JAIST Repository: 制御プログラム構造の違いによる実ロボット動作特性の評価
4
0
0
全文
(2) Evaluation ofrealrobotperform ance w ith differentcontrolprogram structures. Toshikazu Enom oto SchoolofK now ledge Science, Japan A dvanced Institute ofScience and Technology M arch 2000 K eyw ords:behavior-based approach,autonom ous robot,m ultiple robots PU R PO SE Com putersim ulations have been perform ed by m any robotresearches. H ow ever,unlike the virtualw orld,various noises and restrictions existin the actualw orld, and they influence a real robot by them . The purpose of this research is to im plem ent distributed control program s and to estim ate the perform ance.A s a real robot, a consum er productcalled M indStorm s ofLEG O Co.w as used. BA C K G R O U N D H ow to use m ultiple robots has been studied forvarious kinds ofproblem s w hich single robotcannotsolve.There are three kinds ofm ethods to controlm ultiple robots as follow s: 1) centralcontrol:The centralcontrolm ethod collects im m ediately allthe inform ation that the robots acquired in the environm ent ,judges by the globalinform ation,and controls each robot. 2) distributed control(w ith com m unication) :Each robotprocesses an independenttask w hile itcom m unicates each other. 3) distributed control (w ithout com m unication) : Each robot processes an independent task w ithoutcom m unication. Since each system has both of m erits and dem erits, w e cannot conclude that w hich. 1.
(3) system is m ostexcellent.H ow ever,the distributed controlm ethods are expected to be m ore robust than the central control, because each robot controls itself independently. In this research, w e focus on the distributed control w ithout com m unication because of the robustness and low ercost. EX PER IM EN TS In this research, in order to analyze the control program structure suitable for autonom ous m obile robots, the characteristics by the differences in control program are evaluated. The problem of pushing tw ice a target is given to the m ultiple m oving robots and each program structure is evaluated by com paring the results.The goal of this problem is to find the targetand push ittw ice from the frontand the back.The controlprogram ofthe robot,in w hich. behavior-based approach. is im plem ented, consists of three m odules for. corresponding basic actions as follow s : search, evasion and pursuit. Especially the search action and pursuitaction are very im portantforthis problem .Then fourcontrolprogram s w ith differentkinds ofprogram structures to m ake these actions w ere designed as described below . ①1 task type:The search action and the pursuitaction are executed in one task. ②2 task type:The search action and the pursuitaction are executed in tw o independenttasks, respectively. ③m ultiple task type:The search action is executed in m ultiple tasks,and the pursuitaction is executed in one task. ④O nlyTouch type:This type does notexecute the pursuitaction,butthe search action only. R ESU LT Fourkinds ofcontrolprogram s w ere im plem ented in the realrobot,and the perform ance in the problem w as evaluated by average tim e to reach the gorl.W hen m ultiplex com parison w as perform ed by statisticalhypothesis at5% ofsignificantlevels,the results are obtained as follow s. ①and②,①and④, ②and④, ③and④:These com binations have significantdifferences. ①and③, ②and③:Butthese com binations have no significant differences.A nalyzing these results,w e conclude thatitis effective to w atch the environm entcontinuously by the sensor, even ifthe controlm odule have m ore load by the increase in the num beroftasks. C O N C LU SIO N From the above-m entioned result, the structure that assigns all actions to respectively. 2.
(4) independenttasks is suitable foran autonom ous m oving robot’s controlprogram . In the nearfuture,m oving robots w illbe sold as a consum erproduct.Itw illbe required to balance the costagainstthe perform ance to produce such a consum errobot.In this research, autonom ous m oving robot w as im plem ented using M indStorm s, w hich have lim ited restrictions.Even ifthe hardw are has low capability,w e show ed thatthe robotcan adaptitself to the environm ent dynam ically.The m ethod using a consum er level robot in this research suggests one approach to low costs ofm oving robotproducts.. 3.
(5)
関連したドキュメント
The notion of free product with amalgamation of groupoids in [16] strongly influenced Ronnie Brown to introduce in [5] the fundamental groupoid on a set of base points, and so to give
We describe a little the blow–ups of the phase portrait of the intricate point p given in Figure 5. Its first blow–up is given in Figure 6A. In it we see from the upper part of
Next, using the mass ratio m b /m t 100 as in Figure 5, but with e 0.67, and e w 1, we increase the acceleration parameter to a sufficiently large value Γ 10 to fluidize the
We show that a discrete fixed point theorem of Eilenberg is equivalent to the restriction of the contraction principle to the class of non-Archimedean bounded metric spaces.. We
I give a proof of the theorem over any separably closed field F using ℓ-adic perverse sheaves.. My proof is different from the one of Mirkovi´c
Keywords: continuous time random walk, Brownian motion, collision time, skew Young tableaux, tandem queue.. AMS 2000 Subject Classification: Primary:
Thus, we use the results both to prove existence and uniqueness of exponentially asymptotically stable periodic orbits and to determine a part of their basin of attraction.. Let
We present sufficient conditions for the existence of solutions to Neu- mann and periodic boundary-value problems for some class of quasilinear ordinary differential equations.. We
