外文文献.pdf_第1页
外文文献.pdf_第2页
外文文献.pdf_第3页
外文文献.pdf_第4页
外文文献.pdf_第5页
已阅读5页,还剩3页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

Proceedings of the IEEE-RAS International Conference on Humanoid RobotsCopyright 2001Mechanical System and Control System of a DexterousRobot HandDirk Osswald, Heinz WrnUniversity of KarlsruheDepartment of Computer ScienceInstitute for Process Control and Robotics (IPR)Engler-Bunte-Ring 8 - Building 40.28D-76131 Karlsruheemail: osswaldira.uka.de , woernira.uka.deAbstract: In recent years numerous robot systemswith multifingered grippers or hands have beendeveloped all around the world. Many differentapproaches have been taken, anthropomorphic andnon-anthropomorphic ones. Not only themechanical structure of such systems wasinvestigated, but also the necessary control system.With the human hand as an exemplar, such robotsystems use their hands to grasp diverse objectswithout the need to change the gripper. The specialkinematic abilities of such a robot hand, like smallmasses and inertia, make even complexmanipulations and very fine manipulations of agrasped object within the own workspace of thehand possible. Such complex manipulations are forexample regrasping operations needed for therotation of a grasped object around arbitraryangles and axis without depositing the object andpicking it up again. In this paper an overview onthe design of such robot hands in general is given,as well as a presentation of an example of such arobot hand, the Karlsruhe Dexterous Hand II. Thepaper then ends with the presentation of some newideas which will be used to build an entire newrobot hand for a humanoid robot using fluidicactuators.Keywords: Multifingered gripper, robot hand, finemanipulation, mechanical system, control system1 IntroductionThe special research area Humanoid Robotsfounded in Karlsruhe, Germany in July 2001 isaimed at the development of a robot system whichcooperates and interacts physically with humanbeings in normal environments like kitchen orliving rooms. Such a robot system which isdesigned to support humans in non-specialized,non-industrial surroundings like these must, amongmany other things, be able to grasp objects ofdifferent size, shape and weight. And it must alsobe able to fine-manipulate a grasped object. Suchgreat flexibility can only be reached with anadaptable robot gripper system, a so calledmultifingered gripper or robot hand.The humanoid robot, which will be built in theabove mentioned research project, will be equippedwith such a robot hand system. This new hand willbe built by the cooperation of two institutes, theIPR (Institute for Process Control and Robotics) atthe University of Karlsruhe and the IAI (Institutefor Applied Computer Science) at the KarlsruheResearch Center. Both organizations already haveexperience in building such kind of systems, butfrom slightly different points of view.The Karlsruhe Dexterous Hand II (see figure 1)built at the IPR, which is described here in detail, isa four fingered autonomous gripper. The handsbuilt at the IAI (see figure 17) are built asprosthesis for handicapped people.The approach taken so far will be presented anddiscussed in the following sections, as it founds thebasis for the novel hand of the humanoid robot.Figure 1: Karlsruhe Dextrous Hand II from IPRProceedings of the IEEE-RAS International Conference on Humanoid RobotsCopyright 20012 General structure of a robot handA robot hand can be split up in two majorsubsystems: The mechanical system The control systemThe mechanical system, further described in section3, can be subdivided into:- The mechanical design- The actuator system- The sensor systemAnd the control system described in section 4consists at least of :- The control hardware- The control softwareFor each of these parts we will describe theconsiderations for a robot hand in general and thenpresent the exemplary implementation in theKarlsruhe Dexterous Hand II.3 Mechanical systemThe mechanical system describes how the handlooks like and what kind of components it is madeof. It defines the mechanical design, e.g. thenumber of fingers and the kind of materials used.Additionally actuators, e.g. electric motors, andsensors, e.g. position encoders, are settled.3.1 Mechanical designThe mechanical design determines the fundamentaldexterousness of the hand, i.e. what kind ofobjects can be grasped and what kind ofmanipulations can be performed with a graspedobject. Three basic aspects have to be settled whendesigning a robot hand: The number of fingers The number of joints per finger The size and placement of the fingersTo be able to grasp and manipulate an object safelywithin the workspace of the hand at least 3 fingersare required. To achieve the full 6 degrees offreedom (3 translatory and 3 rotatory DOF) for themanipulation of a grasped object at least 3independent joints are needed for each finger. Thisapproach was taken for the first KarlsruheDexterous Hand 1,2. However, to be able toregrasp an object without having to release it andthen pick it up again, at least 4 fingers arenecessary.To determine the size and the placement of thefingers two different approaches can be taken: Anthropomorphic Non-anthropomorphicIt then depends on the objects to manipulate and onthe type of manipulations desired which one ischosen. An anthropomorphic placement allows toeasily transfer e.g. grasp strategies from a humanhand to the robot hand. But the different sizes ofeach finger and their asymmetric placement makesthe construction more expensive and the controlsystem more complicated, because each finger hasto be treated separately.When a non-anthropomorphic approach is takenmost often identical fingers are arrangedsymmetrically. This reduces the costs for theconstruction and simplifies the control systembecause there is only one single finger module tobe constructed and controlled.3.2 Actuator systemThe actuation of the finger joints also has a greatinfluence in the dexterousness of the hand, becauseit determines the potential forces, precision