機(jī)械手外文翻譯---機(jī)械手的機(jī)械和控制系統(tǒng)

1、<p>　　中文6400字，3700單詞，18400英文字符</p><p>　　出處：Osswald D, Wrn H. Mechanical System and Control System of a Dexterous Robot Hand[C]// 2004.</p><p>　　本科畢業(yè)設(shè)計(jì)(論文)</p><p>　　外文翻譯（附外文原文）&

2、lt;/p><p>　　學(xué) 院： 機(jī)械與控制工程 </p><p>　　課題名稱： </p><p>　　專業(yè)(方向)： 機(jī)械設(shè)計(jì)制造及其自動(dòng)化 </p><p>　　班 級(jí)：

3、 </p><p>　　學(xué) 生： </p><p>　　指導(dǎo)教師： </p><p>　　日 期： </p><p><

4、;b>　　外文原文：</b></p><p>　　Mechanical and control system for Manipulators</p><p>　　D Osswald，H Wrn</p><p>　　Abstract: Recently, in the world with a clip or a hand robot system

5、 has been developed, a variety of methods is applied on the, quasi humanized and non-personification. Not only the mechanical structure of these systems is investigated, but also the necessary control system is also incl

6、uded.. As the staff, these robots can use their hands to grasp different objects, without changing the clip. These manipulators possess special athletic abilities (such as small mass and inertia), which enable the</p&

7、gt;<p>　　Keywords: multi robot manipulator; robot hand; finishing operation; mechanical system; control system</p><p>　　1 .Introduction</p><p>　　In June 2001 in Karlsruhe, Germany to carr

8、y out special study a humanoid robot, in order to develop in a normal environment (such as kitchen or the living room) and human cooperation and interaction of the robot system. The design of these robots is designed to

9、help us capture objects of size, shape and weight in a non -professional, non - industrial condition, such as in many objects. At the same time, they must be able to manipulate the object very well. This flexibility can

10、only be through a </p><p>　　The research project mentioned above is to create a humanoid robot, which will equip this robot hand system.. This novice will be produced by two organizations, which are IPR (pro

11、cess control and Robotics Research Institute) and C (Computer Science Institute), University of Karlsruhe.. These two organizations have the experience of making such systems, but slightly different views.</p><

12、;p>　　IPR made Karlsruhe dexterous hand II (as shown in Figure 1), is a four finger gripper is independent of each other, we will be introduced in detail in this paper. The hand made by IAI (as shown in Figure 17) is u

13、sed as prosthetic for the disabled.</p><p>　　Figure 1.IPR Karlsruhe smart hand Figure 2 fluid hand developed by IAI</p><p>　　2 .general structure of robot hand</p><p>　　A r

14、obot hand can be divided into two major subsystems: mechanical systems and control systems.</p><p>　　The mechanical system can be divided into the structure design, the drive system and the sensor system, we

15、 will further introduce in the third part. In the fourth part of the introduction of the control system at least by the control of hardware and software components.</p><p>　　We will be on the two system prob

16、lems of a basic introduction, and then use the Karlsruhe dexterous hand II demonstration.</p><p>　　3 .Mechanical Systems</p><p>　　The mechanical system will describe how the hand looks and what

17、components. It determines the structure design, the number of fingers and the use of materials. In addition, the position of the actuator (such as the motor) and the sensor (e.g. position encoder) is also determined.<

18、/p><p>　　3.1 structure design</p><p>　　The structure design will have the very important function to the manipulator's flexibility, namely it can grasp which kind of object and can carry on to

19、the object to carry on what kind of operation. When designing a robot hand, three basic elements must be determined: the number of fingers, the number of fingers, the size and placement of the fingers..</p><p&

20、gt;　　In order to crawl and operate the object safely within the manipulator, at least three fingers. In order to operate the object being grasped for 6 degrees of freedom (3 translational and 3 rotational degrees of free

21、dom), each finger must have 3 separate joints.. This method was used in the first generation of Karlsruhe's smart hand.. However, in order to catch an object without the need to release it first to pick up, at least

22、4 fingers.</p><p>　　Two methods: the human and the non - human are to determine the size and placement of the finger.. Then it will depend on the object and the type of operation to which the desired operati

23、on is selected. It is easy to transfer grasping intention from hand to robot hand.. However, the placement of different sizes and asymmetric positions of each finger will increase the processing cost, and it is the contr

24、ol system becoming more complicated, because each finger must be controlled separately. For t</p><p>　　3.2 drive system</p><p>　　The flexibility of the actuator is also greatly affected by the d

25、rive of the joints, because it determines the potential strength, precision and speed of the joint movement.. Two aspects of the mechanical movement need to be considered: the movement source and the movement direction.

