This preview shows page 1-2-3-4 out of 13 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 13 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 13 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 13 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 13 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 13 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

Eiichi YoshidaSatoshi MurataAkiya KamimuraKohji TomitaHaruhisa KurokawaShigeru KokajiDistributed System Design Research GroupIntelligent Systems InstituteNational Institute of Advanced IndustrialScience and Technology (AIST)1-2-1 Namiki, Tsukuba-shi, Ibaraki 305-8564, [email protected] Self-ReconfigurableModular Robot:ReconfigurationPlanning andExperimentsAbstractIn this paper we address a reconfiguration planning method for lo-comotion of a homogeneous modular robotic system and we conductan experiment to verify that the planned locomotion can be realizedby hardware. Our recently developed module is self-reconfigurable.A group of the modules can thus generate various three-dimensionalrobotic structures and motions. Although the module itself is a simplemechanism, self-reconfiguration planning for locomotion presents acomputationally difficult problem due to the many combinatorialpossibilities of modular configurations. In this paper, we develop atwo-layered planning method for locomotion of a class of regularstructures. This locomotion mode is based on multi-module blocks.The upper layer plans the overall cluster motion called flow to realizelocomotion along a given desired trajectory; the lower layer deter-mines locally cooperative module motions, called motion schemes,based on a rule database. A planning simulation demonstrates thatthis approach effectively solves the complicated planning problem.Besides the fundamental motion capacity of the module, the hard-ware feasibility of the planning locomotion is verified through aself-reconfiguration experiment using the prototype modules we havedeveloped.KEY WORDS—self-reconfigurable robot, modular robotics,planning, hierarchical planner, experimental robotics1. IntroductionIn recent years, the feasibility of reconfigurable robotic sys-tems has been examined through hardware and software ex-periments in two dimensions (Fukuda and Nakagawa 1988;The International Journal of Robotics ResearchVol. 21, No. 10–11, October-November 2002, pp. 903-915,©2002 Sage PublicationsMurata, Kurokawa, and Kokaji 1994; Chirikjian, Pamecha,and Ebert-Uphoff 1996; Hosokawa et al. 1998; Yoshida et al.1999a; Tomita et al. 1999; Walter, Welch, and Amato 2000;Yoshida et al. 2000) and three dimensions (Murata et al. 1998;Kotay et al. 1998; Kotay and Rus 1998; Hamlin and Sander-son 1998; Yim, Duff, and Roufas 2000; Yoshida et al. 1999b;Murata et al. 2000; Castano, Chokkalingam, and Will 2000).Specifically, a self-reconfigurable robot can adapt itself to theexternal environments. It can also repair itself by using sparemodules owing to the homogeneity of the module. In this pa-per we focus on the reconfiguration planning for a new type ofhomogeneous, self-reconfigurable modular robot that enablesmovement in three dimensions by changing the configuration.Its various potential applications include structures or robotsthat operate in extreme environments inaccessible to humans,for example, in space, deep sea, or nuclear plants.The hardwareofthree-dimensional(3D)self-reconfigurablemodular robots is classified into two types: the lattice type(Murata et al. 1998; Kotay et al. 1998; Ünsal, Kılıççöte,and Khosla 2001; Rus and Vona 2001; Yim et al. 2001)and the linear type (Yim, Duff, and Roufas 2000; Castano,Chokkalingam, and Will 2000). The former corresponds toa system where each module has several fixed connectiondirections, and a group of them can construct various staticstructures such as a jungle gym. However, it is difficult forsuch a system to generate wave-like motions involving manymodules at the same time. In contrast, the latter (linear type)has a shape that can generate various robotic motions such asa snake or a caterpillar, but self-reconfiguration is difficult.There have been a number of studies on the softwareof lattice-type reconfigurable modular robots. Distributedself-reconfiguration methods have also been developed fortwo-dimensional (2D) and 3D homogeneous modular robots903904 THE INTERNATIONAL JOURNAL OF ROBOTICS RESEARCH / October-November 2002(Yoshida et al. 1999a, 1999b; Tomita et al. 1999; Walter,Welch, and Amato 2000; Yim et al. 2001). The first method(Yoshida et al. 1999a) has already been implemented in a 2Dhardware system with more than ten modules to demonstratethe self-assembly and self-repair capacity. The others havebeen investigated in simulations to be implemented in hard-ware in future developments. There are also other methodsbased on centralized planning. Kotay and Rus (1998, 2000)have developed robotic modules and described a global mo-tion synthesis method for a class of module groups to movein arbitrary directions. Ünsal, Kılıççöte, and Khosla (2001)have reported on two-level motion planners for a bipartitemodule composed of cubes and links, based on a heuristicgraph search among module configurations.We have recently developed a new type of modular roboticsystem that has both lattice-type and linear-type features (Mu-rata et al. 2000). The module has a simple bipartite structurein which each part rotates about an axis parallel to the oth-ers by a servomotor and has three magnetic connecting faces.This recent module can form various shapes, such as a leggedwalking robot or a crawler-type robot (see Multimedia Exten-sion 4 for an example of transformation between these loco-motion modes). However, its reconfiguration planning is notstraightforward because of restricted degrees of freedom andnon-isotropic geometrical properties of mobility of a mod-ule unlike lattice-type modules in previous research. When amodule moves from one position to another, a sequence ofnecessary motions must be duly planned for each individuallocal configuration. In addition, a vast search space must beexplored to examine the interchangeability between two arbi-trary module configurations and to avoid collisions betweenmodules in 3D space. These properties of the module makeit difficult to identify generic laws of motion planning and toapply our distributed methods directly.Therefore, in this paper we concentrate on developing fea-sible reconfiguration planning for locomotion of a particularclass of lattice-type module clusters by narrowing the mo-tion search space as the first step to a more general planningmethod. The module cluster to be investigated is a serial col-lection of cube-like blocks, each of which is composed of fourmodules. Using this locomotion mode, the self-reconfigurablerobot can surmount large obstacles and


View Full Document

USC CSCI 584 - 903

Documents in this Course
Load more
Download 903
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view 903 and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view 903 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?