喜歡這套資料就充值下載吧。資源目錄里展示的都可在線(xiàn)預(yù)覽哦。下載后都有，請(qǐng)放心下載，文件全都包含在內(nèi)，有疑問(wèn)咨詢(xún)QQ:414951605 或 1304139763 AdvancedRobotics, Vol. 17, No. 4, pp. 339358 (2003)VSP and Robotics Society of Japan 2003.Also available online - Full paperThe concept and research of a pipe crawling rescue robotZHELONGWANGand ERNEST APPLETONSchool of Engineering,Universityof Durham, Durham DH1 3LE, UKReceived 19 June 2002; accepted 31 July 2002AbstractThis paper presents the control algorithm and design of a pipe crawling robot which canbe used for the purpose of earthquake rescue and pipeline maintenance. The robot is designed tobe able to intelligently alter its body shape to ? t the pipe or tunnel-like voids within rubble. Thepaper introduces a simulation to test how the robot alters its body shape to ? t voids. The controlalgorithmuses a look-uptablemethod combinedwith the methodof leastsquaresto predictthe shapeof the robot under the in?uence of actuators. The paper also presentsthe design of both the hardwareand software systems of the robot, and laboratoryexperiments on the robot body module. Computersimulationresultsby using MATLAB and the experimentalresultsindicatethe feasibilityof the robotbody shape changecontrol algorithmproposal.Keywords: Pipe robot; controlalgorithm;method of leastsquares;deformable body; rescue robot.1. INTRODUCTION1.1. A pipe crawling robotInternational pipeline systems used for the transmission of oil, gas and water aregrowing in age, and some installations have already been in operation beyond theservice life they had originally been designed for. Inspection of short pipes is animportant task faced by many industries. The deterioration of the inner surfaceof pipes is especially common in re? neries and steam plants.It is therefore ofever-increasing importance that pipeline operators are provided with the means toaccurately and reliably inspect their pipelines, and obtain the information neededfor decision making regarding safe operation, rehabilitation and repair 1. The useof automatic pipe inspection could reduce the downtime and manpowerrequired forthe pipe inspection process. This highly demanding requirement has resulted in thedevelopment of various kinds of pipe robots.To whom correspondenceshouldbe addressed. E-mail: zhelong.wangdurham.ac.u k340Z. Wang and E. Appleton1.2. Earthquake rescue robotsEarthquakes are unfortunately frequent happenings and very dangerous naturalphenomena.In almost every major earthquake many victims are buried undercollapsed buildings, bridges, roadways, etc.It is very dif? cult to rescue thesepeople, who may well be injured, hungry and weak. In addition, the structuralconditions under the rubble can be complex, dangerous and unknown. Thus, thereis a primary need to explore these conditions and determine the victims location andtheir condition. The matter is also urgent as it is important to rescue those victimsas soon as possible. As a result, rescue robots have been developed to undertakethese tasks. In an article by Tokuda et al. 2, a prototype CUL (Carry and powerassist robot for Unspeci? ed Landform) which will be used for earthquake rescuemissions has been described. The paper also describes results of experiments for afeasibility study. The experimental results show the effectiveness of the CUL rescuerobot proposed.2. PROBLEM PRESENTATION2.1. The shortcomings of existing pipeline robot designPipeline robots are mentioned in a large number of papers. For example, a robotsystem named KARO 3 (Entwicklung eines ? exible einsetzbaren Robots zurintelligenten sensor-basierten Kanalinspektion) is equipped with intelligent multi-sensors. These allow automatic and reliable detection of damage location, its typeand size, and give superior performancecompared withthe majority ofCCTV-basedsystems. The system has been developed for the protection of groundwater and soilagainst contaminating materials and liquids. Toyomiand Koichi 4 have developeda micro-inspection robot for 1-inch diameter pipes. The robot can undertake visualinspections in pipes and also retrieve small objects. Glen and Devon 5 describe arobot that undertakes automatic pipe inspection and has been proved to be a suitablealternative to current pipe inspection techniques.The advantage of this systemis that it utilizes established mechatronic principles to produce a low-cost devicecapable of detecting inner pipe defects. However, all these robots can only work inpipes of ? xed diameters and cannot work if the wall of the pipe is broken or badlydamaged, if the pipe diameter varies or if the pipe collapsed partially.2.2. A Brush robotA pipe crawling robot has been developed at the Center of Industrial Automationand Manufacturing (CIAM), School of Engineering, University of Durham. Therobot utilizes a unique, innovative and patented traction system. The principle ofthe drive system is simple. That is, if a brush with a diameter slightly larger thanthe bore of a pipe is inserted into a pipe, its bristles are swept back at an angle.Under