andspeed of the joint movements. Two differentaspects of the mechanical movement have to beconsidered: Movement generation Movement forwardingSeveral different approaches for these aspects aredescribed in the literature. E.g. the movement canbe generated by hydraulic or pneumatic cylinders3 or, as in most cases, by electric motors.As the movement generators (motors) are in mostcases to big to be integrated in the correspondingfinger joint directly, the movement must beforwarded from the generator (most times locatedin the last link of the robot arm) to the finger joint.Again different methods can be used, like tendons4,5,6, drive belts 1,2 or flexible shafts. The useof such more or less indirect actuation of the fingerjoint reduces the robustness and the precision of thesystem and it complicates the control systembecause different joints of one finger are oftenmechanically coupled and must be decoupled insoftware by the control system. Due to thesedrawbacks an integration of miniaturizedmovement generators directly into the finger jointsis desirable.3.3 Sensor systemThe sensor system of a robot hand provides thefeedback information from the hardware back tothe control software. This is necessary to perform aclosed loop control of the fingers or a graspedobject. Three types of sensors are used in robothands 7,8: Gripper state sensors determine the position ofthe finger joints, and hence the finger tip, andthe forces which act upon the finger. Knowingthe exact position of the fingertip makes preciseposition control possible, which is necessary fordexterous fine manipulations. With theknowledge of the forces applied to a graspedobject by the fingers it is possible to grasp afragile object without breaking it.Proceedings of the IEEE-RAS International CCopyright 2001 Grasp state sensors provide information aboutthe contact situation between the finger and theobject. This tactile information can be used todetermine the point in time of the first contactwith the object while grasping, and to avoidundesired grasps, like grasping at an edge or atip of the object. But it can also be used todetect slippage of an already grasped object,which might lead to a loss of the object. Object state or pose sensors are used todetermine the shape, position and orientation ofan object in the workspace of the gripper. Thisis necessary if these data is not known exactly,prior to grasping the object. If the object statesensors still works on a grasped object it can beused to control the pose (position andorientation) of a grasped object too, e.g. todetect slippage.Depending on the actuator system the geometricalinformation about the finger joint position can bemeasured at the movement generator or directly atthe joint. For example if there is a stiff couplingbetween an electric motor and the finger joint thenthe joint position can be measured by an angleencoder at the axis of the motor (before or after thegear). This is not possible if the coupling is lessstiff and a high position precision is desired.3.4 The mechanical system of the KarlsruheDexterous Hand IIIn order to permit more complex manipulations likeregrasping the current Karlsruhe Dexterous Hand II(KDH II) was built with 4 fingers and 3independent joints per finger. It is designated forapplications in industrial environments (seefigure 2) and for manipulation of objects likeboxes, cylinders, screws or nuts. Therefore asymmetric, non-anthropomorphic configuration offour identical fingers, each rotated by 90 waschosen (see figure 3).Due to the experiences gained with the firstKarlsruhe Dexterous Hand, like e.g. mechanicalproblems caused by the drive belts or controllingproblems caused by large friction factors, somedifferent design decisions were chosen for theKDH II. The dc-motors for joint 2 and 3 of eachfinger are integrated into the previous finger limb(see figure 4). This permits the use of very stiffball-spindle-gears for the forwarding of themovement to the finger joint. Angle encodersdirectly on the motor axis (before the gear) are usedas very precise position state sensors.Figure 4: Side view of the KDH II3 lasersensorsfixation frameone complete fingercontrol hardware(microcontroller)Figure 3: Top view of the KDH IIFigure 2: KDH II mounted on an industrial robotonference on Humanoid RobotsFor sensing the forces applied to an object by afinger a prototype of a 6 dimensional force torquesensor has been developed (see figure 5). It can beused as the last finger limb and is equipped with aspherical finger tip. It is able to grasp light objectsas well as relatively heavy objects up to 3 to 5 kg.The sensor is able to measure forces in x- y- and z-direction and torques around these axes.Additionally 3 colinear laser triangulation sensorsare mounted in the palm of the KDH II (seefigure 4) 11. Because there are three such sensorsnot only the distances of 3 single points can bemeasured, but also the distance and orientation ofthe surface of a grasped object, if the shape of theobject is known. This object pose sensor worksTo cope with the requirements the control hardwareis usually distributed among several specializedprocessors. For example the input/output on thelowest level (motors and sensors) can be handledby a simple microcontroller, which is also of smallsize and thus can be integrated more easily into themanipulation system. But the higher levels ofcontrol need more computing power and thesupport of a flexible real time capable operatingsystem. This can be achieved most easily with PC-like components.Therefore the control hardware often consist of anon-uniform, distributed computing system withmicrocontrollers on the one end and more powerfulprocessors on the other. The different computingunits then have to be connected with acommunication system, like for example a bussystem.4.2 Control softwarewith a frequency of 1 kHz which allows thedetection and avoidance of a slipping object.4 Control systemThe control system of a robot hand determineswhich of the potential dexterous skills