JAIST Repository: 自律移動ロボットの協調行動に関する研究
3
0
0
全文
(2) A Study on Cooperative Actions of Autonomous Mobile Robots Youichi Takahashi School of Knowledge Science, Japan Advanced Institute of Science and Technology March, 2001. autonomous mobile robot, dynamic environment, subsumption architecture,cooperative action, arti
(3) cial neural network.. Keywords:. Many simulations on a computer have been performed in the conventional robot research. However, various noises in the real world and restrictions conditions. Especially an autonomous mobile robot does not escape the in uence. In this study, it aims at carrying out clearing up that cooperative action ,especially "following another robot" action in the case of not using an explicit communication means by two or more autonomous mobile robots in dynamic environment I implemented control program which a type di ers from to an autonomous mobile robot,respectively, and by making it operate in a real environment considered in order to clarify control program structure suitable for communication-less "following another robot" action . It uses as the control architecture which mounts to a real robot, and is the subsumptiontion architecture. It adopted. An "following another robot" action is decomposed into each module, and fundamental operation is performed. In the experiment, 1. Two kinds of di erent module which judges a wall and the other object were implemented in the real robot. One is a control program only based on the design technique of subsumption. Another is the control program which used the feed forward type neural network that learned the discernment from a wall and the other object by the sensor value from static environment. 2. The performance of a control program was compared by measuring the number of times of "following another robot" action within
(4) xed time.. Copyright c 2001 by Youichi Takahashi. 1.
(5) Consequently, it is shown that the control program implemented recognition module designed based on experience of a designer is a high "following another robot" action performance relatively. And interesting various actions were observed. On the one hand, it is shown that the control program implemented feed forward neural network is low performance relatively and "following another robot" action was hardly able to be taken, in dynamic environment, although high recognition accuracy is shown in static environment In this study, the validity of the subsumption architecture was checked anew. If it experiments in a dynamic environment using control module included neural network which only map from static sensor information to output , it turns out that environmental recognition eÆciency becomes bad and the performance of imitation action becomes low relatively. The re ected that a motion of the robot, such as the evolutional calculation te chnique, itself to recognition mechanism acquisition process. approach is taken, It is considered that robust system using the interaction of system and environment can be built also under dynamic environment. Therefore, it is mutual interaction with environment study how from now on. It becomes a subject whether it should take in.. 2.
(6)
関連したドキュメント
The study of the eigenvalue problem when the nonlinear term is placed in the equation, that is when one considers a quasilinear problem of the form −∆ p u = λ|u| p−2 u with
The variational constant formula plays an important role in the study of the stability, existence of bounded solutions and the asymptotic behavior of non linear ordinary
7.1. Deconvolution in sequence spaces. Subsequently, we present some numerical results on the reconstruction of a function from convolution data. The example is taken from [38],
In this paper relatively realcompact sets are defined, and it is shown that a space is nearly pseudocompact iff every relatively realcompact open set is relatively compact..
In Section 2 we show that the infinite measure-preserving transformation that was shown in [DGMS99] to be power weakly mixing is not multiply recurrent; we in fact show that it
We will study the spreading of a charged microdroplet using the lubrication approximation which assumes that the fluid spreads over a solid surface and that the droplet is thin so
The aim of the present section is to prove that the Orthogonality Logic is complete (for all classes of morphisms) in all locally presentable categories iff the following
But in fact we can very quickly bound the axial elbows by the simple center-line method and so, in the vanilla algorithm, we will work only with upper bounds on the axial elbows..
