一种新的方法通过移动的2D激光测距仪检测管线变形Word文档下载推荐.docx

1、 Pipeline deformation; Laser Range Finder （LRF）INTRODUCTIONBuried infrastructure such as water and gas supply pipes are subject to surface live loads as a result of vehicle loading and construction of surface facilities. The live loads, ground movements due to earthquake or similar causes could make

2、 the pipelines failure, including pipe deflection, bending and deformation etc. Pipeline deformation is also applied to estimate the stress in the pipeline, and thus to keep thepipeline stress below the critical level 1. Therefore, it is necessary to detect pipeline deformation regularly.Traditional

3、 techniques for deformation detection include:on-site visual inspection, photogrammetric surveys （terrestrial or aerial）, precise conventional surveys, and geotechnical measurements using either continuous data collection or observation epochs 2. The most recent approaches utilize the Global Positio

4、ning System （GPS）, optical fibre gyroscope 3, sensor network, etc. to help the detecting. The method with GPS can not be applied for underground pipelines. The cost of the detection system with an optical fibre gyroscope hinders it be used widely. A sensor system installed with remote wireless has t

5、he capabilities to monitor pipelines. This system is relative costly and should be installed beforehand.In-pipe robots with CCTV, which have a long history of development, have been employed as major tools to maintain pipeline utilities. However, pipeline deformation is hard to detect accurately by

6、visual inspection. In this paper, we fix a LRF on the in-pipe inspection robot to find the distance between the robot and the inside wall of the pipeline. In this way, we expect to get the three-dimensional reconstruction of the pipelines. We find it is a good way to detect pipeline deformation. Alt

7、hough the LRF can get accurate range information, the robot posture changes instantaneous when the robot is running in the pipeline, which makes the deformation detection difficult.This paper is structured as follows: In sectionwe present the robot system with a moving 2D LRF. The frames of the LRF,

8、 inspection robot and pipe are installed and their transition matrixes are analyzed in section. In section,we describe how to calculate the pipeline deformation with the received data from the LRF. In section we discuss the experimental results and demonstrate the accuracy of the system and methods.

9、 Finally, we present conclusion and future work in section. ROBOTIC SYSTEM AND CALIBRATIONA The Robotic SystemAs shown in Fig 1, the robotic system is a wheeled mobile robot with CCTV for underground pipelines inspection. A PTZ camera is fixed on the arm of the robot.The pipeline can be monitored an

10、d inspected from the visual information captured by the camera as the robot runs in the pipeline. However􀋈the pipe deformation can not be measured by eyesight from this visual information. To detect the pipe deformation, a LRF is used to find the distance from the sensor to the inside wall.

11、Fig. 1 The robotic systemWe choose HOKUYO UBG-04LX 4 produced by Japan as the range find sensor. It can measure 682 steps on 240 per one rotation, and therefore the angular resolution is 0.352 on a 2D plane. Although it cant measure the whole 360, this range is enough for our application. The below

12、120 of the pipeline which cant detect is often filled with water. This system is also equipped with encoders to measure the distance which the robot covered and an inclinometer to observe the posture of the robot.B CalibrationBecause the LRF finds the distance from the sensor to the wall or the obst

13、acles, its installation position is important. The LRF is fixed horizontally in the middle body of the robot. To make maximize use the effective scope, the sensor is fixed on the robot where it can find the up 240. In this way, when there is some water in the pipe, it still can find most of the insi

14、de wall of the pipeline.After fixation, we design a simply method to get and calibrate the height of the sensor. As shown in Fig. 2, we use the table in our laboratory as calibrate plane which is horizontal. Since the brightness of the target influences the precision of the LRF 5, we choose a white

15、plane which is good for reflectivity. Furthermore, to get vertical distance between the LRF and planes, we only select the front step whose index number is 382. This characterization is observed by Chan-Soo 5.Fig. 2 CalibrationAs shown in Fig.2, The distance between the plane of the table and ground

16、 is d0 , which can be measured by a metre rule. The distance between the LRF and the plane of the table is di , which can be detected by the LRF. We calculate the height of the sensor d j as follows:d j = d0 - di （1）Where di is the sensor reading of the 382nd step. However, the value of di is differ

17、ent every time when the LRF measures 682 steps on 240 per one rotation. To get the minimum measuring error, we detect di n times and use the arithmetic mean, which is given by （2）Then we get the height of the sensor d j as follows:d j = d0 di （3） TANSFORMATION AND POSTURE ANALYSISA The three framesW

18、e describe the posture of the inspection robot moving in round pipes with the coordinates （x, y, z） of one of their points with respect to the inertial basis and their Euler angles （ , , ）.Fig.3 The three frames are establishedAn inertial base frame XWYWZW is fixed in the pipes with axis-ZW aligned

19、with the axis of the round pipes, while a moving frame XRYRZR is attached to the inspection robot. The robot posture can be described in terms of the three coordinates x0 , y0 , z0 of the origin P of the moving frame and the orientation angle （ , , ） of the moving frame, both with respect to the bas

20、e frame with the origin at P. The quantities （ , , ） are the body yaw, pitch, and roll of the robot, which is shown as in Figure 3. Another frame XLYLZL is attached on the LRF.B Transformation MatrixThe data collected by the LRF is based on the coordinate of the LRF, namely in the frame XLYLZL, but

21、the real pipeline deformation is respect to the base frame. Therefore, the transformation matrix from the frame of the LRF to the base frame should be deduced. As shown in Fig.3, the transformation matrix from the frame of the LRF to the frame of the robot is as following:where （xL0 , yL0 , zL0） is

22、the coordinates of the centre of the LRF in the frame of the robot.The transformation matrix from the frame of the robot to the base frame isWhere （6）which can be deduced according to the definition of Euler angles6, andTherefore, the transformation matrix from the frame of the LRF to the base frame

23、 is given by For the wheeled mobile robot working in round pipes, it has 3 degrees of freedom if the posture is described in terms of three coordinates x0, y0, z0 and the orientation angle （ , , ） 6. Two of the orientation angle （ , , ） can be observed by a two-axis inclinometer. The distance of the

24、 robot ran, namely z0 can be detected by the encoder.Therefore, the transformation matrix can be calculated instantaneously.C Posture analysis of the inspection robots in round pipesThe posture of wheeled mobile robots in round pipes can be analyzed by three cases as follows.（1） The robot is horizon

25、tal. In this case, the posture angles satisfy = = = 0 as shown in Fig.4 （a）. Four wheelsof the robot keep in touch with the wall of the pipes in this case.（2） The robot is slanted but its direction is parallel to the axis of the pipes as shown in Fig. 4（b）. In this case, the posture angles satisfy =

26、 0 , = 0 and 0 . When the posture angle is relatively large, the robot will slide to the bottom or even be overturned. （3） The robot is slanted but its direction is not parallel to the axis of the pipes/ducts, which is equivalent to the conditions 0 , 0 and 0 as shown in Fig.4 （c）. Only three wheels

27、 of the four-wheel robot stay in touch with the wall at the same time in this case.（a）First case = = = 0（b） Second case = 0 , = 0 and 0 （c） Third case 0 , 0 and 0Fig.4 Different posture of the robot in the pipelineAs we see from Fig.4, the first and the second case have perfect posture for collecting the accurate information and are easy to calculate the pipeline deformation. In these cases, the data collected by the LRF need not rotate transformation if we only calculate the deformation rate. Unfortunately, these two cases are special condit