機(jī)械電子畢業(yè)設(shè)計外文文獻(xiàn)翻譯-- 一個簡單的方法來控制運動中的自重構(gòu)機(jī)器人

1、<p><b>　　英文原文</b></p><p>　　A simple approach to the control of locomotion in self-reconfigurable robots</p><p>　　K. Støy a,?, W.-M. Shen b, P.M. Will b</p><p>

2、　　The Adaptronics Group, The Maersk Institute, University of Southern Denmark, Campusvej 55, DK-5230 Odense M, Denmark </p><p>　　USC Information Sciences Institute and Computer Science Department, 4676 Admir

3、alty Way, Marina del Rey, CA 90292, USA </p><p><b>　　Abstract</b></p><p>　　In this paper we present role-based control which is a general bottom-up approach to the control of locomot

4、ion in self-reconfigurable robots. We use role-based control to implement a caterpillar, a sidewinder, and a rolling track gait in the CONRO self-reconfigurable robot consisting of eight modules. Based on our experiments

5、 and discussion we con-clude that control systems based on role-based control are minimal, robust to communication errors, and robust to recon-figuration.</p><p>　　© 2003 Elsevier Science B.V. All right

6、s reserved.</p><p>　　Keywords: Self-reconfigurable robots; Locomotion; Role-based control</p><p>　　1. Introduction</p><p>　　Reconfigurable robots are robots made from a pos-sibly la

7、rge number of independent modules connected to form a robot. If the modules from which the re-configurable robot is built are able to connect and disconnect without human intervention the robot is a self-reconfigurable r

8、obot. Refer to Fig. 1 for an exam-ple of a module of a self-reconfigurable robot or refer to one of the other physical realized systems described in [7,8,10–15,17,21,23] .</p><p>　　Several potential advanta

9、ges of self-reconfigurable robots over traditional robots have been pointed out in literature:</p><p>　　Versatility. The modules can be combined in differ-ent ways making the same robotic system able to perf

10、orm a wide range of tasks. </p><p>　　Adaptability. While the self-reconfigurable robot performs its task it can change its physical shape to adapt to changes in the environment. </p><p>　　Robust

11、ness. Self-reconfigurable robots consist of many identical modules and therefore if a module fails it can be replaced by another. </p><p>　　Cheap production. When the final design for the basic module has be

12、en obtained it can be mass pro-duced. Therefore, the cost of the individual module can be kept relatively low in spite of its complexity. </p><p>　　Self-reconfigurable robots can solve the same tasks as trad

13、itional robots, but as Yim et al. [23] point out; in applications where the task and environment are given a priori it is often cheaper to build a special purpose robot. Therefore, applications best suited for self-recon

14、figurable robots are applications where some leverage can be gained from the special abilities of self-reconfigurable robots. The versatility of these</p><p>　　Fig. 1. A CONRO module. The three male connecto

15、rs are located in the lower right corner. The female connector is partly hidden from view in the upper left corner.</p><p>　　robots make them suitable in scenarios where the robots have to handle a range of

16、tasks. The robots can also handle tasks in unknown or dynamic environ-ments, because they are able to adapt to these envi-ronments. In tasks where robustness is of importance it might be desirable to use self-reconfigura

17、ble robots. Even though real applications for self-reconfigurable robots still are to be seen, a number of applications have been envisioned [17,23]: fire fighting, search and rescue after an earthq</p><p>　

18、　The potential of self-reconfigurable robots can be realized if several challenges in terms of hardware and software can be met. In this work we focus on one of the challenges in software: how do we make a large number o

19、f connected modules perform a coor-dinated global behavior? Specifically we address howto design algorithms that will make it possible for self-reconfigurable robots to locomote efficiently. In order for a locomotion alg

20、orithm to be useful it has to preserve the special properties</p><p>　　It is an open question if a top-down or a bottom-up approach gives the best result. We find that it is diffi-cult to design the system a

21、t the global level and then later try to make an implementation at the local level,because often properties of the hardware are ignored and a slow robotic system might be the result. There-fore, we use a bottom-up approa

