资源目录
压缩包内文档预览：(预览前20页/共28页)
编号：78164730
类型：共享资源
大小：2.01MB
格式：ZIP
上传时间：2020-05-09
上传人：柒哥
认证信息
个人认证
杨**（实名认证）
湖南
IP属地：湖南
40
积分
- 关 键 词：
-
曲轴振动自动抛光机的设计
曲轴
振动
自动
抛光机
设计
- 资源描述：
-
购买设计请充值后下载，，资源目录下的文件所见即所得，都可以点开预览，，资料完整，充值下载可得到资源目录里的所有文件。。。【注】：dwg后缀为CAD图纸，doc,docx为WORD文档，原稿无水印，可编辑。。。具体请见文件预览，有不明白之处，可咨询QQ:12401814
- 内容简介：
-
Robotics and Computer-Integrated Manufacturing 23 (2007) 371379Robotic sanding system for new designed furniturewith free-formed surfaceFusaomi Nagataa,?, Yukihiro Kusumotob, Yoshihiro Fujimotob, Keigo WatanabecaDepartment of Electronics and Computer Science, Tokyo University of Science, Yamaguchi, Daigaku-Dori 1-1-1, Sanyo-Onoda 756-0884, JapanbInterior & Design Research Institute, Fukuoka Industrial Technology Center, Agemaki 405-3, Ohkawa, Fukuoka 831-0031, JapancDepartment of Advanced Systems Control Engineering, Graduate School of Science and Engineering, Saga University, Honjomachi 1, Saga 840-8502, JapanReceived 10 December 2005; received in revised form 10 March 2006; accepted 7 April 2006AbstractIn this paper, a sanding system based on an industrial robot with a surface following controller is proposed for the sanding process ofwooden materials constructing furniture. Handy air-driven tools can be easily attached to the tip of the robot arm via a compact forcesensor. The robotic sanding system is called the 3D robot sander. The robot sander has two novel features. One is that the polishing forceacting between the tool and wooden workpiece is delicately controlled to track a desired value, e.g., 2kgf. The polishing force is definedas the resultant force of the contact force and kinetic friction force. The other is that no complicated teaching operation is required toobtain a desired trajectory of the tool. Cutter location (CL) data, which are tool paths generated by a CAD/CAM system, are directlyused for the basic trajectory of the handy tool attached to the robot arm. The robot sander can be applied to the sanding task of free-formed curved surface with which conventional sanding machines have not been able to cope. The effectiveness and promise are shownand discussed through a few experiments.r 2006 Elsevier Ltd. All rights reserved.Keywords: Robotic sanding; CAD/CAM; Cutter location data; Non-taught operation; Surface following control; Polishing force1. IntroductionIn manufacturing industry of wooden furniture, CAD/CAM systems and NC machine tools have been introducedwidely and generally, so that the design and machiningprocesses are rationalized drastically. However, the sand-ing process after machining is hardly automated yet,because it requires delicate and dexterous skills so as notto spoil the beauty and quality of the surface. Up to now,several sanding machines have been developed for woodenmaterials. For example, the wide belt sander as shown inFig. 1 is used for flat workpieces constructing furniture.The profile sander as shown in Fig. 2 is suitable for thesanding around the edge. However, these conventionalmachines cannot be applied to the sanding task of theworkpiece with free-formed surface. Accordingly, we mustdepend on skilled workers who can not only performappropriate force control of sanding tools but also dealwith complex curved surface as shown in Fig. 3. Skilledworkers usually use handy air-driven tools such as a doubleaction sanding tool and an orbital sanding tool as shown inFig. 4. In order to produce a better surface quality, thedouble action sander simultaneously performs rotationaland eccentric motions. As can be guessed, such toolsspread out unhealthy noise, vibration and dust. The mostserious problem in the sanding process is that the sandingtask in such a bad working environment is extremely hardfor skilled workers. From this reason, an advanced sandingmachine which can even partially replace the skilledworkers is being required in the furniture manufacturingindustry.Industrial robots have been progressed remarkably andapplied to several tasks such as painting, welding, handlingand so on. In these cases, it is important to preciselycontrol the position of the end-effector attached to the tipof the robot arm. On the contrary, when the robots areapplied to polishing, deburring or grinding task, it isARTICLE IN PRESS/locate/rcim0736-5845/$-see