電気学会論文誌C(電子・情報・システム部門誌)
Online ISSN : 1348-8155
Print ISSN : 0385-4221
ISSN-L : 0385-4221
124 巻, 10 号
選択された号の論文の40件中1~40を表示しています
特集:知能および進化システム
特集論文
<電子・集積回路>
  • Hidenori Kawamura, Takashi Kataoka, Koichi Kurumatani, Azuma Ohuchi
    2004 年 124 巻 10 号 p. 1922-1929
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    We focus on the simple theme park problem, where there are two attractions and visitor agents which select their destination attraction based on congestion disregarding behavior and congestion avoiding behavior. According to the computer simulation, the result shows that the growth of individual congestion avoiding behavior is not always effective for improving global performance, and this phenomenon is caused by the oscillation of successive selection switching of the same destination by many congestion avoiding agents. Although the model and setting of this paper is simpler than other related works, we consider each phenomenon in those works has the same characteristic based on the ineffectiveness caused by the homogeneity of congestion avoiding behavior and information sharing.
<ロボティクス>
  • Keiko Motoyama, Hidenori Kawamura, Masahito Yamamoto, Azuma Ohuchi
    2004 年 124 巻 10 号 p. 1930-1937
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    The final goal of this study is to achieve an autonomous indoor blimp robot. While blimp robots are attractive entities moving safely in three dimensional space, they are hard to control autonomously as they are subjected to strong inertial forces and air resistance, and have many nonlinear characteristics. As blimp robots are used not only in vast halls but also in complex buildings with obstacles such as pillars and walls, they need to plan suitable motion for avoiding them. Furthermore, robots need to conserve energy with restricted power supply because they are unable to carry heavy batteries. We have to design an action-value function for planning motion in a blimp robot, considering constraints imposed by these characteristics of an actual robot. In this paper, we designed an action-value function for motion planning based on the potential field method, and evaluated its effectiveness in a simulated environment.
  • Komei Sugiura, Takayuki Shiose, Hiroshi Kawakami, Osamu Katai
    2004 年 124 巻 10 号 p. 1938-1943
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    In this paper we investigate the evolutionary development of embodied agents that evolve not only control mechanisms but also the sensitivity and temporal resolution of their sensors. In our system, agents can adapt their sensory parameters to their morphology and task environments, while no explicit cost for sensors is used. We have carried out two kinds of experiments in which agents perform phototactic and pursuit/evading behaviors. The experimental results indicate that the sensors and controller co-evolve in an agent through interaction with the environments.
<ソフトコンピューティング>
  • Joarder Kamruzzaman, Ruhul A Sarker
    2004 年 124 巻 10 号 p. 1944-1951
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    Previous studies have demonstrated superior performance of artificial neural network (ANN) based forex forecasting models over traditional regression models. This paper applies support vector machines to build a forecasting model from the historical data using six simple technical indicators and presents a comparison with an ANN based model trained by scaled conjugate gradient (SCG) learning algorithm. The models are evaluated and compared on the basis of five commonly used performance metrics that measure closeness of prediction as well as correctness in directional change. Forecasting results of six different currencies against Australian dollar reveal superior performance of SVM model using simple linear kernel over ANN-SCG model in terms of all the evaluation metrics. The effect of SVM parameter selection on prediction performance is also investigated and analyzed.
  • Ding-zhong Feng, Mitsuo Yamashiro
    2004 年 124 巻 10 号 p. 1952-1957
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    A decision-making model is developed for an optimal production policy in a two-stage production-delivery system that incorporates a fixed quantity supply of finished goods to a buyer at a fixed interval of time. First, a general cost model is formulated considering both supplier (of raw materials) and buyer (of finished products) sides. Then an optimal solution to the problem is derived on basis of the cost model. Using the proposed model and its optimal solution, one can determine optimal production lot size for each stage, optimal number of transportation for semi-finished goods, and optimal quantity of semi-finished goods transported each time to meet the lumpy demand of consumers. Also, we examine the sensitivity of raw materials ordering and production lot size to changes in ordering cost, transportation cost and manufacturing setup cost. A pragmatic computation approach for operational situations is proposed to solve integer approximation solution. Finally, we give some numerical examples.