22、ch where the single module is the basic unit of design. That is, we move from a global design perspective to a bottom-up one where the important</p><p>　　2. Related work</p><p>　　In the related

23、work presented here we focus on con-trol algorithms for locomotion of self-reconfigurable robots.</p><p>　　Yim et al. [22,23] demonstrate caterpillar-like loco-motion and a rolling track. Their system is con

24、trolled based on a gait control table. Each column in this table represents the actions performed by one module. Mo-tion is then obtain by having a master synchronizing the transition from one row to the next. The proble

25、m with this approach is that the amount of communica-tion needed between the master and the modules will limit its scalability. Another problem is the need for a central controller</p><p>　　Shen et al. [17]

26、propose to use artificial hormones to synchronize the modules to achieve consistent global locomotion. In earlier versions of the system a hor-mone is propagated through the self-reconfigurable system to achieve synchron

27、ization. In later work the hormone is also propagated backward making all modules synchronized before a new action is initiated [16,18]. This synchronization takes time O(n), where n is the number of modules. This slows

28、down the system considerably, because it ha</p><p>　　Butler et al. [4] propose a method inspired by cel-lular automata. In their approach modules respond to state changes of neighbor modules. Their approach

29、is a bottom-up approach related to ours, but in cellular automata there is no concept of time only of sequence. Timing is important in locomotion, because it is the key to produce smooth and life-like locomotion and avoi

30、d jerky locomotion.</p><p>　　In our system all modules repeatedly go through a cyclic sequence of joint angles describing a motion. This sequence could come from a column in a gait con-trol table, but in our

31、 implementation the joint angles are calculated using a cyclic function with period T . Every time a module has completed a specified frac-tion d of the period a message is sent through the child connectors. If the signa

32、l is received the child module resets its action sequence making it delayed d com-pared to the parent.</p><p>　　3. Role-based control</p><p>　　We assume that the modules are connected to form a

33、tree structure, that a parent connector is specified, and that this connector is the only one that can connect to child connectors of other modules. Furthermore, we assume that the modules can communicate with the module

34、s to which they are connected.</p><p>　　The algorithm is instantiated by specifying three components. The first component is a cyclic action sequence A(t), where t ∈ [0 : T ]. T is the second component that

35、needs to be specified and is the pe-riod of the action sequence. A(t) describes the ac-tions that each module repeats cycle after cycle. In our implementation A(t) returns joint angles to control the two degrees of freed

36、om of the CONRO module, but the action sequence could also be used to trigger dif-ferent behaviors at different t</p><p>　　t=0 while(true) {</p><p>　　if(t=d)then <signal child modules> if

37、<parent signals> then t=0 <perform action A(t)></p><p>　　t=(t+1) modulus T</p><p><b>　　}</b></p><p>　　Ignoring the first two lines of the loop, the module re

38、peatedly goes through a sequence of actions param-eterized by the cyclic counter t. This part of the al-gorithm alone can make a single module repeatedly perform the specified sequence of actions. In order to coordinate

39、the actions of the individual modules to produce the desired global behavior the modules need to be synchronized. Therefore, at step t = d a signal is sent through all child connectors. If a child receives a signal it kn

40、ows t</p><p>　　From the time the modules are connected it takes time proportional to d times the height of the con-figuration tree for all the modules to synchronize. To avoid problems with uncoordinated mod

41、ules initially we make sure the modules do not start moving until they receive the first synchronization signal. After the start-up phase the modules stay synchronized using only constant time.</p><p>　　4. E

42、xperimental setup</p><p>　　To evaluate our algorithm we conducted several ex-periments using the CONRO (CONfigurable RObot) modules of which one is shown in Fig. 1. The CONRO modules have been developed at U

43、SC/ISI [5,9]. The modules are roughly shaped as rectangular boxes mea-suring 10 cm × 4.5 cm × 4.5 cm and weigh 100 g. The modules have a female connector at one end and three male connectors located at the othe