26、In this case, there are several different methods, such as the paper [3], which can be produced by hydraulic cylinders or pneumatic cylinders, or, as most of the case, the motor is used.. In most cases, motor drive,</

27、p><p>　　3.3 sensing system</p><p>　　The sensing system of the robot hand can transfer the feedback information from the hardware to the control software.. It is necessary to establish a closed loop

28、 control for finger or object.. 3 types of sensors were used in the machine.</p><p>　　1）Hand state sensors determine the position of fingertip and finger joint and finger force situation. Know the precise po

29、sition of the fingertips will make precise control possible. In addition, knowing that the finger is the force that is grasped at the object, you can grab a fragile object without breaking it.</p><p>　　2）The

30、 grasping state sensor provides the contact information between the finger and the object. This kind of tactile information can be determined in the process of grasping the first contact with the object in time, and can

31、also avoid incorrect grasping, such as the edge and tip of the object.. It can also detect whether the object has been caught, so as to avoid the object due to fall and damage.</p><p>　　3）The object or pose

32、sensor is used to determine the shape, position and direction of the object in a finger. This sensor is very essential if it is not clear to the case of the object.. If this sensor can also act on the object that has bee

33、n grasped, it can also control the pose (position and direction) of the object, thus monitoring whether or not.</p><p>　　Depending on the drive system, the geometrical information about the joint position ca

34、n be measured at a motion drive or directly at the joint. For example, if there is a rigid shaft coupling between the motor and the knuckle, then the position of the joint can be measured by a motor shaft (before the gea

35、r or after the gear).. However, if this coupling stiffness is not enough or to get a high accuracy, it can not use this method.</p><p>　　3.4 the mechanical system of the robot hand in Karlsruhe</p>&l

36、t;p>　　In order to obtain more complex operation such as heavy grasping, the Karlsruhe smart hand II (KDH II) is composed of 4 fingers, and each finger is composed of 3 independent joints.. The hand is designed for app

37、lications in industrial environments (Figure 3) and a control box, cylinder and screw nut and other objects. Therefore, we selected four identical fingers. They are symmetric, non-anthropomorphic configuration and each f

38、inger can rotate 90 DEG (Fig. 4).</p><p>　　View from the first generation of Karlsruhe dexterous hand design by experience, for example, the problem of mechanical caused by the drive belts and larger frictio

39、n factor leading to the control problem of Karlsruhe dexterous hand II uses a number of different design decisions. The DC motor between the joint 2 and the joint of each finger is integrated into the anterior part of th

40、e finger (Figure 5).This arrangement can be used with hard ball gears to transmit motion to the joints of the finger</p><p>　　Figure 3 KDH II on industrial robot Figure 4 KDH II top view </p&g

41、t;<p>　　In order to perception of the role of finger force on the object, we invented a six axis force torque sensor (Figure 6). The sensor can be used as a fingertip for the end of a finger and is equipped with a

42、 spherical fingertip.. It can grab lighter objects, but also can grab 3-5kg similar heavier objects. This sensor can measure the force of the direction of X, Y and Z and the torque of the winding axis.. In addition, the

43、laser triangulation sensor 3 collinear is placed in the hands of KDH II (Fi</p><p>　　Figure 5 KDH II side view. Figure six a 6 degree of freedom torsion sensor with a strain gage sensor</p><p>

44、;　　4 control system</p><p>　　The robot's control system determines which potential dexterity skills can actually be used, and those skills are provided by mechanical systems.. As mentioned before, the co

45、ntrol system can be divided into the control computer, namely, the hardware and the control algorithm is the software.</p><p>　　The control system must meet the following conditions:</p><p>　　1)

46、 Must have sufficient input and output ports. For example, a low level hand with 9 degrees of freedom, its drive needs at least 9 way to simulate the output port, and there are 9 paths from the angle encoder input port.