this condition, it is easier to push the brush forwards though the pipe thanThe conceptand research of a pipe crawling rescue robot341it is to pull it backwards. Thus, if two brush are interconnected by a reciprocatingcylinder, then, by cycling the cylinder, it is possible for the robot to crawl along thepipe 6. The drive system has been awarded patents on a near worldwide basis.More details about the drive system are illustrated in Stutchbury 6 and Han 7.Most of the robots developed in Durham are powered by pneumatics and grip thewall of the pipe by means of many bristle clusters, hence named Brush robots. Justlike other pipeline robots mentioned earlier, a Brush robot can only work in pipeswithin a limited diameter range although, unlike many of the other traction systems,the bristle mechanism is capable ofdealing with broken on partially collapsed pipes.2.3. An improved Brush robot for earthquake rescue and pipeline maintenanceThis report outlines some improvements on the established Brush robot principle,and adapts it to work for the purpose of earthquake rescue and in severely damagedor broken pipelines. Thus, it is designed to be able to alter its body shape to ? tthe variable void shapes in a collapsed building or different diameter pipes whosewalls might be broken or in a bad condition. To realize these functions, a sensorsystem for detecting the hole shape and a control system for altering the robot bodyshape is necessarily equipped. For the purpose of rescue, a CCD camera needs tobe equipped to see the conditions in the hole. A CO2detector is used for detectingwhether victims are alive or not and for locating their positions. A microphone isequipped for a victim to communicate with rescue personnel. In addition, an airhose will be carried by the robot for conditions of air de?ciency.3. RESEARCH METHODOLOGY3.1. OutlineThe key point for a Brush robot is to produce enough friction to drive itself andcarry the necessary loads. For the purpose of earthquake rescue, the robot should beable to pass through holes and cracks in the ruins made by the earthquake. However,the holes and cracks in the ruins usually have irregular shapes and sizes. Thus, therobot must be able to alter its body shape and ? t the hole shape to produce enoughfrictional force. To realize the requirement, it is essential to install a sensor systemon the robots head, which is used to roughly detect the hole and crack shapes. Inthe design of this Brush robot, four groups of actuators need to be installed aroundthe robots body. These actuators will change the robots body shape according tosignals from the sensor system. During this procedure a control system will recog-nize the sensors signals, performing calculations based on the signal informationand sending appropriate commands to change the actuators. Using a CCD cameramounted on the head of the robot, the conditions in the holes and cracks can beinvestigated visually. A manual control function is also required to deal with excep-tional circumstances in the case of failure of the robot automatic control function. A342Z. Wang and E. Appletonsoftware system based on a PC is also required to run data communication and con-trol, store data, etc. More speci? cally, the software is required to be able to recordthe robots routes and the hole shapes at different travel stages.3.2. Robot working mechanism theory3.2.1.The mechanism theory of an old Brush robot.Before building a real robotmodel, it is necessary to do some computer simulations to prove the feasibility ofthe control algorithm. Movement of a Brush robot is achieved by the utilization ofcurved bristle as the means of propulsion and support, as illustrated in Fig. 1.When the cylinder opens, the leading brush, offering lower resistance becauseof the bristle curvature, moves forward easily; the trailing brush, because of itshigher resistance tobackwardforces, remains stationary. However,whenthe reversehappens, i.e. the cylinder closes, the leading brush remains stationary, whereas thetrailing brush, now offering low resistance, is pulled forward.Based on this theory, the resultant traction depends entirely on the bristlemechanism set-up and can be illustrated in the following way.Considering asingle bristle for the purpose of simplicity, when a bristle is put into a pipe, andbecause of its effective lateral dimension, it is bent by the pipe wall, there will bea perpendicular force P acting at the tip of the bristle, as shown in Fig. 2. Whenmoving the core of bristles, traction F should equal P. The projection of a bristlein the direction of the y-axis is marked as h. The length of a bristle is l. The chordbetween the two tips of the bristle is expressed as L. In the thesis 6, Stutchburygivesthe conclusion that the optimum angelbetween the bristle and pipe wall shouldbe between 30and 40to achieve the best traction F, although this angel will varydepending upon a number of factors, e.g. lubrication. In Ref. 7, Han gives us therelation between h and l:hlD 2E./Ql