44、r. Each connectorhas a infra-red transmitter and receiver used for local communication and sensing. The module</p><p>　　5. Experiments</p><p>　　In this section we describe three different locomo

45、-tion gaits implemented using role-based control. For each gait we have chosen to report the length of our programs as a measure of the complexity of the control algorithm. These results are used to support our claim tha

46、t the implemented control systems are minimal. We also report the speed of the locomotion patterns, but this should only be considered an example, the reason being that in our system the limiting factors are how robust t

47、he modules</p><p>　　5.1. Caterpillar locomotion</p><p>　　We connect eight of our modules in a chain and designate the male opposite the female connector to be the parent connector. We then impl

48、ement the algorithm described above with the following parameters.</p><p><b>　　T = 180,</b></p><p><b>　　_ </b></p><p>　　Pitch and yaw is measured in a coordi

49、nate system where a yaw and a pitch of zero mean that the joints are straight. The motor control of our modules makes the joint go to the desired angle as fast as possible. This means that way-points have to be specified

50、 to avoid jerky motion. The period T can be used to control the number of way-points and therefore the smoothness and speed of the motion. The action sequence is an oscillation around 0? with an amplitude of 50? for the

51、pitch angle and the yaw</p><p>　　The modules are connected and after they synchro-nize a sine wave is traveling along the length of the robot. Refer to Fig. 2. This produces caterpillar-like locomotion at a

52、speed of 4 cm/s. Note that it is easy to adjust the parameters of this motion. For instance, the length of the wave can be controlled using the de-lay. The program is simple. The main loop contains 13 lines of code exclu

53、ding comments and labels (shown in Fig. 2). The initialization including variable and constant declaration</p><p>　　Fig. 3. A snapshot of sidewinder-like locomotion.</p><p>　　lines of code. The

54、parameters for the eight module rolling track are:</p><p>　　Unlike the controller for the sidewinder gait and the caterpillar gait this controller only works with eight modules, because of the physical const

55、raint. It might be possible to make a more general solution by mak-ing pitch(t) and d a function of the number of mod-ules. The number of modules in the loop could be ob-tained by the leader by including a hop count in t

56、he signal.</p><p>　　6. Handling a general configuration</p><p>　　We saw in the previous section that we had to in-troduce IDs to find a unique leader in a configuration that contains loops. Intr

57、oducing the ID mechanism un-fortunately ruins the opportunity to use the synchro-nization algorithm to automatically find a leader in a tree structure, because synchronization signals are only propagated down in the conf

58、iguration tree. In fact, the loop algorithm will fail in this situation unless the module with the highest ID also happens to be the root. In order to </p><p>　　7. Discussion</p><p>　　An importa

59、nt issue in the design of control algo-rithms for self-reconfigurable robots is that the algo-rithms should still be efficient in systems consisting of many modules. Role-based control is only initially dependent on the

60、number of modules, because it de-cides how long it takes for the synchronization sig-nal to be propagated through the system. After this start-up phase it takes constant time to keep the mod-ules synchronized implying th

61、at the algorithm scales.</p><p>　　In role-based control all modules run identical pro-grams and there is no representation of a modules po-sition in the configuration. Therefore, the system is highly robust

62、to reconfiguration. In fact, the caterpil-lar can be divided in two and both parts still work. If they are reconnected in a different order they will quickly synchronize to behave as one caterpillar again. This also impl

63、ies that the system is robust to module failure. If a module is defect and it can be detected this module c</p><p>　　In role-based control the synchronization signal is only sent once per period. This means

64、that in order for the modules to stay synchronized the time to complete a period has to be the same for all modules. In the experiments presented here the cycles take the same amount of time, but in more complex control

65、systems where other parts of the control system use random amounts of computation time this cannot be assumed to be true. This problem can easily be handled by using timers. Even though time</p><p>　　In the

66、work presented here we have shown what can be achieved using as simple a control algorithm as possible. In related work we have investigated how we can extend the algorithm to handle more complex locomotion gaits. We hav

