besides, although the optic fiber has been proven to have many advantages such as the high speed, broad

bandwidth, few weight, etc, the soil layer can corrode the optic fiber and influence the performance of this temperature sensor network.

2.6.2 Wireless communication

There are many types of wireless communication methods which have been adopted in pipeline robots or systems. According to the frequency channel, these methods can be generally classified into many categories: radio frequency (RF), Wi-Fi, Zigee, and Bluetooth, etc.

Recently these communication approaches are commonly used because of low power consumption and suitable transmission rate for information exchanging in the pipe environment.

For internal inspection system, especially the pipeline robot, the wireless communication enable the robot to move flexible with few friction and limitation from the copper cable if a suitable battery is chosen as the power source for robot. However, a new environmental factor inside pipeline occurs and influences the performance of these wireless inspection methods.

The problem is that the wireless communication distance maybe shorter than in the open air because of the much bigger signal attenuation in the pipe. Considering that the pipe is a cylindrical closed system, when the wireless signals are transmitted in the pipe, many waves are absorbed or refracted by the metal pipe wall. This effect will be strengthened when the waves are transmitted in the elbow of pipe. Besides, after multiple reflection by the pipe wall, the energy loss of the waves can also not be ignored. The most detailed description and experiments will be revealed in Chapter 4. In conclusion, the wireless communication for internal inspection system will meet a big challenge brought by the signal attenuation by the pipeline.

For external and hybrid inspection system, we will meet the same problems. Although hybrid inspection techniques based on sensor relay nodes have revealed a well leak detection performance. The signal can be realized in real time with relay sensor nodes. However, such approach can only be applied to shallow pipes and non-metal pipes, in fact, in order to avoid buildings or the underground facilities, most of the pipes are buried in the deep soil. Wireless signals cannot pass through thick soil layers, metal pipe wall and underground facilities because wireless signal propagation in soil incurs much higher path loss than that in air due to the soil material absorption.

2.6.3 Discussion

We obtain a summary of these above communication methods as shown in Table 2-2. In this table, the typical wired robot such as ALSTOM and MRINSPECT can realize the high communication rate and relatively long communication distance whereas they own high frame loss ratio (FLR) and data error ratio (DER). Since these parameters are the most important indication to reflect the communication quality. For wireless robots such as KANTARO and EXPLORER can obtain the relatively low FLR and DER with high transmission rate. However due to the wireless signal attenuation in the pipe, the maximum wireless communication for robot is much shorter than wired communication. The WSN technology seems to be able to optimize the communication performance through combining the advantages of wired and wireless technology. However it owns a fatal weakness of impossibility of communication in the pipe with a depth of more than 5 m. Based on such investigation of communication approaches, we can propose the necessity of research on wireless relay communication (WRC) methods which will be described in the chapter 4.

TABLE 2-2 Communication method

2.7 Chapter Summary

To design a coordinated and cooperative pipe inspection robot chain system, we investigated and analysed the conventional pipe inspection methods. Corresponding to our purposes, we got the following results:

1) The chapter summarized the three main pipe inspection methods, briefly introduced the principle and characteristics.

2) Through further analysing the operation method and communication method, we can basically describe the two important environmental-factors and technical-factors from research questions in Chapter 1.

3) We analysed the drawbacks of former operation and communication methods. For the operation solution, we will describe in detail in chapter 3, and for the communication solution, we will introduce in chapter 4.

3 S ^{ELF - NAVIGATION}

A ^PPROACHES

Current pipe inspection technology don’t have the enough capability to deal with the problems such as failure of inspection due to human operation error or imprecise manual operation, and the communication barrier caused by special environment of pipeline. We have briefly described these problems in Chapter 2, and then will give one solution to overcome the first technical problem in this chapter.

In general, the navigation methods and defects of pipeline robots can be summarized in two basic aspects:

 Without self-navigation: the robot is usually completely operated by operator. Due to the complexity of the pipeline environment, it is difficult for robot to pass through elbows, junctions and miters only through human assessment.