front matter r 2006 Elsevier Ltd. All rights reserved.doi:10.1016/j.rcim.2006.04.004?Corresponding author. Tel.: +81836884547; fax: +81836883400.E-mail address: nagataed.yama.tus.ac.jp (F. Nagata).indispensable to use some force control strategy withoutdamaging the object. For example, polishing robots andfinishing robots were presented in 15. Automated roboticdeburring and grinding were also introduced in 610.Two representative force control methods have been everproposed. They are impedance control 11 and hybridposition/force control 12. The impedance control is one ofthe most effective strategies for a manipulator to desirablyreduce or absorb the impact force with an object. It ischaracterized by an ability that controls the mechanicalimpedance such as inertia, damping and stiffness actingbetweentheend-effectoranditsenvironment.Theimpedance control does not have a force control mode ora position control mode but it is a combination of the forceand velocity of the end-effector. On the other hand, thehybrid position/force control method simultaneously con-trols the position and force of a robot manipulator.However, the six constraints which consist of 3-DOFpositions and 3-DOF forces in a constrained frame cannotbe simultaneously satisfied. In order to avoid the inter-ference between the force control system and positioncontrol system, either force control mode or positioncontrol mode is selected in each direction.Surface following control is a basic sanding strategy forindustrial robots. It is known that two control schemes areneeded to realize the surface following control system. Oneis the position/orientation control of the sanding toolattached to the tip of the robot arm. The other is the forcecontrol to stably keep in contact along the curved surfaceof the workpiece. It should be noted that if the geometricinformation on the workpiece is unknown, then it is sodifficult to satisfactorily control the contact force movingwith a higher speed 13. To suppress overshoots andoscillations, for example, the feed rate must be given asmall value. Furthermore, it is also difficult to control theorientation of the sanding tool, keeping in contact with theworkpiece from normal direction.ARTICLE IN PRESSFig. 2. Conventional profile sander.Fig. 3. Sanding scene by a skilled worker.Fig. 4. Handy air-driven sanding tools usually used by skilled workers.Fig. 1. Conventional wide belt sander.F. Nagata et al. / Robotics and Computer-Integrated Manufacturing 23 (2007) 371379372Theauthorshaveconductedrelevantfundamentalstudies. As for force control, impedance model followingforce control method was proposed for an industrial robotwith open architecture concept 14. The force controlleradjusts the contact force acting between a sanding tool andworkpiece through a desired impedance model. In 15,fuzzy environment model was presented for environmentswith unknown physical property. The fuzzy environmentmodel is learned with genetic algorithm and can estimatethe stiffness of unknown environment. The effectivenesswas evaluated through simulations using a dynamic modelof PUMA560 manipulator. In 16, a gravity compensatorwas considered to remove the influence of tool weight frommeasured force. In 17, concerning tool position andorientation control, it was further considered how torealize non-taught operation for industrial robots. Further-more, hyper CL data were also presented to deal with newstatements about the regulation of sanding parametersin 18.In this paper, a robotic sanding system is integrated fornew designed furniture with free-formed curved surface.The robotic sanding system provides a practical surfacefollowing control that allows industrial robots not only toadjust the polishing force through a desired impedancemodel in Cartesian space but also to follow a curvedsurface keeping contact with from normal direction. Thepolishing force is assumed to be the resultant force ofcontact force and kinetic friction force. We also describehow to apply the sanding system to a sanding task ofwooden workpiece without complicated teaching process.A few sanding experiments are shown to demonstrate theeffectiveness and promise of the proposed robotic sandingsystem using the surface following controller.2. Robotic