Modeled Ditensions an MassinuS Si.

Inthecaseofsimulators ofmobile robots, itis regard tothetraversing abilities, thesizes oftheobstacles necessary toclarify thesimilarity between actual andsimulated robotsthataretraversed rather thanthedynamics oftherobots are inorder toimprove theefficiency andtransparency ofsimulators co d asresearch tools. Because ofitsreal-time performance, USAR- comparedi Sim(1)hasbeenpopularly usedinstudies involving human-robotTherearethree majorreasons whythedimensions andtotal interactions. However, since USARSimusesapproximations and massofthesimulated robots areaccurately modeled andthe abbreviations initsphysical computations, there aresomedoubtsphysical parameters arenot.First, theauthors haveassumed over its computational accuracy. that theability oftheresponse robots totraverse around simple Theobjective ofthis study istocompare theactual andsimulated obstacles significantly depends on thedimensions andmass robots inorder toinvestigate whether thesimulated environments ofsthe s Hnce, by usin thesim edion s with canbeusedasatraining tool. Theauthors modeled twotypes of oftherobots. Hene,byusingthesimulated robots with response robots -IRSSoryu (2)andKenaf(3)- andcompared theaccurately modeled dimensions andmass,we candetermine abilities ofthesimulated robots andtheactual robots innavigating thecapabilities oftheactual robots. Second, measuring the obstacles such assteps andtrenches. Thedimensions such asheights, physical parameters oftherealrobots takes timeandalsois widths andlengths andtotal massofthereal robots wereaccurately highincost.Finally, theapproximations andabbreviations modeled inthesimulation andtheother physical parameters were assigned default values. Itwasobserved that themaximum error in ofthephysical computations inUSARSimaresignificant. thesizes ofthetraversable objects was18% ofthebodylength of Somephysical parameters suchaskinetic friction andstatic therobot. Itwasalso found outthat theangular velocities ofthefriction coefficients arenotconfigurable. Eveniftheconfig- robots weresignificantly different forthereal andsimulated robots. urable parameters areaccurately reproduced inthesimulator, Theconclusion isthatatraining system canbebuilt basedon thedynamic interactions between therobots andthevirtual USARSimwhenthesimulated environments involve simple obstacles because there arenotsignificant discrepancies between thebasicenvironments arenotexpected tomatchthose ofthereal robots traversing abilities oftheactual robots andthose ofthesimulated inanactual environment. robots. However, interms ofthedynamic motions oftherobots the operators needfurther trainings using thereal robots. B.USARSimandRelated Stadies about Validation