  • Dat Tran, Michael Wagner, Yee Wah Lau, Mitsuo Gen
    2004 年 124 巻 10 号 p. 1958-1963
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    In this paper, we present normalization methods based on fuzzy set theory to enhance the performance of voice-based person authentication systems. For an input utterance and a claimed identity, a similarity score is calculated and compared with a given threshold to accept or reject the claimant. Using fuzzy set theory, the similarity score can be viewed as a fuzzy membership function, which denotes the degree of belonging of the input utterance to the claimant model. We present fuzzy similarity scores based on fuzzy entropy and fuzzy c-means membership functions. Furthermore, the noise clustering method supplies a very effective modification to all methods, which can overcome some of the problems of ratio-type scores and reduce the false acceptance rate. Experiments were performed to evaluate proposed normalization methods for speaker verification using the ANDOSL database and utterance verification using the TI46 database. Experiments showed better results for fuzzy similarity scores.
  • Ruhul Sarker, Akira Namatame
    2004 年 124 巻 10 号 p. 1964-1971
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    Evolutionary Algorithms have attracted increasing attention in recent years, as powerful computational techniques, for solving many complex real-world problems. The successful application of Evolutionary algorithms to optimization problems is dependent on the methods and parameters used for the algorithms. In this paper, we develop three evolutionary algorithms for solving a transshipment problem. We investigate the effect of population sizes on the quality of solutions to be obtained, the computational time to be required and the size of search spaces of the problems under consideration. We also use a well-known conventional optimization package to compare the quality of solutions. The numerical results are analyzed and the interesting findings are discussed.
  • Mitsuo Gen, Lin Lin, Runwei Cheng
    2004 年 124 巻 10 号 p. 1972-1978
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    Network optimization is being an increasingly important and fundamental issue in the fields such as engineering, computer science, operations research, transportation, telecommunication, decision support systems, manufacturing, and airline scheduling. In many applications, however, there are several criteria associated with traversing each edge of a network. For example, cost and flow measures are both important in the networks. As a result, there has been recent interest in solving Bicriteria Network Optimization Problem. The Bicriteria Network Optimization Problem is known a NP-hard. The efficient set of paths may be very large, possibly exponential in size. Thus the computational effort required to solve it can increase exponentially with the problem size in the worst case. In this paper, we propose a genetic algorithm (GA) approach used a priority-based chromosome for solving the bicriteria network optimization problem including maximum flow (MXF) model and minimum cost flow (MCF) model. The objective is to find the set of Pareto optimal solutions that give possible maximum flow with minimum cost. This paper also combines Adaptive Weight Approach (AWA) that utilizes some useful information from the current population to readjust weights for obtaining a search pressure toward a positive ideal point. Computer simulations show the several numerical experiments by using some difficult-to-solve network design problems, and show the effectiveness of the proposed method.
  • Fulya Altiparmak, Mitsuo Gen, Berna Dengiz, Alice E. Smith
    2004 年 124 巻 10 号 p. 1979-1985
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    In this paper, we develop a network-based genetic algorithm with fuzzy logic controller (flc-NBGA) to the design of computer communication networks under reliability constraint, which is a well-known NP-hard problem. A new encoding based on Prüfer numbers, two-point crossover and local search operator as a mutation have been used in flc-NBGA. The algorithm results are compared to optimum results found by branch and bound (B&B), GA based on binary representation (SGA) and NBGA without fuzzy logic controller (NBGA) on a suite of test problems.
  • Minoru Mukuda, YoungSu Yun, Mitsuo Gen
    2004 年 124 巻 10 号 p. 1986-1990
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    This paper proposes an adaptive genetic local search (aGLS) algorithm for effectively solving reliability optimization problems. The proposed aGLS hybridizes a local search technique and genetic algorithm (GA). The local search technique is incorporated into the GA loop and its scheme is adaptively regulated according to the change of the average fitness value at every generation in GA.