 Self-navigation based on multi-sensor detection: the assessment of the pipe condition is partially per-formed by calculation from sensor data. Such navigation methods based on sensor information analysis might be associated with imprecision brought by sensors’ low sensitivity and error, furthermore the limited space also restricts the use of a large number of sensors.

To overcome the first technical challenge as mentioned before. This chapter introduces the intelligent robot for pipeline inspection based on self-navigation. Since most robots which rely heavily on manual operation are incapable of self-navigation in pipe. Moreover incorrect operations would degrade the efficiency, and sometimes damage the robots especially when they pass through elbows or junctions. Some robots can realize navigation based on multi- sensor, but navigation performance will be greatly influenced by the performance of these sensors, and space to install large number of sensors is limited. To develop a pipeline robot with high self-navigation, we will do a further and detailed analysis in this chapter [3].

3.1 Chapter Purpose

From the above analysis, we propose two self-navigation methods to reduce the workload of operator and increase the operation accuracy and efficiency. The purposes of this chapter are listed as below:

1) Firstly we developed a new tracked pipeline robot which can travel in the gas pipe for the inspection task. In the following parts, the new structure and specifications will be introduced in detail, include mechanical parts, drive and control parts, pipe inspection parts.

2) Secondly, in order to solve one technical problem caused by manual error operation and inaccurate operation, we propose an approach of pipeline robot’s self-navigation based on feature extraction and image processing. Compared with CCD-based former navigation methods, we use image process fuzzy logic control (FLC) algorithm for robot to pass through elbows or T-junctions. A CCD camera installed on the robot is used for locating region of interest (ROI) in elbow or junction. Moreover, ROIs formed by reflection of robot’s LED light and edge of pipe’s dark hole are considered as input variables in the FLC system. By analysing system outputs, we can control the robot’s speed and yaw angle in real time. Compared with conventional studies on pipeline robot’s navigation method, the proposed method can be more precise and faster by using FLC algorithm and analysing ROI with fewer sensors. Finally, we conducted a simulation validation, and the results showed that the robot was capable of adapting to known pipe environments and realizing navigation in straight part, elbow, and junction of pipe.

3) Finally, I present another method for self-navigation based on ultrasonic detection and evaluation. Based on ultrasonic detection, it is capable of moving in straight pipe, turn at elbow/ T-junction and also climb the inclined pipe flexibly and efficiently.

Compared with previous robots, it owns two working modes: self-navigation (automated control) and manual control mode for pipeline inspection, which can be more convenient for operator to operate the robot. Moreover, this chapter also presents a preliminary experiment with a typical M-pipeline in 10cm diameter/4m length. The experiments results revealed that such self-navigation method could improve the efficiency of inspection.

3.2 Robot Platform

As stated above, pipeline robots require good flexibility, mobility and stability at the same time. Hence based on comprehensive performances of climbing ability, speed, difficulty and stability, we consider that the tracked type mechanism is the most suitable one for our research.

So we determine to design two types of pipeline robots based on such mechanism. These robots are developed and implemented in our research on navigation and communication technology.

In the following chapters, we will introduce in detail [2].

Figure 3-1 System diagram of pipeline robot

3.2.1 Mechanical system

As shown in Figure 3-1, the pipeline robot owns modular architecture in its mechanical structures. The whole robot is built on Pololu Zumo chassis. Such chassis is a small tracked high-performance platform which owns the size of 90 x 90 mm, allowing it enter into 100 mm diameter pipeline freely. The drive system consists of two black silicone tracks, one on each side, that are each supported by a freely spinning idler sprocket and a motor-driven drive sprocket. The chassis is composed of black ABS and features a compartment for four AA batteries and sockets for two micro metal gear motors. A black acrylic plate is included with the chassis. This plate holds the motors in place and can be used for mounting electronic device.