47、Such as force sensor, tactile sensor and object sensor, then port number will be increased by several times.</p><p>　　2) The ability to have a quick and real-time response to external events. For example, wh

48、en the detected object falls, the corresponding measures can be taken immediately.</p><p>　　3) With a higher computational power to address some of the different tasks. Such as path planning, coordinate conv

49、ersion, and closed-loop control for multi - object and object - parallel execution.</p><p>　　4) The volume of the control system is small so that it can be integrated directly into the operating system..<

50、/p><p>　　5) In the control system and between the drive and sensor must be electrically short. Especially for the sensor, if there is no word, a lot of interference will interfere with the sensor signal.</p&

51、gt;<p>　　4.1 control hardware</p><p>　　In order to meet the requirements of the system, the hardware is distributed in several special processors.. The controller can be easily integrated into the ope

52、rating system, such as the low input output interface (motor and sensor), which is handled by a simple microcontroller.. However, the higher level of the control port requires a higher computing power, and a flexible rea

53、l-time operating system is needed.. This can be easily resolved through the PC.</p><p>　　Therefore, the control hardware is often composed of a distributed computer system, which is a microcontroller, and th

54、e other is a powerful processor. Different computing units are connected by a communication system, such as bus system.</p><p>　　4.2 control software</p><p>　　Robot hand control software is quit

55、e complex. Must be real-time and parallel control of the fingers, but also plan the new trajectory of the fingers and objects. Therefore, in order to reduce the complexity of the problem, it is necessary to divide this p

56、roblem into several sub problems to deal with..</p><p>　　On the other hand, software development.. Robot hand is a research project, its programming environment such as user interface, programming tools and

57、debugging facilities must be very strong and flexible. These can only be met using a standard operating system. The hierarchical control system method is widely used in the robot after pruning, in order to meet the speci

58、al requirements of the manipulator control.</p><p>　　4.3 control system of the robot hand in Karlsruhe</p><p>　　As said in Section 4.1, a distributed method (Figure 7) is adopted for the hardwar

59、e of the control hardware of the smart hand of Karlsruhe.. A microcontroller controls a finger drive and sensor respectively, and a microcontroller is used to control the object state sensor (laser triangulation).. These

60、 microcontrollers (Figure 7 the left and right side of the box) are directly mounted on the hand, so the shorter electrical connection between the driver and the sensor can be guaranteed. These micr</p><p>　

61、　Figure 7 II KDH control hardware architecture Figure 8 parallel master computer for controlling II KDH</p><p>　　A three maximum levels of online planning regarding this hand control system are bei

62、ng planned. The ideal object displacement command can be obtained by the superior robot control system and can be used as the precise programming of the object path.. According to the target path, the feasible fetching b

63、ehavior of the finger can be planned (the feasible grasping position of the object) is feasible.. Now that the object movement plan can be obtained by finger path planning, and the real-time capab</p><p>　　F

64、igure 9 hand control system for KDH II5 experimental results</p><p>　　To verify the ability of the Karlsruhe smart hand, we chose two requirements for operation.. One problem is the control of the pose (posi

65、tion and direction) of the object under the influence of the Internet.. Another problem is that the object must be able to rotate around any angle, which can only be realized by heavy catch.. This can reflect the operati

66、on ability of the complex task of the robot hand in Karlsruhe..</p><p>　　5.1 object attitude control</p><p>　　The object of the attitude controller is to determine the position and direction of

67、the object to fit the given trajectory.. This task must be obtained in real-time conditions, despite the presence of internal changes and external disturbances.. Internal changes such as the rolling of the ball fingers o

68、n the object when the object is moving. This situation is shown in Figure 10, figure 11. This will cause unwanted additional movement and tilt of the object. The pose of these errors is hard to es</p><p>　　F

69、igure 10 additional displacement due to rolling Figure 12 No state controlled object tilt</p><p>　　Figure 11 an additional undesired tilt due Figure 13 object state control to reduce </p>

70、<p>　　to the rolling of a ball finger tip over an object the object Tilt</p><p>　　The object state controller is also necessary for the compensation of external interference.. For example, the robot

71、 (arm, hand or finger) or the collision between objects and the outside may cause the slide of the object. This is more likely to lead to the loss of the object, which is not the case. In order to avoid the loss of an ob