    For more various comparisons with the proposed aGLS, conventional GLS algorithm with local search that does not use any adaptive scheme is also suggested. These two algorithms are tested and analyzed using two complex reliability optimization problems. Numerical result shows that the proposed aGLS outperforms the conventional HGA.
  • Jinwu Gao, Baoding Liu, Mitsuo Gen
    2004 年 124 巻 10 号 p. 1991-1998
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    In this paper, a decentralized decision-making problem in stochastic environment is formulated as expected value multilevel programming, and chance-constrained multilevel programming according to different decision criteria. In order to solve the proposed stochastic multilevel programming models for finding the Stackelberg-Nash equilibrium, genetic algorithm, neural network and stochastic simulation are integrated to produce a hybrid intelligent algorithm. Finally, two numerical examples are provided to illustrate the effectiveness of the hybrid intelligent algorithm.
  • Jiuping Xu, Masato Sasaki
    2004 年 124 巻 10 号 p. 1999-2005
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    The aim of this paper is to extend the technique of order preference by similarity to ideal and anti-ideal alternative to the Grey system. In general, multi-attribute problems in real-life situation have uncertain and imprecise data while Grey system theory is to deal with a linguistic decision process to solve the decision-making problems under uncertainty environment. In this paper, the rating of each alternative and the weight of each criterion are described by linguistic terms that can be expressed in Grey numbers. Further, vertex method is proposed to calculate the distance between Grey numbers. According to the concept of the technique of order preference by similarity to ideal and anti-ideal alternative, a closeness coefficient is defined to determine the ranking order of all alternatives by calculating the distances to both the Grey ideal solution (GIS) and Grey anti-ideal solution (GAIS) simultaneously. Finally, an example is shown to highlight the procedure of the proposed method.
<システム>
  • Kazuaki Abe, Tetsuo Yamada, Masayuki Matsui
    2004 年 124 巻 10 号 p. 2006-2013
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    Under the BTO environment, stochastic assembly lines require design methods which shorten not only the production lead time but also the ready time for the line design. We propose a design method for Assembly Line Systems (ALS) in Yamada et al. (2001) by using Genetic Algorithm (GA) and Adam-Eve GA, in which all design variables are determined in consideration of constraints such as line length related to the production lead time. First, an ALS model with a line length constraint is introduced, and an optimal design problem is set to maximize the net reward under shorter lead time. Next, a simulation optimization method is developed using Adam-Eve GA and traditional GA. Finally, an optimal design example is shown and discussed by comparing the 2-stage design by Yamada et al. (2001) and both the GA designs. It is shown that the Adam-Eve GA is superior to the traditional GA design in terms of computational time though there is only a slight difference in terms of net reward.
  • Yukiko Orito, Takaaki Motoyama, Genji Yamazaki
    2004 年 124 巻 10 号 p. 2014-2018
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    It is well known that index fund selections are important for the risk hedge of investment in a stock market. The`selection’means that for`stock index futures’, n companies of all ones in the market are selected. For index fund selections, Orito et al.(6) proposed a method consisting of the following two steps : Step 1 is to select N companies in the market with a heuristic rule based on the coefficient of determination between the return rate of each company in the market and the increasing rate of the stock price index. Step 2 is to construct a group of n companies by applying genetic algorithms to the set of N companies. We note that the rule of Step 1 is not unique. The accuracy of the results using their method depends on the length of time data (price data) in the experiments. The main purpose of this paper is to introduce a more`effective rule’for Step 1. The rule is based on turnover. The method consisting of Step 1 based on turnover and Step 2 is examined with numerical experiments for the 1st Section of Tokyo Stock Exchange. The results show that with our method, it is possible to construct the more effective index fund than the results of Orito et al.(6). The accuracy of the results using our method depends little on the length of time data (turnover data). The method especially works well when the increasing rate of the stock price index over a period can be viewed as a linear time series data.