sanding systemRecently, open architectural industrial robots have beenproposed to comply with users various requests withregard to application developments. The industrial robothas an open programming interface for Windows or Linux,so that we can try to program new functions such as forcecontrol, compliance control and so on. The 6-DOFindustrial robot shown in Fig. 5 is a FS20N with a PC-based controller provided by Kawasaki Heavy Industries.The proposed robotic sanding system is developed basedon the industrial robot whose tip has a compact forcesensor. A handy sanding tool can be easily attached to thetip of the robot arm via the force sensor. A PC is connectedto the PC-based controller via an optical fiber cable. ThePC-based controller provides several Windows API (ap-plication programming interface) functions, such as servocontrol with joint angles, forward/inverse kinematics andso on. By using such API functions, for instance, theposition and orientation at the tip of the robot arm can becontrolled easily and safely. In the following section, thesurface following controller is implemented for roboticsanding by using the Windows API functions.3. Surface following control for robotic sanding systemThe robotic sanding system has two main features: one isthat neither conventional complicated teaching tasks norpost-processor (CL data ! NC data) is required; the otheris that the polishing force acting on the sanding tool andtool position/orientation are simultaneously controlledalong free-formed curved surface. In this section, a surfacefollowing control method indispensable for realizing thefeatures are described in detail.3.1. Desired trajectoryRobotic sanding task needs a desired trajectory so thatthe sanding tool attached to the tip of the robot arm canfollow the objects surface, keeping contact with the surfacefrom the normal direction. In executing a motion using anindustrial robot, the trajectory is generally obtained inadvance, e.g., through conventional robot teaching pro-cess. When the conventional teaching for an object withcomplex curved surface is conducted, the operator has toinput a large number of teaching points along the surface.ARTICLE IN PRESSFig. 5. Robotic sanding system developed based on an open architectural industrial robot FS20N.F. Nagata et al. / Robotics and Computer-Integrated Manufacturing 23 (2007) 371379373Such a teaching task is complicated and time-consuming.However,iftheobjectisfortunatelydesignedandmanufactured by a CAD/CAM system and an NC machinetool, then the CL data can be referred as the desiredtrajectory. In order to realize non-taught operation, wehave already proposed a generalized trajectory generator19,20 using the CL data, which yields the desiredtrajectory rk at the discrete time k given byrk xTdk oTdk?T,(1)wherexdk xdxk xdyk xdzk?Tandodk odak odbk odgk?Tare the position and orientationcomponents, respectively. odk is the normal vector at theposition xdk. In the following, we detail how to make rkusing the CL data.A target workpiece with curved surface is generallydesigned by a 3D CAD/CAM, so that the CL data can becalculated by the main-processor of the CAM. The CLdata are sequential points along the model surface given bya zigzag path or a whirl path. In this approach, the desiredtrajectory rk is generated along the CL data. The CL dataare usually calculated with a linear approximation alongthe model surface. The ith step is written byCLi pxi pyi pzi nxi nyi nzi?T,(2)fnxig2 fnyig2 fnzig2 1,(3)wherepi pxi pyi pzi?Tandni nxi nyinzi?Tare position and orientation vectors, respectively.rk is obtained by using linear equations and a tangentialvelocity vtk represented byvtk vtxk vtyk vtzk?T.(4)A relation between CLi and rk is shown in Fig. 6. In thiscase, assuming rk 2 CLi; CLi 1? we obtain rkthrough the following procedure. First, a direction vectorti is given byti pi 1 ? pi(5)so that each component of vtk is obtained byvtjk kvtkktjiktikj x;y;z.(6)Using a sampling width Dt, each component of the desiredposition xdk is given byxdjk xdjk ? 1 vtjkDtj x;y;z.(7)Next, the desired orientation odk is considered. Wedefine two angles y1i;y2i as shown in Fig. 7. y1i andy2i are the tool angles of inclination and rotation,respectively. Using y1i and y2i, each component of niis represented byai siny1icosy2i,(8)bi siny1isiny2i,(9)gi cosy1i.