The whole detail specification of structure is explained in Table 3-1, the robot mainly consists of six parts: track chassis (Pololu Zumo Co., Ltd.), DRV8835 motor driver module (TI Co, Ltd), power converter module, core STM32F103 embedded controller module (ST Co., Ltd.), NRF24L01+ wireless communication module (Nordic Co, Ltd), HC-SR04 ultrasonic sensor, inertial measurement unit, air pressure/temperature sensor for pipe inspection, 3.7V Li-Po battery, and CCD camera (Thanko Co., Ltd.) [3-4].

3.2.2 Drive and control system

The robot uses TI. Co DRV8835 to drive two gear motors, DRV8835 is a dual H-bridge driver module which can be used for bi-directional DC motor control at voltage from 2.7 V to 11V. It can continuously supply the motor up to about 1200 mA per channel. This electrical performance making it suitable choice for the self-navigation, cooperative and coordinated movement of robot system in this research. It can assist the robot to realize the mode of forward, backward, and rotation precisely by controlling the speed of two gear motors.

Control system adopts MCU of STM32F407ZGT to control the motors and communicate with controller outside the pipe. This MCU is based on Cortex-M4 Core. The MCU owns 8 Timers for 32 channels PWM outputs, which are enough for the motor control. It also has two SPI bus interfaces for connecting with wireless communication module and IMU device. Two I2C interfaces for inspection sensor connection are also provided by this MCU [4].

3.2.3 Pipe inspection system

This tracked robot can carry two types of sensors. One is a gas concentration sensor, the other is the gas pressure and temperature sensor. The CCS811 is an ultra-low power digital gas sensor solution which integrates a metal oxide (MOX) gas sensor to detect a wide range of Volatile Organic Compounds (VOCs) for indoor air quality monitoring with a microcontroller unit (MCU), which includes an analog-to-digital converter (ADC), and an I2C interface.

CCS811 is based on an unique micro-hotplate technology which enables a highly reliable solution for gas sensors, very fast cycle times and a significant reduction in average power consumption. The integrated MCU manages the sensor drive modes and raw sensor data measured while detecting VOCs. The I2C digital interface significantly simplifies the hardware and software design, enabling a faster time to market [5].

BMP280 is an absolute barometric gas pressure and temperature sensor especially designed for mobile applications. The pressure can be detected range from 300-1100 hPa with excellent relative accuracy ±0.12 HPa. The ASIC performs A/D conversions and provides the conversion results and sensor specific compensation data through a digital interface. BMP280 is equipped with a built-in IIR filter in order to minimize short-term disturbances in the output data.

TABLE 3-1 System diagram of pipeline robot

3.3 Self-navigation based on Feature Extraction & Image Processing

In this research, a new CCD-based evaluation method combining with the fuzzy logic control (FLC) algorithm is implemented to the gas pipe robot’s navigation, especially for travelling through the special situations such as elbows and T-junctions. CCD has been applied for detection, localization, and classification of the region of interest (ROI) generated by the LED light reflection and edge areas of dark hole at elbow or T-junction. We only consider the relationship between these ROI’s features and the navigation of this robot. By adopting the FLC algorithm, we have further quantified these characteristics of the ROI, and regard them as the precise input variables for the FLC system. Finally, with MATLAB simulation, we could control the robot’s velocity and attitude in the real-time. Such approach requires fewer sensors and even without positioning systems, and it could enable the navigation of pipe robot much faster and more precisely [6].

3.3.1 Landmark calculation and detection

The gas pipe contains various landmarks such as the straight parts, Y-miters, elbows, and T-junctions. To enable the robot to travel through these special parts, the recognition of these landmarks is very necessary. Since the robot’s LED light could be reflected at the elbow or junction, and different pipe conditions could reflect the different characteristics as depicted in Figure 3-2. CCD installed on this pipe robot could capture and classify the ROI of light reflection at elbow and T-junction. Moreover, ROI could represent the correct direction and position of the gas pipe bend. Figure 3-3 (a) and (b) illustrate that the region of ROI could be changed when the robot comes closer to the elbow and junction. We choose the ROI as the navigation variable in this FLC system [7-8]. Another basic element is the ROI generated by

Figure 3-2 Direction and location of light reflection’s ROI in elbow and T-junction