  • Kiyohiko Hattori, Keiki Takadama, Hiroshi Ueno, Mitsushige Oda
    2004 年 124 巻 10 号 p. 2019-2026
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    This paper analyzes the capabilities of two methods that specify a failure area in a large-scale space structure: (1) the adjoining module decision (AMD) method proposed by our previous research; and (2) a Byzantine-based decision (BBD) method which is a general method in the context of a distributed processing approach. In this paper, we investigate methods that specify a failure area from the viewpoint of a distributed processing approach. Through intensive simulations, we finally concluded that the adjoining module decision method has better capability than a Byzantine-based decision method. Other implications are summarized as follows; (1) A Byzantine-based decision method cannot specify a failure area if the number of broken modules is over 1/3 of the connected modules, while the adjoining module decision method can; (2) A Byzantine-based decision method requires more time to decide a failure area than the adjoining module decision method; (3) Neither methods can specify a failure area stably when methods do not disconnect broken modules; and (4) an AMD method does not depend on the shape of the failure area, which indicates that the AMD method is robust in many cases.
<ソフトウェア・情報処理>
  • Jyunichi Taniguchi, Xiaodong Wang, Mitsuo Gen, Takao Yokota
    2004 年 124 巻 10 号 p. 2027-2033
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    Location-allocation problem is known as one of the important problems faced in Industrial Engineering/Operations Research fields. One of important logistic tasks is transfer of manufactured products from plants to customers. If there is a need to supply products to large number of customers in a wide area, it is disadvantageous to deliver products from the only central distribution center or direct from plants. It is suitable to build up local distribution centers. In literature, different location models have been used according to characteristics of a distribution area. However, most of them related the location problem without obstacle. In this paper, an extended location-allocation problem with obstacles is considered. Since this problem is very complex and with many infeasible solutions, no direct method is effective to solve it, we propose a hybrid Genetic Algorithm (hGA) for effectively solving this problem. The proposed hGA combines two efficient methods based on Lagrangian relaxation and Dijkstra’s shortest path algorithm. To improve the performance of the proposed hGA, a Fuzzy Logic Controller (FLC) approach is also adopted to auto-tune the GA parameters.
  • Atsushi Wada, Keiki Takadama, Katsunori Shimohara, Osamu Katai
    2004 年 124 巻 10 号 p. 2034-2039
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    As a first step toward an analysis of the capabilities of adaptive systems, including learning and evolution, we focus on the Learning Classifier System (LCS) and compare it with Reinforcement Learning (RL) that adopts the Function Approximation (FA) method. An analysis of this comparison found an equivalence of learning processes between both the two models, which brings the mathematical framework of the LCS’s learning process to the level of RL with FA. Our analysis also clarified the limitations of the results.
<エレクトロニック・コマース>
  • Jonghun Park
    2004 年 124 巻 10 号 p. 2040-2043
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    New developments in service-oriented computing allow the creation of Web services-enabled business processes that can be defined through the composition of a set of existing services. Recently many proposals have emerged along this line, including BPSS, BPML, and BPEL4WS, with little effort spent on the development of systematic methods to effectively support composition of partner business processes that dynamically discover and interact with publicly available business processes at runtime. Motivated by this, this paper proposes a framework for automatically synthesizing the partner business processes, named Co-BP, from the formal description of a target business process, and discusses preliminary work on implementing the proposed ideas by use of XSLT. The proposed framework bases its foundation on the well-known Petri net techniques such as reachability analysis and theory of regions. It is expected that the work presented in this paper can promote the rapid adoption of Web services composition languages by providing an automated tool for facilitating seamless integration of distributed business processes.
論文
<電子・集積回路>
  • 福原 雅朗, 吉田 正廣
    2004 年 124 巻 10 号 p. 2044-2050
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    A neuron MOS transistor has been proposed which behaves much more intelligently than a conventional MOS transistor does. In this paper, we propose a Hamming distance detector with a large noise margin using the neuron MOS transistors. The proposed circuit accepts two bit-streams to be compared in parallel, and is capable of detecting if the two bit-streams are identical (“exact match"), and if the Hamming distance between the two bit-streams is within a certain range (“near match"). Moreover, the “acceptable" range of the Hamming distance (in case of “near match") can be soft-programmed. The operating characteristics of the circuit are also analyzed in detail. Furthermore, these analyses are fully confirmed by simulation using the circuit analysis program HSPICE.