(10)The desired tool angles yr1k, yr2k of inclination androtation at the discrete time k can be calculated asyrjk yji fyji 1 ? yjigkxdk ? pikktik,(11)where j 1;2. If (11) is substituted into (8), (9), (10), wefinally obtainodak sinyr1kcosyr2k,(12)odbk sinyr1ksinyr2k,(13)odgk cosyr1k.(14)xdk and odk mentioned above are directly obtainedfrom the CL data without any conventional complicatedteaching, and used for the desired position and orientationof a sanding tool attached to a robot arm.3.2. Polishing forceIn this section, a sanding strategy dealing with polishingforce is described in detail. The polishing force vectorFk Fxk Fyk Fzk?Tis assumed to be the resultantforce of contact force vector fk fxk fyk fzk?TandkineticfrictionforcevectorFrk FrxkFryk Frzk?Tthat are given to the workpiece as shownARTICLE IN PRESSWorkpieceCL(i-1)r (k)r (k + 1)r (k + 2)CL(i+1)CL(i)Fig. 6. Relation between CL data CLi and desired trajectory rk.YXZO?2 (i)?1 (i)? (i)? (i )? (i )Fig. 7. Normalized tool vector ni represented by y1i and y2i in robotbase coordinate system.F. Nagata et al. / Robotics and Computer-Integrated Manufacturing 23 (2007) 371379374in Fig 8, where the sanding tool is moving along on thesurface from (A) to (B). Frk is written byFrk diagmx;my;mzkf kkvtkkvtkk diagZx;Zy;Zzvtk,15where diagmx;my;mzkf kkvtk=kvtkk is the Coulombfriction, and diagZx;Zy;Zzvtk is the viscous friction. miand Zii x;y;z are the i-directional coefficients ofCoulomb friction per unit contact force and of viscousfriction, respectively. Each friction force is generated byfk and vtk, respectively. Fk is represented byFk f k Frk.(16)The polishing force magnitude can be easily measured byusing a 3-DOF force sensor attached between the tip of thearm and the sanding tool, which is given bykFkk ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffifSFxkg2 fSFykg2 fSFzkg2q,(17)whereSFxk,SFyk andSFzk are the each directionalcomponentofforcesensormeasurementsinsensorcoordinate system. In the following section, the errorEfk of polishing force magnitude is calculated byEfk Fd? kFkk,(18)where Fdis a desired polishing force.3.3. Feedback control of polishing forceIn the manufacturing industry of wooden furniture,skilled workers usually use handy air-driven tools to finishthe surface after machining or painting. These types oftools cause high frequency and large magnitude vibrations,so that it is so difficult for the skilled workers to sand theworkpiece keeping the polishing force a desired value.Consequently, undesirable unevenness tends to appear onthe sanded surface. In order to achieve a good surfacefinishing, it is fundamental and effective to stably controlthe polishing force. When the robotic sanding system runs,the polishing force is controlled by the impedance modelfollowing force control with integral action given byvnormalk vnormalk ? 1e?Bd=MdDt e?Bd=MdDt? 1KfBdEfk KfiXkn1Efn,19where vnormalk is the velocity scalar; Kfis the forcefeedback gain; Kfiis the integral control gain; Mdand Bdare the desired mass and desired damping coefficients,respectively. Dt is the sampling width. Using vnormalk, thenormal velocity vector vnk vnxk vnyk vnzk? at thecenter of the contact point is represented byvnk vnormalkodkkodkk.(20)3.4. Feedforward and feedback control of positionCurrently, wooden furniture are designed and machinedwith 3D CAD/CAM systems and NC machine tools,respectively. Accordingly the CL data generated from themain-processor of the CAM can be used for the desiredtrajectory of the sanding tool. The tool path (CL data) asshown in Fig. 9, which are calculated in advance based on azigzag path, is considered to be a desired trajectory of thesanding tool. Fig. 10 shows the block diagram of thesurface following controller implemented in the robotsander. The position and orientation of the tool attached tothe tip of the robot arm are feedforwardly controlled by thetangential velocity vtk and rotational velocity vrk,respectively, referring xdk and odk. vtk is giventhrough an open-loop action so as not to interfere withthe force feedback loop. The polishing force is regulated byvnk which is perpendicular to vtk. vnk is given to thenormal direction referring the orientation vector odk.It should be noted, however, that using only vtk is notenough to precisely carry out desired trajectory controlalong the CL data: actual trajectory tends to deviate fromARTICLE IN PRESSFig. 