72、ject in this case, it is necessary to detect the slide of the object and act quickly to stabilize the object..</p><p>　　In order to validate the Karlsruhe dexterous hand II control system of this interferenc

73、e processing capability, we do the following experiments: object to be caught, the finger contact force constant is reduced until the object began to fall. After the laser triangulation sensor detects the slide, the obje

74、ct state controller takes measures to re regulate the object to the desired position.. Figure 15 shows an example of this experiment.. In particular, Figure 14, which shows that the object fall</p><p>　　Figu

75、re 14: the actual object position of the X direction Figure 15 slide experiment: the actual object direction of the Z axis</p><p><b>　　5.2 catch</b></p><p>　　Although the Karlsruh

76、e smart hand is very flexible, it can not get every ideal object manipulation in the first operation.. This stems from the fact that fingers are small relative to the normal industrial robot, so the working range is very

77、 limited.. If the object is caught by fingers, it can be manipulated for the first time in the remainder of all fingers.. The condition of the feasible operation is that all the contact points must be in the working rang

78、e of the associated finger for a long t</p><p>　　Figure 16 the catch rotating nut shaped object Figure 17 the catch is pulled out from the holes in the wood</p><p>　　6 conclusions</p>

79、<p>　　In order to enable the manipulator to perform the flexible and accurate operation, the appropriate mechanical system and the control system are required.. The criteria for the introduction of the standards are

80、 considered, as the text said. Karlsruhe smart hand II is very successful. This manipulator is capable of capturing a large range of objects of different shapes, sizes and weights.. The attitude of the object being grasp

81、ed can also be reliably controlled, even in the case of external interf</p><p>　　機(jī)械手的機(jī)械和控制系統(tǒng)</p><p>　　摘要: 最近，全球內(nèi)帶有多指夾子或手的機(jī)械人系統(tǒng)已經(jīng)發(fā)展起來(lái)了，多種方法應(yīng)用其上，有擬人化的和非擬人化的。不僅調(diào)查了這些系統(tǒng)的機(jī)械結(jié)構(gòu),而且還包括其必要的控制系統(tǒng)。如同人手一樣，這些機(jī)

82、械人系統(tǒng)可以用它們的手去抓不同的物體，而不用改換夾子。這些機(jī)械手具備特殊的運(yùn)動(dòng)能力（比如小質(zhì)量和小慣性），這使被抓物體在機(jī)械手的工作范圍內(nèi)做更復(fù)雜、更精確的操作變得可能。這些復(fù)雜的操作被抓物體繞任意角度和軸旋轉(zhuǎn)。本文概述了這種機(jī)械手的一般設(shè)計(jì)方法，同時(shí)給出了此類機(jī)械手的一個(gè)示例，如卡爾斯魯厄靈巧手Ⅱ。本文末介紹了一些新的構(gòu)想，如利用液體驅(qū)動(dòng)器為類人型機(jī)器人設(shè)計(jì)一個(gè)全新的機(jī)械手。</p><p>　　關(guān)鍵詞：多指機(jī)

83、械手；機(jī)器人手；精操作；機(jī)械系統(tǒng)；控制系統(tǒng)</p><p><b>　　引言</b></p><p>　　2001年6月在德國(guó)卡爾斯魯厄開(kāi)展的“人形機(jī)器人”特別研究，是為了開(kāi)發(fā)在正常環(huán)境（如廚房或客廳）下能夠和人類合作和互動(dòng)的機(jī)器人系統(tǒng)。設(shè)計(jì)這些機(jī)器人系統(tǒng)是為了能夠在非專業(yè)、非工業(yè)的條件下（如身處多物之中），幫我們抓取不同尺寸、形狀和重量的物體。同時(shí)，它們必須能夠很好