<医用電子・生体工学>
  • 小矢 美晴, 溝尻 勳
    2004 年 124 巻 10 号 p. 2051-2056
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    The arteriosclerosis is on the increase with an aging or change of our living environment. For that reason, diagnosis of the common carotid artery using echocardiogram is doing to take precautions carebropathy. The arteriosclerosis of the common carotid artery is diagnosed using Intima-Media Thickness (IMT) which is obtained from echocardiogram. In order to measure IMT from echocardiogram, it is required to detect a border which is a boundary between vessel tissue layers. The method of border detection requires reproducibility, high accuracy and high-speed.
    In this paper, we propose the high-accuracy and high-speed detection method by Gauss-Markov estimation. About high-accuracy, it realized by attaching importance to high reliable candidate point of border. And, about high-speed, it realized by calculating matrix only one time.
  • 山下 和彦, 野本 洋平, 梅沢 淳, 宮川 晴妃, 川澄 正史, 小山 裕徳, 斎藤 正男
    2004 年 124 巻 10 号 p. 2057-2063
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    There is the increasing concern of the society to prevent the tumbling of the aged. The study of the static, as well as dynamic aspects, such as the muscular strength of the lower-limb and the postural stability, should be developed, especially from the viewpoint of the aged. This paper focuses on the external observation of the foot and toenail, as being correlated to the physical functions of the lower-limb against tumbling. The lower-limb functions are evaluated in terms of the 10 m walking time, the toe-gap force and single-foot standing period. The correlation to the personal tumbling experiences is also examined. It is seen that the groups, which exhibit external abnormalities in the foot and the toenail, generally decline in the muscular strength and postural stability. They also have more frequent tumbling experiences and express in their concern of the danger of tumbling. It seems that those shapes abnormalities can indicate, to some extent, the tumbling danger of the aged.
<音声画像処理・認識>
  • 廣川 洋一, 阿部 重夫
    2004 年 124 巻 10 号 p. 2064-2071
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    In this paper, we propose a new method for training L1 (the linear sum of slack variables) and L2 (the square sum of slack variables) support vector regressors, which is based on the steepest ascent method previously developed for pattern classification. We test our method using two benchmark data sets and show that training of L2 support vector regressors is faster than that of L1 support vector regressors when polynomial and dot product kernels are used and that the working set selection by the exact Karush-Kuhn-Tucker (KKT) conditions does not always result in faster training than by the inexact KKT conditions.
  • Masanori Sugisaka, Xinjian Fan
    2004 年 124 巻 10 号 p. 2072-2078
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    Fast and robust face detection is necessary for many practical applications, especially for a high-performance face recognition system. However, most existing robust algorithms are computationally expensive. To reduce computational cost while retaining high detection accuracy and robustness, in this paper, we propose a new face detection method which combines a neural network (NN) and a genetic algorithm (GA). The NN serves as a face filter while the GA is used to search an image efficiently. The GA searches the image with a population of individuals, each representing a subwindow in the image. The subwindows are evaluated by how well they match the NN-based face filter. A face is indicated when the filter response of the best individual is above a given threshold. Experimental results show the effectiveness of the proposed method.
  • 大山 航, 稲見 真樹, 若林 哲史, 木村 文隆, 鶴岡 信治, 関岡 清次
    2004 年 124 巻 10 号 p. 2079-2086
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    In this paper, we propose a new automatic tracking method for 2-D motion of Regional myocardial motion in high-frame rate echocardiography by using the correlation method with connective multiple ROIs. Several automatic tracking methods by means of correlation method and image gradient method are proposed. However, these method tend to fail to track the motion of object because of a considerable speckle noise contained in echocardiography. The proposed method in this paper is a combination method of maximization for correlation of brightness distribution and optimization the geometric location of ROIs. We compared tracking accuracies of conventional method and proposed method in evaluation test. Results of these tests shows that proposed method is able to derive more accurate trajectories from echocardiography than conventional method.