9. Zigzag path generated from main-processor of CAM.Tip of robot armForce sensorSanding toolWorkpiecevtFfFr(A)(B)Fig. 8. Polishing force Fk composed of contact force fk and kineticfriction force Frk.F. Nagata et al. / Robotics and Computer-Integrated Manufacturing 23 (2007) 371379375the desired one, so that the constant pick feed (e.g., 20mm)cannot be performed. This undesirable phenomenon leadsto the lack of uniformity on the surface. To overcome thisproblem, a simple position feedback loop with small gainsis added as shown in Fig. 10 so that the tool does notdeviate from the desired pick feed. The position feedbackcontrol law generates another velocity vpk given byvpk SpKpEpk KiXkn1Epn(),(21)where Sp diagSx;Sy;Sz is a switch matrix to realize aweakcouplingcontrolineachdirection.IfSpdiag1;1;1, then the coupling control is active in alldirections; whereas if Sp diag0;0;0, then the positionfeedback loop does not contribute to the force feedbackloop in all directions. Epk xdk ? xk is the positionerror vector. xk is the current position of the sanding toolattached to the tip of the arm and is obtained from theforward kinematics of the robot. Kp diagKpx;Kpy;Kpzand Ki diagKix;Kiy;Kiz are the position feedback gainand its integral gain matrices, respectively. Each compo-nent of Kpand Kimust be set to small values so as notto obviously disturb the force control loop. Finally,recomposedvelocities vnk vTnk 0 0 0?T, vtk vTtk vTrk?Tand vpk vTpk 0 0 0?Tare summed up,and those of which are given to the reference of theCartesian-based servo controller of the industrial robot.It is known that the six constraints, which consist of 3-DOF positions and 3-DOF forces in a constraint frame,cannot be simultaneously satisfied 21. However, thedelicate cooperation between the position feedback loopand force feedback loop is an important key point tosuccessfully achieve a robotic sanding with curved surface.3.5. Hyper CL dataOne of the features of the robot sander is that the CLdata are referred as the desired trajectory of the sandingtool attached to the tip of the arm. Therefore, thecomplicated teaching process can be completely omitted.It is also useful that no post-processor included in CAM isrequired to transform the CL data into the NC data, i.e.,the robot sander does not run based on the NC data butthe CL data. The conventional CL data mainly deal withstatic positions and orientations along the curved surfaceof a model. In order to realize such a skillful sanding asskilled workers perform, we propose hyper CL data thatcan describe serviceable items as shown in Fig. 11. Forexample, following conditions can be specified:(1) Desired polishing force acting between a sanding tooland a workpiece;(2) Sanding power such as motor torque or air pressure;(3) Feed rate (velocity tangential to curved surface);(4) Task mode change (contact mode 2 non-contactmode), etc.The desired polishing force Fdkgf is recognized byFORCE=Fd.(22)The air pressure Pkgf=cm2of an air-driven sanding tool isregulated byPOWER=P.(23)The feed rate norm kvtkmm/s in contact state is set byVELOCITY=kvtk.(24)The task mode change from contact mode to non-contactmode is switched byNON?CONTACT=vm;d,(25)where the sanding tool takes off from the workpiece withvmmm=s to the normal direction; dmm is the distance to bemoved. On the other hand, the task mode change fromnon-contact mode to contact mode is switched byCONTACT=v,(26)where vmm=s is the approaching velocity from the normaldirection. Due to the hyper CL data, it has been possible todetailedly record sanding skills.ARTICLE IN PRESSForce FeedbackControlLawPositionFeedbackControlLawCartesian-BasedServo ControllerRobot Sander+Fdod (k)xd (k)x (k)F (k)xd (k) : desired positionod (k) : desired positionFd : desired polishing forcePosition/Orientation FeedforwardControl LawBased on CL DataSpSp : switch matrixvp(k)vt (k)vn (k)Fig. 10. Block diagram of the surface following controller implemented inthe robot