巻号一覧
42 巻, 7 号
選択された号の論文の18件中1～18を表示しています
- |<
- <
- 1
- >
- >|
-
小島 史男, 依藤 大輔, 伊東 一文2006 年 42 巻 7 号 p. 705-711
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーThis paper is concerened with a computational method for recovering induced magnetic fields due to the existence of surface or subsurface cracks on material in critical structure. The inversion formula can be simply represented by an integral equation of the first kind. However, in practical magnetic measurements, the severe ill-posed problem is caused by the observation noise. The problem is transformed into a well-posed difference equation using output least squares approach with a regularization of bounded variations. A fast computational algorithm is also proposed using discrete Fourier transform. Results of numerical experiments to demonstrate the efficacy of the proposed algorithm are reported.抄録全体を表示PDF形式でダウンロード (3855K) -
寺本 顕武, 鶴田 浩輔2006 年 42 巻 7 号 p. 712-718
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーThis paper presents an autonomous directivity microphone system based on the newly proposed spatio-temporal gradient blind source separation. The blind source separation principally uses no a priori knowledge about parameters of convolution, filtering and mixing. In the simplest case of the blind source separation problems, observed mixed signals are linear combinations of unknown mutually statistically independent, zero-mean source signals. The blind signal separation algorithm utilizes the linearity among the four signals: (1) the sound pressure, (2) x, (3) y, and (4) z-directional particle velocities, all of which are governed by the wave equation and the equation of motion. The proposed method, therefore, has an ability to simplify the convolution blind source separation problems into the instantaneous blind source separation over the spatio-temporal gradient space. Several acoustical experiments are performed with the particle velocity microphone (Microflown) successfully instead of sound pressure microphones.抄録全体を表示PDF形式でダウンロード (2297K) -
高橋 秀典, 田中 孝之, 金子 俊一2006 年 42 巻 7 号 p. 719-726
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーIn this paper, the authors achieved a 3D motion capture system by using one camera with triangle markers, named as Mono-MoCap (MMC). The camera parameters were calibrated by means of the modified Direct Linear Transformation (DLT) method. The relation between the image coordinate value and the absolute coordinate value was expressed by the calibrated camera parameters. The absolute coordinate values of apexes of a triangle marker were calculated on the basis of the perspective n point (PnP) problem. As compared with a conventional stereo vision typed motion capture system, MMC had enough accuracy for 3D measurement in the depth of 1.0 to 1.3m from a camera. Also, by putting a triangle marker on each human joint, MMC could capture a human walking motion, and MMC had the wider measuring range than the stereo vision system in the parallel plane of the image frame.抄録全体を表示PDF形式でダウンロード (2900K) -
星 貴之, 篠田 裕之2006 年 42 巻 7 号 p. 727-735
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーIn this paper, we propose a new tactile sensor element to realize practical robot skins. The element acquires not only a contact force but also a contact area within its several-cm-sq. sensing area. By arraying the elements, we can cover a large area with a small number of them, and collect rich tactile information. The structure of the element is very simple; two layers of compressible insulators (urethane foam) which are sandwiched between three pieces of stretchable conductive sheets (conductive fabric). The structure enables us to obtain the contact force and area from the capacitances between the conductive layers. We also propose a method to connect the elements with no long wires. Sensor/communication chips are arranged at the boundary of the elements. The chips measure the capacitances between the conductive layers and send signals through the same layers.抄録全体を表示PDF形式でダウンロード (2942K) -
石原 義之, 千田 有一2006 年 42 巻 7 号 p. 736-743
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーIn this paper, we consider a HH∞ control problem with constraints on closed-loop pole location. In this problem, it is difficult to locate closed-loop poles in preferable region if weighting functions include unstabilizable or undetectable poles close to the imaginary axis. In order to overcome this problem, we present a solvability condition of extended H∞ control problem with pole placement constraints via LMI approach. Finally, we show an application result of the proposed method to a vibration isolation control and verified the performance of the designed controller by experiments.抄録全体を表示PDF形式でダウンロード (2112K) -
大屋 勝敬, 小林 敏弘2006 年 42 巻 7 号 p. 744-749
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーWhen we control real devices by using model reference adaptive controllers, we have an experience that high frequency oscillations with large amplitude occur in the control input signals. This is one of serious problems in actual applications. In this paper a new model reference adaptive controller is developed under the recognition. To generate a control input signal without oscillations, a new estimator using an observer for the output error signal is proposed. Using the estimator, it is shown theoretically that the control input signal converges with any convergent rate to an ideal input signal given by adding the exact model matching input for no initial values and an exponential function. From the property it can be expected that meaningless oscillations are hard to occur in control input signals. Moreover, it is also proved theoretically that the transient output error responses can be specified by setting design parameters.抄録全体を表示PDF形式でダウンロード (1675K) -