<制御・計測>
  • 小林 正樹, 長坂 保典, 江口 治
    2004 年 124 巻 10 号 p. 2087-2092
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    In this paper we propose a design method of the various filters utilizing an adaptive filter that has the restricted algebraic moment of the weight. When an amplitude sampling filter such as the surface acoustic wave filter or a digital filter are realized, the dispersion of the weight of these filters becomes a problem. This dispersion is given by the algebraic moment of degree N of the weight in this paper. The adaptive system consists of the ideal filter and an adaptive filter that estimates the impulse response of the ideal filter. The adaptive algorithm is a gradient algorithm by which the algebraic moment of the weight was restricted. We prove the existence of the unique convergence value of the weight and formulate the convergence condition of the adaptive algorithm. As a result, the design of the filter that has the restricted algebraic moment of the weight becomes easy, and the application of this design technology extends widely. Finally, the convergence performances of this adaptive filter are verified by the computer simulation.
  • 内藤 浩行, 東 剛人, 西村 昭彦, 藤田 政之
    2004 年 124 巻 10 号 p. 2093-2100
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    The purpose of this paper is to design a congestion controller for TCP/AQM networks based on the H control theory, and validate the effectiveness of the designed H congestion controller via a testbed which is the small-scale network for verification of the designed H congestion controller. Firstly, a nonlinear and the linearized model of TCP/AQM networks are introduced, and control objectives are described. The H congestion controller is designed by solving the mixed sensitivity problem for the linearized model. In order to verify the designed H congestion controller, we show simulation results using SIMULINK and ns-2 simulator, and compare the congestion controller via PI control with the designed H congestion controller. Finally, we develop a testbed for verification of the effectiveness of the designed H congestion controller which is implemented to the testbed by improving Linux kernel and carry out an experiment. As experimental results using the developed testbed, we show the queue size in the router, where the designed H congestion controller is implemented, and confirm that the control objectives are achieved.
  • 中村 浩二, 河辺 徹, 寅市 和男, 片岸 一起, 夏 広詣
    2004 年 124 巻 10 号 p. 2101-2107
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    A continuous-time signal in the real world must be approximated within a discrete-time signal space for processing in digital devices in many actual engineering applications. One of the effective discrete approximation methods is to employ signals composed of the spline functions. In order to use such signals effectively, however, the sampling bases of such signal spaces play an important role in signal processing. On the other hand, recently, it becomes commonplace to use digital controllers in control engineering field. Therefore, it is important to assume input signals be composed of piecewise polynomials in the state functional spaces. The existence of sampling bases in the state functional spaces had been shown in previous research conducted by the authors. Therefore, in this paper, a computation method of sampling bases in the state functional spaces by using spline functions of general degrees is proposed. Furthermore, the sampling bases composed by spline functions of 2nd and 3rd degrees are concretely calculated as practical examples.
  • 生田 顕, 太田 光雄, 小川 仁士
    2004 年 124 巻 10 号 p. 2108-2115
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    In this study, a signal processing method considering not only linear correlation but also the higher order nonlinear correlation information is proposed on the basis of fuzzy observation data, in order to find the mutual relationship between sound and electromagnetic waves leaked from an electronic information equipment. More specifically, by applying the well-known fuzzy probability to an expression on the multi-dimensional probability distribution in an orthogonal expansion series from reflecting hierarchically various type correlation information, a method to estimate precisely the correlation information between each variable from the conditional moment statistics of fuzzy variables is proposed as a trail. The effectiveness of the proposed theory is experimentally confirmed by applying it to the observation data leaked from VDT periphery in the actual work environment.