67、e shown how to implement a hexapod walking gait in the CONRO system in [19].</p><p>　　Another issue is that if the self-reconfigurable robot is to locomote autonomously in a real complex envi-ronment the con

68、trol algorithm has to be able to take feedback from the environment into account. We have done some initial work on how sensor feedback can be included in role-based control in [20].</p><p>　　8. Summary</

69、p><p>　　We have presented role-based control a general control algorithm for controlling locomotion in self-reconfigurable robots. The algorithm has the follow-ing properties: distributed, scalable, homogeneous

70、, and minimal. We have shown how the algorithm easily can be used to implement a caterpillar- and sidewinder-like locomotion pattern. Furthermore, we have seen that by giving modules IDs it is possible to handle loop con

71、figurations. We have demonstrated this using the rolling track as an example.</p><p>　　Acknowledgements</p><p>　　This work is supported under the DARPA contract DAAN02-98-C-4032, the EU contract

72、 IST-20001-33060, and the Danish Technical Research Council contract 26-01-0088.</p><p>　　References</p><p>　　H. Bojinov, A. Casal, T. Hogg, Emergent structures in modular self-reconfigurable ro

73、bots, in: Proceedings of the IEEE International Conference on Robotics and Automation (ICRA’00), San Francisco, CA, 2000, vol. 2, pp. 1734–1741. </p><p>　　H. Bojinov, A. Casal, T. Hogg, Multiagent control of

74、 self-reconfigurable robots, in: Proceedings of the Fourth International Conference on MultiAgent Systems, Boston, MA, 2000, pp. 143–150. </p><p>　　J.W. Burdick, J. Radford, G.S. Chirikjian, A ‘sidewinding’

75、locomotion gait for hyper-redundant robots, in: Proceedings of the IEEE International Conference on Robotics and Automation, Atlanta, GA, 1993, pp. 101–106. </p><p>　　Z. Butler, R. Fitch, D. Rus, Exper

76、iments in locomotion </p><p>　　with a unit-compressible modular robot, in: Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems, Lausanne, Switzerland, October 2002,</p

77、><p>　　pp. 2813–2818. </p><p>　　A. Castano, R. Chokkalingam, P. Will, Autonomous and </p><p>　　self-sufficient conro modules for reconfigurable robots,</p><p>

78、　　in:Proceedings of the Fifth International Symposium</p><p>　　on Distributed Autonomous Robotic Systems (DARS’00), Knoxville, TX, 2000, pp. 155–164.</p><p>　　E. Chang, R. Roberts, An impr

79、oved algorithm for decentralized extrema-finding in circular configurations of processes, Communications of the ACM 22 (5) (1979) 281– 283. </p><p>　　G.S. Chirikjian, Metamorphic hyper-redundant manupulators

80、, in: Proceedings of the 1993 JSME International Conference on Advanced Mechatronics, Tokyo, Japan, 1993, pp. 467–472. </p><p>　　K. Hosokawa, T. Tsujimori, T. Fujii, H. Kaetsu, H. Asama, </p><p>

81、;　　Kuroda, I. Endo, Self-organizing collective robots with morphogenesis in a vertical plane, in: Proceedings of the IEEE International Conference on Robotics and Automation (ICRA’98), Leuven, Belgium, 1998, pp. 2858–286

82、3. </p><p>　　B. Khoshnevis, B. Kovac, W.-M. Shen, P. Will, Reconnectable joints for self-reconfigurable robots, in: Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IRO

83、S’01), Maui, HI, 2001, pp. 584–589. </p><p>　　K. Kotay, D. Rus, M. Vona, C. McGray, The self-reconfiguring robotic molecule, in: Proceedings of the IEEE International Conference on Robotics and Automation (I

84、CRA’98), Leuven, Belgium, 1998, pp. 424–431. </p><p>　　S. Murata, H. Kurokawa, S. Kokaji, Self-assembling machine, in: Proceedings of the IEEE International Conference on Robotics and Automation (ICRA’94), S