provided bythe mechanical system can actually be exploited.As mentioned before the control system can besubdivided in the control computer or hardware andthe control algorithms or software.The control system must meet several conflictingrequirements:Proceedings of the IEEE-RAS InterCopyright Many input/output resources like actor orsensor signals must be attached. For examplefor a minimum hand with 9 degrees of freedom,at least 9 analog outputs to the motors and 9inputs from angle encoders must be estimated.With force and tactile sensors for every fingerand additional object state sensors the numberof inputs quickly increases to several dozens. Quick reactions in real-time to external eventsare required. If for example a slipping of thegrasped object is detected immediate countermeasures must be taken. High computing power for several differenttasks must be available. For example pathplanning, coordinate transformations, closedFigure 5: 6 DOF force torque sensor with straingage sensors used as the last finger limbof the KDH IIloop control in software are executed in parallelfor multiple fingers as well as for the object. Small physical size is needed to be able tointegrate the control system into themanipulation system. Short electrical connections between the controlsystem and the actuators and sensors should beused. This is especially relevant for the sensorsbecause otherwise massive interference mightdisturb the sensor signal.4.1 Control hardwarenational Conference on Humanoid Robots2001The control software of a robot hand is quitecomplex. Several fingers must be controlled in real-time and in parallel while new trajectories for thefingers and the object must be planned at the sametime. Therefore it is necessary to reduce thecomplexity by dividing the problem into subproblems.Another aspect concerns software development. Asa robotic hand is usually a research project for mostof its lifetime, the programming environment, likeuser interface, programming tools and debuggingfacilities, should be powerful and flexible. This canonly be achieved if a standard operating system isused.Figure 6: control hardware architecture of the KDH IITun4.3AtDexmsenm(mrthanare conseTgasPC2mfofiOnOncalcuismthTconfiOnsobject movement commands are received from thesuperior robot control system and used for a fineplanning of the object path. According to thegenerated object path feasible grasps (possiblegrasp points for fingers on the object) are planned.Now that the grasps and the object movements areknown the trajectories for each finger are plannedby the finger path planning and forwarded to thereal-time capable part of the system.If an object is already grasped, then the fingermovement paths are forwarded to the object posecontroller. This controller controls the actual objectpose, determined by the gripper and object statesensors, to reach the desired object pose. If a fingeris not attached to an object, then its movement pathis forwarded directly to the hand controller.The hand controller coordinates the movements ofall fingers by forwarding correspondent desiredfinger positions to all finger controllers. These inturn drive the finger actuators with the help of thefinger sensors.Fighe usual hierarchical control system approachsed in robotics has to be trimmed to fit the specialeeds of the controlling of a robot hand.The control system of the KarlsruheDexterous Hand IIs suggested in section 4.1 a distributed approacho the control hardware was taken for the Karlsruheterous Hand II (KDH II) (see figure 6) 8. Oneicrocontroller is used to control the actuators andsors of one finger respectively. An additionalicrocontroller is used for the object state sensorlaser triangulation sensors). Theseicrocontrollers (the outer boxes to the left andight in figure 6) are mounted directly on the hand,us short electrical connections to the actuatorsd sensors are guaranteed. The microcontrollersnected to the main control computer byrial bus systems (CAN-bus).he main control computer of the KDH II (the lightrey box in figure 6, and figure 7) is implementeda parallel computer consisting of 6 industrialProceedings of the IEEE-RAS InterCopyright s (PC104 standard). These PCs are arranged in aD-plane. Neighboring PC-modules (a PC has atost 8 neighbors) use a dual ported RAM (DPR)r fast communication (the dark grey boxes ingure 6).e PC is used to control a finger respectively.e PC controls the object state sensors andlates the objects position. The remaining PCplaced such that it neighbors all the previouslyentioned PCs. It is used for the coordination ofe whole control systemhe structure of the control software reflects thetrol hardware architecture. It is shown ingure 8.the three top levels of the local hand controlystem an on-line planning is performed. Desiredure 7: parallel main computer used to control theKDH IIobjectsensorshand controller (finger coordination)object pose controllerlocal grasp planning.local object path planninglocal handcontrol systemFingfinger 1sensorsfinger 1actuatorsfinger controller 1onlineplanningonlinecontrolling (real-time)finger path planningobject movement free movementFingfinger nsensorsfinger nactuatorsfinger controller nFigure 8: local hand control system of theKarlsruhe Dextrous Hand IInational Conference on Humanoid Robots20015 Experimental resultsTo validate the capabilities of the KarlsruheDexterous Hand II two demanding manipulationproblems were chosen. One problem is the on-linecontrolling of the pose (position and orientation) ofa grasped object under external influences. Herethe hard real-time conditions reveal the controllingcapabilities of the approach chosen.For the other problem a grasped object must berotated around arbitrary angles, which can only beachieved with regrasping. This reveals thecapability of the Karlsruhe Dexterous Hand II toperform very complex manipulation tasks.5.1 Object pose controlThe objective of the object pose controller is tocorre

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。

评论

0/150

提交评论