  • 舩曳 繁之, 西山 晋司, 田中 俊彦, 藤原 順一, 馬庭 壯吉, 酒井 康生
    2004 年 124 巻 10 号 p. 2116-2122
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    There are differences between walkway walking and walking on a treadmill. It is considered that these differences are based on the fact that the walking on the treadmill is a passive motion, while the walkway walking is an active motion. The differences in walking between on a floor and on a treadmill are investigated using the electromyograph and on the oral questionnaires from subjects. The obtained knowledge is as follows.
    (1) The muscular activity of the legs in walking on the treadmill without the tractive force is smaller than that in walking on the floor.
    (2) The walking on the treadmill with 60% of the tractive force being equivalent to the walkway walking from the rear downward of 30 degrees becomes similar to the usual walking on the floor.
    This paper proposes a novel treadmill with a function of simulating walkway-walking. The developed treadmill has a walking-load device towing the subject from the rear downward and controlling the walking speed according to the position of subject on the treadmill. The verification experiment of walking on the developed treadmill shows the availability to gait training and rehabilitation.
  • 後藤 信夫, 宮崎 保光
    2004 年 124 巻 10 号 p. 2123-2131
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    Precise patterning of optical waveguides is required in fabrication of integrated-optic circuits. In this paper, stage control for laser drawing system with 10nm precision is investigated. The precise control is performed by analog and digital feedback with pulse stages and piezo stages. Mechanical fluctuation and vibration due to pulse drive by stepping motors are compensated by piezo stage. Further, cooperation of pulse and piezo stages is demonstrated to make smooth translation of stage possible.
  • Zi-Jiang Yang, Hideto Iemura, Shunshoku Kanae, Kiyoshi Wada
    2004 年 124 巻 10 号 p. 2132-2139
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    This paper considers the identification problem of continuous-time systems with unknown time delay from sampled input-output data. By using a digital pre-filter, an approximated discrete-time estimation model is first derived, in which the system parameters remain in their original form and the time delay need not be an integral multiple of the sampling period. Then an unseparable nonlinear least-squares (UNSEPNLS) method and a separable nonlinear least-squares (SEPNLS) method for identification of transfer function parameters and time delay are derived. Furthermore, an unseparable nonlinear instrumental variable (UNSEPNIV) method and a separable nonlinear instrumental variable (SEPNIV) method are proposed, to eliminate estimate bias due to measurement noise. Simulational results show that the UNSEPNIV and SEPNIV methods yield consistent estimates in the presence of measurement noise.
<ロボティクス>
  • 橋本 岳, 熊岡 洋介, 山本 茂広, 牧野 秀成, 安陪 稔
    2004 年 124 巻 10 号 p. 2140-2147
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    We have been investigating the flexible transporting system that can be realized by cooperative tasks of multiple mobile robots. This paper discusses the navigation method of mobile robots by beams of lights, we call them “light lines". Our method is simple and stable. The system is composed of a main robot and multiple sub robots. The main robot measures positions of the sub robots and a target, and guides the sub robots to the target. The sub robots move along light lines that projected from the main robot based on the navigation rule of “line trace robot". The features of our method are as follows: The sub robots are continuously steered by output of optical sensors. Therefore, the equipment of the sub robots is simple and low cost for practical applications. In this paper, the validity of this system is verified by some experiments of such a s the simultaneous control of four sub robots for the cleaning tasks. Then, it is shown that this method can easily detect obstacles that exist between the sub robots and the target by the light section method as the secondary application of the light lines.
  • 孫 悠洛, 鎌野 琢也, 安野 卓, 鈴木 茂行, 原田 寛信
    2004 年 124 巻 10 号 p. 2148-2154
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    This paper describes the generation of adaptive gait patterns using new Central Pattern Generators (CPGs) including motor dynamic models for a quadruped robot under various environment. The CPGs act as the flexible oscillators of the joints and make the desired angle of the joints. The CPGs are mutually connected each other, and the sets of their coupling parameters are adjusted by genetic algorithm so that the quadruped robot can realize the stable and adequate gait patterns. As a result of generation, the suitable CPG networks for not only a walking straight gait pattern but also rotation gait patterns are obtained. Experimental results demonstrate that the proposed CPG networks are effective to automatically adjust the adaptive gait patterns for the tested quadruped robot under various environment. Furthermore, the target tracking control based on image processing is achieved by combining the generated gait patterns.