84、的操縱被抓物體。這種極強(qiáng)的靈活性只能通過(guò)一個(gè)適應(yīng)性極強(qiáng)的機(jī)械人手抓系統(tǒng)來(lái)獲得，即所謂的多指機(jī)械手或機(jī)器人手。</p><p>　　上文提到的研究項(xiàng)目，就是要制造一個(gè)人形機(jī)器人，此機(jī)器人將裝備這種機(jī)器人手系統(tǒng)。這個(gè)新手將由兩個(gè)機(jī)構(gòu)合作制造，它們是卡爾斯魯厄大學(xué)的IPR（過(guò)程控制和機(jī)器人技術(shù)研究院）和c（計(jì)算機(jī)應(yīng)用科學(xué)研究院）。這兩個(gè)組織都有制造此種系統(tǒng)的相關(guān)經(jīng)驗(yàn)，但是稍有不同的觀點(diǎn)。</p><

85、p>　　IPR制造的卡爾斯魯厄靈巧手Ⅱ（如圖1所示），是一個(gè)四指相互獨(dú)立的手爪，我們將在此文中詳細(xì)介紹。IAI制造的手（如圖17所示）是作為殘疾人的假肢。</p><p>　　圖1.IPR的卡爾斯魯厄靈巧手Ⅱ 圖2. IAI開(kāi)發(fā)的流體手</p><p>　　2.機(jī)器人手的一般結(jié)構(gòu)</p><p>　　一個(gè)機(jī)器人手可以分成兩大主要子系統(tǒng)：機(jī)械

86、系統(tǒng)和控制系統(tǒng)。</p><p>　　機(jī)械系統(tǒng)又可分為結(jié)構(gòu)設(shè)計(jì)、驅(qū)動(dòng)系統(tǒng)和傳感系統(tǒng)，我們將在第三部分作進(jìn)一步介紹。在第四部分介紹的控制系統(tǒng)至少由控制硬件和控制軟件組成。</p><p>　　我們將對(duì)這兩大子系統(tǒng)的問(wèn)題作一番基本介紹，然后用卡爾斯魯厄靈巧手Ⅱ演示一下。</p><p><b>　　3.機(jī)械系統(tǒng)</b></p><

87、;p>　　機(jī)械系統(tǒng)將描述這個(gè)手看起來(lái)如何以及由什么元件組成。它決定結(jié)構(gòu)設(shè)計(jì)、手指的數(shù)量及使用的材料。此外，還確定驅(qū)動(dòng)器（如電動(dòng)機(jī)）、傳感器（如位置編碼器）的位置。</p><p><b>　　3.1 結(jié)構(gòu)設(shè)計(jì)</b></p><p>　　結(jié)構(gòu)設(shè)計(jì)將對(duì)機(jī)械手的靈活度起很大的作用，即它能抓取何種類型的物體以及能對(duì)被抓物體進(jìn)行何種操作。設(shè)計(jì)一個(gè)機(jī)器人手的時(shí)候，必須確定

88、三個(gè)基本要素：手指的數(shù)量、手指的關(guān)節(jié)數(shù)量以及手指的尺寸和安置位置。</p><p>　　為了能夠在機(jī)械手的工作范圍內(nèi)安全的抓取和操作物件，至少需要三根手指。為了能夠?qū)Ρ蛔ノ矬w的操作獲得6個(gè)自由度（3個(gè)平移和3個(gè)旋轉(zhuǎn)自由度），每個(gè)手指必須具備3個(gè)獨(dú)立的關(guān)節(jié)。這種方法在第一代卡爾斯魯厄靈巧手上被采用過(guò)。但是，為了能夠重抓一個(gè)物件而無(wú)需將它先釋放再拾取的話，至少需要4根手指。</p><p>　

89、　要確定手指的尺寸和安置位置，可以采用兩種方法：擬人化和非擬人化。然后將取決與被操作的物體以及選擇何種期望的操作類型。擬人化的安置方式很容易從人手到機(jī)器人手轉(zhuǎn)移抓取意圖。但是每個(gè)手指不同的尺寸和不對(duì)稱的安置位置將增加加工費(fèi)用，并且是其控制系統(tǒng)變得更加復(fù)雜，因?yàn)槊總€(gè)手指都必須分別加以控制。對(duì)于相同手指的對(duì)稱布置，常采用非擬人化方法。因?yàn)橹恍杓庸ず蜆?gòu)建單一的“手指模塊”，因此可減少加工費(fèi)用，同時(shí)也可是控制系統(tǒng)簡(jiǎn)化。</p>&

