1、英文原文 A simple approach to the control of locomotion in self-reconfigurable robots K. Sty a, , W.-M. Shen b, P.M. Will b a The Adaptronics Group, The Maersk Institute, University of Southern Denmark, Campusvej 55, DK-5230 Odense M, Denmark b USC Information Sciences Institute and Computer Science Dep
2、artment, 4676 Admiralty Way, Marina del Rey, CA 90292, USA Abstract In this paper we present role-based control which is a general bottom-up approach to the control of locomotion in self-reconfigurable robots. We use role-based control to implement a caterpillar, a sidewinder, and a rolling track ga
3、it in the CONRO self-reconfigurable robot consisting of eight modules. Based on our experiments and discussion we con-clude that control systems based on role-based control are minimal, robust to communication errors, and robust to recon-figuration. 2003 Elsevier Science B.V. All rights reserved. Ke
4、ywords: Self-reconfigurable robots； Locomotion； Role-based control 1. Introduction Reconfigurable robots are robots made from a pos-sibly large 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 wit
5、hout human intervention the robot is a self-reconfigurable robot. 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 7,8,1015,17,21,23 . Several potential advantages of self-reconfigurable robots over traditi
6、onal robots have been pointed out in literature: Versatility. The modules can be combined in differ-ent ways making the same robotic system able to perform a wide range of tasks. Adaptability. While the self-reconfigurable robot performs its task it can change its physical shape to adapt to changes
7、in the environment. Robustness. Self-reconfigurable robots consist of many identical modules and therefore if a module fails it can be replaced by another. Cheap production. When the final design for the basic module has been obtained it can be mass pro-duced. Therefore, the cost of the individual m
8、odule can be kept relatively low in spite of its complexity. Self-reconfigurable robots can solve the same tasks as traditional 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, a
9、pplications best suited for self-reconfigurable robots are applications where some leverage can be gained from the special abilities of self-reconfigurable robots. The versatility of these Fig. 1. A CONRO module. The three male connectors are located in the lower right corner. The female connector i
10、s partly hidden from view in the upper left corner. robots make them suitable in scenarios where the robots have to handle a range of 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 im
11、portance it might be desirable to use self-reconfigurable 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 earthquake, battlefield reconnais-sance, planetary explo
12、ration, undersea mining, and space structure building. Other possible applications include entertainment and service robotics. 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 challenge
13、s in software: how do we make a large number of 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 algorithm to be useful it has to pres
14、erve the special properties of these robots. From the advantages and applications mentioned above we can extract a number of guidelines for the design of such a control algorithm. The algorithm should be dis-tributed to avoid having a single point of failure. Also the performance of the algorithm sh
15、ould scale with an increased number of modules. It has to be robust to re-configuration, because reconfiguration is a fundamen-tal capability of self-reconfigurable robots. Finally, it is desirable to have homogeneous software running on all the modules, because it makes it possible for any module t
16、o take over if another one fails. 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 at the global level and then later try to make an implementation at the local level,because often properties of the hardware are ig
17、nored and a slow robotic system might be the result. There-fore, we use a bottom-up approach 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 design element is the individual module and its interactions with
18、 its neighbors. The global behavior of the system then emerges from the local interaction between individual modules. A sim-ilar approach is also used by Bojinov et al. 1,2 and Butler et al. 4. 2. Related work In the related work presented here we focus on con-trol algorithms for locomotion of self-
19、reconfigurable robots. Yim et al. 22,23 demonstrate caterpillar-like loco-motion and a rolling track. Their system is controlled 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 transi
20、tion from one row to the next. The problem 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, since it gives the system a single point of failure. If there is no master it
21、 is suggested that the modules can be assumed to be synchronized in time and each module can execute its column of actions open-loop. However, since all the modules are autonomous it is a questionable assumption to assume that all the modules are and can stay synchronized. In order to use the gait c
22、ontrol table each module needs to know what column it has to execute. This means that the modules need IDs. Furthermore, if the con-figuration changes or the number of modules changes the table has to be rewritten. Shen et al. 17 propose to use artificial hormones to synchronize the modules to achie
23、ve consistent global locomotion. In earlier versions of the system a hor-mone is propagated through the self-reconfigurable system to achieve synchronization. In later work the hormone is also propagated backward making all modules synchronized before a new action is initiated 16,18. This synchroniz
24、ation takes time O(n), where n is the number of modules. This slows down the system considerably, because it has to be done before each action. Also, the entire system stops working if one hormone is lost. This is a significant problem, because a hormone can easily be lost due to unreli-able communi
25、cation, a module disconnecting itself before a response can be given, or a module failure. In fact, the system has n points of failure which is not desirable. The earlier version is better in this sense, but still performance remains low because a synchronization hormone is sent before each action.
26、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 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 avoid jerky locomotion. In our system all modules repeatedly go through a cyclic sequence of joint angles describing a motion.