<ソフトコンピューティング>
  • Xinshun Xu, Zheng Tang, Xiaoming Chen, Jiahai Wang
    2004 年 124 巻 10 号 p. 2155-2161
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    The minimum vertex cover(MVC) problem is a classic graph optimization problem. It is well known that it is NP-Complete problem. In this paper, a modified Hopfield neural network is presented for the minimum vertex cover problem. In the modified Hopfield neural network, a correction term is introduced into the motion equation. With this correction term, the modified Hopfield network can find optimal or near-optimal solutions for the minimum vertex cover problem in higher probability. Extensive simulations are performed, and the results show that the modified Hopfield neural network works much better than other existing algorithms for this problem on both random graphs and DIMACS benchmark graphs.
  • Xinshun Xu, Zheng Tang, Jiahai Wang
    2004 年 124 巻 10 号 p. 2162-2168
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    Abstract By analyzing the dynamic behaviors of the transiently chaotic neural network, we present a improved transiently chaotic neural network(TCNN) model for combinatorial optimization problems and test it on the maximum clique problem. Extensive simulations are performed and the results show that the improved transiently chaotic neural network model can yield satisfactory results on both some graphs of the DIMACS clique instances in the second DIMACS challenge and p-random graphs. It is superior to other algorithms in light of the solution quality and CPU time. Moreover, the improved model uses fewer steps to converge to saturated states in comparison with the original transiently chaotic neural network.
  • 松村 幸輝, 後藤 道彦, 浜松 芳夫
    2004 年 124 巻 10 号 p. 2169-2178
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    This paper describes a negotiation agent system based on the fuzzy decision making. The method of seeking appropriate membership functions and a reasonable agreement point was examined by means of the genetic programming technique with the coupled chaos system, which is an intelligent principle. The negotiation rule is based on the negotiation model expressed by the utility theory in the process of decision making. And the concession process was modified with the opponent’s movement and the persistence of each negotiator. In order to search for a membership function more efficiently, the dynamic state of symbiosis between individuals, which was caused by the coupled chaos system, was taken advantage of. Then the effectiveness of the technique was examined by applying it to a practical negotiation case which needs cooperative decision making. As a result, the following findings were obtained. This technique helps discover practicable membership functions in a vast search area, and achieve the solution search with high efficiency. This technique is also considered to be applied to the negotiation support easily.
  • 横瀬 義雄, 泉 照之
    2004 年 124 巻 10 号 p. 2179-2186
    発行日: 2004年
    公開日: 2005/01/01
    ジャーナル フリー
    The global warming and the environmental destruction are caused from the mass consumption of energy in a factory in which robotic manipulators are used. It is necessary to take the trajectory planning into considerations in order to save the consumption energy of the manipulator controlled with Point-to-Point (PTP) motion. An optimal trajectory that minimizes the consumption energy can be obtained by solving a two-point boundary value problem. The optimal trajectory can be analyzed easily if the system is linear. However, the dynamical equation of the multi-link manipulator with Coulomb friction has a strong non-linearity. Then the Genetic Algorithm (GA) is applied to obtain the solution of two-point boundary value problem with the strong non-linearity. The GA is the method for searching discrete solution space by genes. However, the trajectory function is continuous with respect to time. In this paper, the coefficients of the approximation functions are determined as the gene’s phenotype. This method has disadvantages that are long computation time and increase the possibility of lapsing into the local optimal solution. So, a new method is proposed in order to shorten the computation time and to lessen the possibility of lapsing into the local optimal solution. It is seen from application to the minimization of the consumption energy of a manipulator with Coulomb friction that a proposed method is effective.
<システム>
研究開発レター
<制御・計測>
<ソフトウェア・情報処理>
feedback
Top