90、lt;p><b>　　3.2 驅(qū)動(dòng)系統(tǒng)</b></p><p>　　指關(guān)節(jié)的驅(qū)動(dòng)器對(duì)手的靈活度也有很大的影響，因?yàn)樗鼪Q定潛在的力量、精度及關(guān)節(jié)運(yùn)動(dòng)的速度。機(jī)械運(yùn)動(dòng)的兩個(gè)方面需加以考慮：運(yùn)動(dòng)來(lái)源和運(yùn)動(dòng)方向。在這方面，文獻(xiàn)里描述了有幾種不同的方法，如文獻(xiàn)[3]中說(shuō)可由液壓缸或氣壓缸產(chǎn)生運(yùn)動(dòng)，或者，正如大部分情況一樣使用電動(dòng)機(jī)。在多數(shù)情況下，運(yùn)動(dòng)驅(qū)動(dòng)器（如電機(jī)）太大而不能直接與相應(yīng)的指關(guān)節(jié)結(jié)

91、合在一起，因此，這個(gè)運(yùn)動(dòng)必須由驅(qū)動(dòng)器（一般位于機(jī)器臂最后的連接點(diǎn)處）轉(zhuǎn)移過(guò)來(lái)。有幾種不同的方法可實(shí)現(xiàn)這種運(yùn)動(dòng)方式，如使用鍵、傳動(dòng)帶以及活動(dòng)軸。使用這種間接驅(qū)動(dòng)指關(guān)節(jié)的方法，或多或少地降低了整個(gè)系統(tǒng)的強(qiáng)度和精度，同時(shí)也使控制系統(tǒng)復(fù)雜化，因?yàn)槊扛种傅牟煌P(guān)節(jié)常常是機(jī)械地連在一起，但是在控制系統(tǒng)的軟件里卻要將它們分別獨(dú)立控制。由于具有這些缺點(diǎn)，因此小型化的運(yùn)動(dòng)驅(qū)動(dòng)器與指關(guān)節(jié)的直接融合就顯得相當(dāng)必要。</p><p>

92、<b>　　3.3 傳感系統(tǒng)</b></p><p>　　機(jī)器手的傳感系統(tǒng)可將反饋信息從硬件傳給控制軟件。對(duì)手指或被抓物體建立一個(gè)閉環(huán)控制是很必要的。在機(jī)器手中使用了3種類型的傳感器：</p><p>　　1）手爪狀態(tài)傳感器確定指關(guān)節(jié)和指尖的位置以及手指上的作用力情況。知道了指尖的精確位置將使精確控制變得可能。另外，知道手指作用在被抓物體上的力，就可以抓取易碎物件而不

93、會(huì)打破它。</p><p>　　2）抓取狀態(tài)傳感器提供手指與被抓物體之間的接觸狀態(tài)信息。這種觸覺(jué)信息可在抓取過(guò)程中及時(shí)確定與物體第一次接觸的位置點(diǎn)，同時(shí)也可避免不正確的抓取，如抓到物體的邊緣和尖端。另外還能察覺(jué)到已抓物體是否滑落，從而避免物體因跌落而損壞。</p><p>　　3）物體狀態(tài)或姿態(tài)傳感器用于確定手指內(nèi)物體的形狀、位置和方向。如果在抓取物體之前并不清楚這些信息的情況下，這種傳感

94、器是非常必要的。如果此傳感器還能作用于已抓物體上的話，它也能控制物體的姿態(tài)（位置和方向），從而監(jiān)測(cè)是否滑落。</p><p>　　根據(jù)不同的驅(qū)動(dòng)系統(tǒng)，有關(guān)指關(guān)節(jié)位置的幾何信息可以在運(yùn)動(dòng)驅(qū)動(dòng)器或直接在關(guān)節(jié)處出測(cè)量。例如，如在電動(dòng)機(jī)和指關(guān)節(jié)之間有一剛性聯(lián)軸器，那么就可以用電機(jī)軸上的一個(gè)角度編碼器（在齒輪前或齒輪后）來(lái)測(cè)量關(guān)節(jié)的位置。但是如果此聯(lián)軸器剛度不夠或者要獲得很高的精度的話，就不能用這種方法。</p>