sander.FORCE/1.0POWER/4.0CONTACT/2.0VELOCITY/50.0GOTO/20.0008,300.0000,-6.8004,0.0096482,0.0893400,0GOTO/32.1883,300.0000,-6.9395,0.0130900,0.0850632,0GOTO/44.3758,300.0000,-7.1206,0.0165358,0.0769387,0GOTO/56.5633,300.0000,-7.3438,0.0199903,0.0649396,0GOTO/68.7508,300.0000,-7.6090,0.0234558,0.0490319,0FORCE/0.000,0.000,2.000,0.000,0.000,0.000POWER/6.0VELOCITY/30.0.9959545.9962896.9968987.9976890.9985218Fig. 11. Example of proposed hyper CL data.F. Nagata et al. / Robotics and Computer-Integrated Manufacturing 23 (2007) 3713793764. Experimental resultsIn this section, experimental results of surface sandingexperiment are given by using the proposed robot sander.The orbital sanding tool is widely used by skilled workersto sand or finish a workpiece with curved surface. The baseof the orbital sanding tool can perform eccentric motion.That is the reason that the orbital sander is not only apowerful sanding tool but also gives good surface finishingwith less scratches. In this experiment, an orbital sandingtool is selected and attached to the tip of the robot arm viaa force sensor. The size of the base and the eccentricity are75 ? 110mm and 4mm, respectively. The weight of thesanding tool is 800g. When a sanding task is conducted, acircular pad with a sanding paper is attached to the base.Fig. 12 shows an example of the target workpieces afterNC machining, which is a representative shape that theconventional sanding machines cannot sand sufficiently.The pick feed in the NC machining is set to 3mm. Theworkpiece before sanding shown in Fig. 12 has undesirablecusp marks higher than 1mm every pick feed. The robotsander first removed the cusp marks using a rough sandingpaper 80, then sanded the surface using a sanding paperwith a middle roughness 220 and finally a smooth paper400. The diameters of the pad and paper were cut to65mm, which were so larger than that of the ball-end mill(17mm) used in the NC machining process. Therefore, weregenerated the CL data with a pick feed 15mm for therobotic sanding as shown in Fig. 9 and substituted the CLdata for the controller shown in Fig. 10. The contourwas made so as to be a small size with an offset 15mmto prevent the edge of the workpiece from over sanding.Table 1 shows the other sanding conditions and theparameters of the surface following controller shown inFig. 10. These semi-optimum values were found throughtrial and error.Fig. 13 shows the sanding scene using the robot sander.In this case, the polishing force was controlled as shown inFig. 14. Handy air-driven tools are usually used by skilledworkers to sand wooden material constructing furniture.These types of tools cause large noise and vibration.Further, the system of force control consists of anARTICLE IN PRESSFig. 12. An example of curved workpiece that conventional sandingmachines cannot deal with.Table 1Sanding conditions and control parametersConditions or parametersValuesWorkpieceJapanese oakSize (mm)1200 ? 425 ? 85Diameter of sand paper (mm)65Grain size of sandpaper80 ! 220 ! 400Desired polishing force Fd(kgf)1.0Feed rate kvtk (mm/s)30Pick feed of CL data (mm)15Air pressure of orbital sanding tool kgf=cm24.0Desired mass coefficient Md(kgf s2=mm)0.01Desired damping coefficient Bd(kgf s=mm)20Force feedback gain Kf1Integral control gain for polishing force Kfi0.001Switch matrix for weak coupling control Spdiag0;1;0Position feedback gain matrix Kpdiag0;0:01;0Integral control gain matrix Kifor positiondiag0;0:0001;0Sampling width Dt (ms)0.01Fig. 13. Sanding scene using an orbital sanding tool.ReferenceResponse2.01.00020406080100120140160Time (s)Polishing force (kgf)Fig. 14. An example of time history of polishing force.F. Nagata et al. / Robotics and Computer-Integrated Manufacturing 23 (2007) 371379377industrial robot, force sensor, attachment, handy air-driventool, zig and wooden material. Because each of them hasstiff property, it is not easy to strictly keep the polishingforce a constant value without overshoot and oscillation.That is the reason why measured value of the polishingforce tends to have spikes and noise as shown in Fig. 14.However, the result would be much better than the one byskilled workers. Although it is so difficult and hard forskilled workers to simultaneously keep the polishing force,tool position and orientation to be desired situation evenfor a few minutes, the