L:(1)Note:E./ D21 122sin221 32 42sin4 231 3 52 4 62sin6 25 ;andQl D21 C122sin22C1 32 42sin4 23C1 3 52 4 62sin6 25C :The value of h can be obtained if l and are known. The thesis by Han givesa table which indicates the results obtained by applying (1), as shown below inTable 1.3.2.2. The mechanism theory of an improved Brush robot.A newly improvedBrush robot used the same brush mechanism as the old one, but the structure ofThe conceptand research of a pipe crawling rescue robot343Figure 1. Brush robot motion principle.Figure 2. Bristle mechanism diagram.Table 1.Spreadsheetfor the relationbetween h and l (deg)P=PEuler=lh=l101.0040.11160.9923201.0160.21930.9698301.0350.25880.9324401.0620.42210.8787601.1520.59300.6973901.3920.79250.4189344Z. Wang and E. AppletonFigure 3. The structureof one robot body module.its body module has been modi?ed. The old Brush robots body module is a solidsteel cylinder mounted with hundreds of steel bristles. Thesteel cylinder is the robotbody core of a ? xed diameter and the body core cannot do any changeto its physicalshape. However, in the improved Brush robot, a thin steel strip circle replaces theold robot body cylinder, as illustrated in Fig. 3. Inside the strip circle, four actuatorsconnect the strip circle by joining the actuator end points with the strip circle. Likethe old robot body module, hundreds of steel bristles are mounted on the surface ofthe strip circle. The new, improved robot body module with such a mechanism canalter its shape by actuators pushing in and out.3.3. Control algorithm for the new Brush robotTomake an improved Brush robot to be able to alter its body shape and ? t the voids,a hybrid control algorithm based on a look-up table method and the method of leastsquares is developed. The method that the robot use to decides how to alter its bodyshape to ? t the hole is reference to a data ? le, stored in a table. The table is com-posed of data ? les and each ? le presents one calculation result obtained by usingABAQUS software. Figure 3 illustrates the structure of one robot body module.Each robot body module is composed of four actuators, a thin spring steel strip cir-cle and hundreds of spring steel bristles mounted on the surface of the strip. Theend point of each actuator is connected to the strip so that the shape of the steel stripcircle can bedeformed by the actuators pushing in and out. If several hundred pointson the strip circle are marked, the position of each point can be recorded as a pair ofcoordinates. The strip circle shape can be uniquely represented if all points coordi-nates can be known. The shape of the strip circle can be acquired from coordinatesof those points when actuators push in/out and the strip circle is deformed by suchpushing. In fact, here the shape of the strip circle represents the actual shape of therobot body module. These shapes are related to actuator loads and consequentlyde?ections that are put on the robot body in the direction of the x-axis and y-axisindependently. Thus, the robot body module shape will be changed with varied ac-tuator loads and de?ections. In the mean time, the coordinates of the points on therobot body module will be changed because of the robot body de?ection. The co-ordinates of those points can be calculated and predicted by using ABAQUS,whenactuator loads are already known. This array of point coordinates presents the robotThe conceptand research of a pipe crawling rescue robot345Table 2.A data ? le stores an array of point coordinatesPOINT1POINT2POINTnXiX1X2XnY1Y1Y2Ynbody module shape under such actuator loads. If the actuator loads are changed, anew array of point coordinates can be acquired and this means that the robot bodymodule will assume a new shape. Thus by changing the actuator loads, which areinput variables in the calculation by ABAQUS,many arrays ofpoint coordinates canbe acquired. Each of them uniquely represents a robot body module shape. Thesearrays of point coordinates can be stored in data ? les and each data ? le is used as arecord to put into a table. As a result, this table includes many robot body moduleshapes under different actuator loads. Similarly a hole shape can also be uniquelyexpressed as an array of coordinates of points around the hole wall. The robotscontrol algorithm is required to be able to ? nd the most appropriate shape to ? t thehole shape from those data ? les in the table.For example, Table 2 expresses an array of point coordinates stored as a data ? le,which is a record in the robot body shape table. A POINTi(i D 1;2;:;n/ meansa pair of coordinates of a point on the robot body after the robot body is de? ected.POINTiis expressed as (Xi;Yi) in the meaning of the coordinates, which is apoint on the robot body. The coordinate (xi;yi) is expressed as the coordinate ofa point on the hole wall. diis the difference between a point on the robot bodyand its corresponding target point on the hole wall along the same direction. If therobot body module shape could ? t the hole shape very