95、;<p>　　3.4卡爾斯魯厄靈巧手Ⅱ的機(jī)械系統(tǒng)</p><p>　　為了能夠獲得如重抓等更加復(fù)雜的操作，卡爾斯魯厄靈巧手Ⅱ（KDHⅡ）由4根手指組成，且每根手指由3個(gè)相互獨(dú)立的關(guān)節(jié)組成。設(shè)計(jì)該手是為了能夠在工業(yè)環(huán)境中應(yīng)用（圖3所示）和操縱箱、缸及螺釘螺帽等物體。因此，我們選用四個(gè)相同手指，將它們作對(duì)稱、非擬人化配置，且每個(gè)手指都能旋轉(zhuǎn)90°（圖4所示）。</p><p

96、>　　鑒于從第一代卡爾斯魯厄靈巧手設(shè)計(jì)中得到的經(jīng)驗(yàn)，比如因傳動(dòng)帶而導(dǎo)致的機(jī)械問(wèn)題以及較大摩擦因數(shù)導(dǎo)致的控制問(wèn)題，卡爾斯魯厄靈巧手Ⅱ采用了一些不同的設(shè)計(jì)決策。每根手指的關(guān)節(jié)2和關(guān)節(jié)3之間的直流電機(jī)被整合到手指前部肢體中（圖5所示）。這種布置可使用很硬的球軸齒輪將運(yùn)動(dòng)傳遞到手指的關(guān)節(jié)處。處在電機(jī)軸上的角度編碼器（在齒輪前）此時(shí)可作為一個(gè)精度很高的位置狀態(tài)傳感器。</p><p>　　圖3.工業(yè)機(jī)器人上的KDHⅡ

97、 圖4. KDHⅡ的頂視圖 </p><p>　　為了感知作用在物體上的手指力量，我們發(fā)明了一個(gè)六維力扭矩傳感器（圖6所示）。這個(gè)傳感器可當(dāng)作手指末端肢體使用，且配有一個(gè)球形指尖。它可以抓取較輕的物體，同時(shí)也能抓取3-5kg相近的較重物體。此傳感器能測(cè)量X、Y和Z方向的力及繞相關(guān)軸的力矩。另外，3個(gè)共線的激光三角測(cè)量傳感器被安置在KDHⅡ的手掌上（圖5所示）。因?yàn)橛?個(gè)這樣的傳

98、感器，因此不僅可以測(cè)量3單點(diǎn)之間的距離，如果知道物體的形狀，還能測(cè)出被抓物體表面之間的距離和方向。物體狀態(tài)傳感器的工作頻率為1kHz，它能檢測(cè)和避免物體的滑落。</p><p>　　圖5. KDHⅡ的側(cè)視圖 圖6. 帶應(yīng)變計(jì)量傳感器的六自由度扭轉(zhuǎn)傳感器</p><p><b>　　4. 控制系統(tǒng)</b></p><p>　　機(jī)器人手的控

99、制系統(tǒng)決定哪些潛在的靈巧技能能夠被實(shí)際利用，這些技能都是由機(jī)械系統(tǒng)所提供的。如前所述，控制系統(tǒng)可分為控制計(jì)算機(jī)即硬件和控制算法即軟件。</p><p>　　控制系統(tǒng)必須滿足以下幾個(gè)的條件：</p><p>　　1) 必須要有足夠的輸入輸出端口。例如，具有9個(gè)自由度的低級(jí)手，其驅(qū)動(dòng)器至少需要9路模擬輸出端口，且要有9路從角度編碼器的輸入端口。如再加上每個(gè)手指上的力傳感器、觸覺(jué)傳感器及物體狀態(tài)

100、傳感器的話，則端口數(shù)量將增加號(hào)好倍。</p><p>　　2) 需具備對(duì)外部事件快速實(shí)時(shí)反應(yīng)的能力。例如，當(dāng)檢測(cè)到物體滑落時(shí)，能立即采取相應(yīng)的措施。</p><p>　　3) 需具備較高的計(jì)算能力以應(yīng)對(duì)一些不同的任務(wù)。如可以對(duì)多指及物體并行執(zhí)行路徑規(guī)劃、坐標(biāo)轉(zhuǎn)換及閉環(huán)控制等任務(wù)。</p><p>　　4) 控制系統(tǒng)的體積要小，以便能夠?qū)⑵渲苯蛹傻讲僮飨到y(tǒng)當(dāng)中。&l