robot sander can perform the taskmore uniformly and perseveringly.Another sanding scene of the wooden table with curvedsurface is also shown in Fig. 15. The touch feelings withboth the fingers and the palm were very satisfactory.Undesirable cusp marks were not observed at all. And also,there was no over sanding around the edge of theworkpiece and no swell on the surface. Furthermore, weconducted a quantitative evaluation by using a stylusinstrument, so that the measurements obtained by thearithmetical mean roughness (Ra) and max height (Ry)were around 1 and 3mm, respectively.Figs. 16 and 17 show new designed furniture using theworkpiece sanded by the robot sander. It was confirmedfrom the experimental results that the proposed roboticsanding system could successfully sand the woody work-pieces with curved surface.5. Conclusions and future workThis paper has dealt with a robotic sanding system forattractively designed furniture with free-formed surface.The system has been developed using an industrial robotwith an open architectural controller. The system has twonovel features. One is that the surface following controllerdevelopedforroboticsandingallowstherobottosimultaneously realize a delicate polishing force controland smooth position/orientation control. The other is thatconventional and complicated teaching task is not requiredat all. When conventional sanding systems based on anindustrial robot are used, desired trajectory for toolsposition/orientation control is obtained through compli-cated teaching process. In this case, operator has to input alarge number of teaching points along objects surface.However, the proposed robot sander directly handles CLdata generated from 3D CAD/CAM, so that such ateaching process can be omitted. Experimentally, thesanding tool attached to the tip of the robot arm couldsmoothly follow the curved surface, keeping a contact fromnormal direction with a desired polishing force. It was alsodemonstrated that the proposed system could successfullysand the curved surface of attractive furniture withextremely good quality.The process parameters such as sanding conditions andcontrol gains given in the experiments were tuned with trialand error in consideration of efficiency and sanded surfacequality, and worked well for the examples shown in thispaper. However, if the robot sander is applied to otherdifferent materials, desirable parameters (i.e., semi-opti-mized parameters) should be found again. Because thedesirable parameters tend to largely depend on the speciesof wooden material. In the future, we plan to accumulate adatabase that has promising relations between the speciesARTICLE IN PRESSFig. 15. Sanding scene of wooden table with curved surface.Fig. 16. Bench with curved surface.Fig. 17. Table with curved surface.F. Nagata et al. / Robotics and Computer-Integrated Manufacturing 23 (2007) 371379378of wooden material and desirable parameters. Though theparameters should be further tuned time-varyingly anddelicately in compliance with intra species variations aswell as directionality and changes in a single workpiece soas to realize an ideal robot sander, no practical on-lineassessment techniques for wooden materials seem to beproposed yet at this stage.References1 Ozaki F, Jinno M, Yoshimi T, Tatsuno K, Takahashi M, Kanda M,et al. A force controlled finishing robot system with a task-directedrobot language. J Robotics Mechatronics 1995;7(5):3838.2 Pfeiffer F, Bremer H, Figueiredo J. Surface polishing with flexiblelink manipulators. Eur J Mech, A/Solids 1996;15(1):13753.3 Takeuchi Y, Ge D, Asakawa N. Automated polishing process with ahuman-like dexterous robot. Proceedings of IEEE internationalconference on robotics and automation; 1993. p. 9506.4 Pagilla PR, Yu B. Robotic su
- 温馨提示:
1: 本站所有资源如无特殊说明，都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
2: 本站的文档不包含任何第三方提供的附件图纸等，如果需要附件，请联系上传者。文件的所有权益归上传用户所有。
3.本站RAR压缩包中若带图纸，网页内容里面会有图纸预览，若没有图纸预览就没有图纸。
4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
5. 人人文库网仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对用户上传分享的文档内容本身不做任何修改或编辑，并不能对任何下载内容负责。
6. 下载文件中如有侵权或不适当内容，请与我们联系，我们立即纠正。
7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。
人人文库网所有资源均是用户自行上传分享，仅供网友学习交流，未经上传用户书面授权，请勿作他用。
川公网安备: 51019002004831号