well, that means that each dishould be as small as possible. To realize this, the robot control algorithm needs to? nd the smallest D, which is the sum of the squared distance di. Finally, the controlalgorithm will search all data ? les in the robot body module shape table and ? ndthe best data ? le to minimize D, which is calculated by using (3). l is the optimallength of the bristle, which is a constant and can be know by using a spreadsheetin 6. n is the total number of points around the robot body module.diDx2iC y2iX2iC Y2i l;.i D 1;2;:;n/;(2)D DniD1x2iC y2iX2iC Y2i l2;.i D 1;2;:;n/:(3)3.4. Computer simulation of the control algorithmTo test the feasibility of the control algorithm presented above, a number ofcomputer simulations have been performed using MATLAB.The simulation resultsshow that the control algorithm proposed will be effective. Figures 46 are based346Z. Wang and E. AppletonFigure 4. Robot body module ? ts a square: C indicates a point of the robot body and indicatesa point of the holes wall.Figure 5. Robot body module ? ts an ellipse: C indicatesa point of the robot body.Figure 6. Robot body module ? ts an irregularshape: indicatesa point of the holes wall.The conceptand research of a pipe crawling rescue robot347on some simulation results. In these ? gures, the hole shape is expressed by an arrayof interconnected symbols and the robot body module shape is expressed by anarray of interconnected symbols C. Also each symbol represents a point onthe hole wall and each C symbol represents a point on the robot body module. InFig. 4, the hole shape is a rectangle and the robot body curve ? ts the hole well. InFig. 5, the hole shape is an ellipse and the robot body curve is similar to the holeshape, but it is relatively smaller than the hole. The bristle around the robot bodycan deal with the little difference by elastic de? ection. Figure 6 shows the robotbody tested to ? t an irregular hole shape and most of the robot body module curvecan ? t the hole by elastic de?ection of the bristles; however, the right upper cornercannot be ? tted by the robot body module curve. To achieve a better ? tting, furtherwork needs to be carried on making a non-symmetric robot body.From the simulation results above, the robot body module could alter its shape to? t some basic geometric shapes and simple irregular shapes. The control algorithmis basically proved to be feasible. More complicated irregular shapes cannot be? tted well by using the current structure of the robot body module. This could besolved by making a non-symmetric robot body module and using more actuators,which will enable robot body module to change into more complex shapes to ? tvarious hole shapes.3.5. Robot control system3.5.1.Controlmodulediagram.It is envisaged that a control program will run ina PC and that a sensor system will send back signals about the constantly changinghole shape. After processing these signals, the PC makes the control decisionsand sends control commands to every control module. Then the control moduleswill control the movements of the actuators according to those control commands.Figure 7 displays that a robot control system that includes two layers of control.One is the control from a PC to each control module; the other is the control fromeach control module to its corresponding actuator. In addition, a sensor systemFigure 7. Control module diagram.348Z. Wang and E. AppletonFigure 8. Control board municates with the PC to collect the information of the hole shape and sendsthe information to the PC.3.5.2. Control board diagram.It is proposed to make a prototype of this robotusing eight stepper motor control boards, 16 stepper motor drive boards, 16 steppermotors and one PC to run the control software. In Fig. 8, a scheme is drawn toexplain the connections between these modules. Each control board can controltwo stepper motors and every stepper motor needs to be driven by a drive board.Four stepper motors are needed for each robot body module, so that a four-body-part Brush robot needs 16 stepper motors in all.4. EXPERIMENTSThis Brush robot is composed of the same four modules. Thus, the laboratoryexperiment focuses on one robot body module.4.1. Laboratory experiment on a robot body moduleTherobot body module experimental device is shown in Fig. 9. The devices includea strain gauge sensor, robot body module, robot actuator control box, DC powersupply and PC. Thestrain gaugesensor is used to detect the void shapes. The sensorincludes 12 ? ngers equipped with strain gauges. The strain gauge can be used todetect the de?ections of the ? ngers. The touching point coordinates of the ? ngerson the void wall can be worked out by these de?ections. With these coordinates, aspline algorithm can estimate the whole void shape.4.1.1. Robot software system.A computer program written in Visual C+ wasdeveloped for the robot prototype experiment to collect sensor signals, controlThe conceptand research of a pipe crawling rescue robot349Figure 9. Robot expe