85、an Diego, CA, 1994, pp. 441–448. </p><p>　　S. Murata, H. Kurokawa, E. Yoshida, K. Tomita, S. Kokaji, </p><p>　　3-d self-reconfigurable structure, in: Proceedings of the IEEE International Confer

86、ence on Robotics and Automation (ICRA’98), Leuven, Belgium, 1998, pp. 432–439. </p><p>　　S. Murata, E. Yoshida, K. Tomita, H. Kurokawa, A. Kamimura, S. Kokaji, Hardware design of modular robotic system, in:

87、Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS’00), Takamatsu, Japan, 2000, pp. 2210–2217. </p><p>　　A. Pamecha, C. Chiang, D. Stein, G.S. Chirikjian, Design and

88、 implementation of metamorphic robots, in: Proceedings of the ASME Design Engineering Technical Conference and Computers in Engineering Conference, Irvine, CA, 1996, pp. 1–10. </p><p>　　D. Rus, M. Vona, Crys

89、talline robots: self-reconfiguration with compressible unit modules, Autonomous Robots 10 (1) (2001) 107–124. </p><p>　　B. Salemi, W. Shen, P. Will, Hormone controlled metamorphic robots, in: Proceedings of

90、the IEEE International Conference on Robotics and Automation, Seoul, Republic of Korea, 2001, </p><p>　　4194–4199. </p><p>　　W.-M. Shen, B. Salemi, P. Will, Hormone-based control for self-reconf

91、igurable robots, in: Proceedings of the International Conference on Autonomous Agents, Barcelona, Spain, 2000, </p><p><b>　　1–8. </b></p><p>　　W.-M. Shen, B. Salemi, P. Will, Hormone

92、s for self-reconfigurable robots, in: Proceedings of the International Conference on Intelligent Autonomous Systems (IAS-6), Venice, Italy, 2000, pp. 918–925. </p><p>　　K. Støy, W.-M. Shen, P. Will, How

93、 to make a self-reconfigurable robot run, in: Proceedings of the First International Joint Conference on Autonomous Agents and Multiagent Systems (AAMAS’02), Bologna, Italy, 2002, </p><p><b>　　813–820.

94、 </b></p><p>　　K. Støy, W.-M. Shen, P. Will, On the use of sensors in self-reconfigurable robots, in: Proceedings of the Seventh International Conference on the Simulation of Adaptive behavior (SA

95、B’02), Edinburgh, UK, 2002, pp. 48–57. </p><p>　　C. Ünsal, P.K. Khosla, Mechatronic design of a modular self-reconfiguring robotic system, in: Proceedings of the IEEE International Conference on Robotic

96、s and Automation (ICRA’00), San Francisco, CA, 2000, pp. 1742–1747. </p><p>　　M. Yim, Locomotion with a unit-modular reconfigurable robot, Ph.D. Thesis, Department of Mechanical Engineering, Stanford Univers

97、ity, 1994. </p><p>　　M. Yim, D.G. Duff, K.D. Roufas, Polybot: a modular reconfigurable robot, in: Proceedings of the IEEE International Conference on Robotics and Automation (ICRA’00), San Francisco, CA, 200

98、0, pp. 514-520. </p><p><b>　　中文譯文</b></p><p>　　一個簡單的方法來控制運動中的自重構(gòu)機(jī)器人</p><p>　　K. Støy a,?, W.-M. Shen b, P.M. Will b</p><p><b>　　摘要</b></

99、p><p>　　在本文中，我們提出了基于角色的控制，這是一般的自下而上的方法，在自重構(gòu)機(jī)器人的運動控制。我們使用基于角色的控制，實施毛蟲，響尾蛇，并滾動軌道中的CONRO 8個模塊組成的自我重構(gòu)機(jī)器人步態(tài)。我們的實驗和討論的基礎(chǔ)上，我們CON包括控制系統(tǒng)，基于角色的控制的基礎(chǔ)是最小的，健壯的通信錯誤，和強(qiáng)大的偵察配置。</p><p>　　關(guān)鍵詞：自重構(gòu)機(jī)器人;步態(tài);基于角色的控制<