伸張型LMIに基づく既存結果の拡張川田 昌克, 蛯原 義雄, 陳 幹2006 年 42 巻 7 号 p. 750-757
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーIn this paper, we shall provide a new redundant descriptor approach to H∞ state-feedback controller synthesis based on the strategy of the recently developed dilated LMIs. It is known that both of the redundant descriptor and the dilated LMI approaches are effective for robust/gain-scheduled H∞ state-feedback controller synthesis when the state-space matrices of the plant are affected by uncertain/scheduling parameters. These approaches are closely related in the sense that they enlarge the size of LMIs and employ additional matrix variables to obtain more tractable LMI conditions. However, in the existing redundant descriptor approaches, some assumptions have been imposed on the the state-space matrices of the plant to avoid difficulties arising from the fictious direct feedthrough term from the disturbance inputs to the performance outputs of the resulting descriptor form. This is in sharp contrast with the dilated LMI approaches that do not require any specific assumptions on the state space matrices. To fill the gap between these two approaches, in the present paper, we revisit the redundant descriptor approaches and provide a new way to recast the state-space representations into descriptor forms by employing yet new additional variables. By employing those new additional variables, it turns out that those LMIs resulting from the descriptor approaches coincide with the known dilated LMI conditions. The role of those additional variables can be seen clearly if we view from the standpoint of the redundant descriptor approaches.抄録全体を表示PDF形式でダウンロード (1592K) -
杉江 俊治, 植田 克樹2006 年 42 巻 7 号 p. 758-764
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーThis paper considers local stability analysis of the systems with input saturation via polynomial programming. First, describing the saturated inputs in terms of logic variables, we derive a local stability condition with the aid of generalized S-procedure of polynomials with higher degree of freedom. Then, we obtain a Lyapunov function using Sum of Squares decomposition, which specifies the region of attraction. Moreover, it is shown that the method can be easily extended to the case where piecewise quadratic Lyapunov functions are adopted. Finally we demonstrate its effectiveness through numerical examples.抄録全体を表示PDF形式でダウンロード (1734K) -
ダオ ミンクアン, 井上 拓哉, 劉 康志2006 年 42 巻 7 号 p. 765-774
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーThis paper aims at developing a practical automatic parking technology for 4-wheeled cars, which is able to park cars subject to parking space constraint and steering angle saturation. The feedback law is designed based on the switching control method proposed by the authors for chained systems. The focus of this paper is to prove a condition on the initial state under which the stability is guaranteed even when the steering angle is saturated. After that, parking algorithms are proposed based on this condition. The proposed method is validated by numerous parking experiments.抄録全体を表示PDF形式でダウンロード (2480K) -
低次の切換時補償器の設計浅井 徹2006 年 42 巻 7 号 p. 775-782
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーThis paper proposes a general synthesis framework to attenuate disturbance responses which are caused due to switching. The objective function to be minimized is the worst-case L2 gain from the past disturbance to the future output. The problem formulation is given in a general manner so that controller switching is a special case. In particular, this paper assumes that some of state variables before switching can be exactly available and also that some of those after switching can be manipulated without affecting control output directly. Necessary and sufficient conditions for synthesis are given in terms of LMIs. It follows that solutions can be obtained efficiently. Moreover, this paper shows that reduced order compensators exist under the above assumptions.抄録全体を表示PDF形式でダウンロード (1453K) -
小林 孝一, 井村 順一2006 年 42 巻 7 号 p. 783-790
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーThis paper proposes an integrated optimal design of a two-level hierarchical system, which is constructed by a discrete-time high-level system such as a production-scheduling system and a continuous-time low-level control system. First, an integrated optimal design problem is formulated using a cost function composed of continuoustime and discrete-time cost functions, and an optimal control input of this problem based on the finite-time optimal control problem is derived. Next, the difference in performance between the proposed and the standard design methods is given explicitly. Finally, the effectiveness of the proposed method is shown by numerical examples on the slab heating control for reheating furnaces.抄録全体を表示PDF形式でダウンロード (1919K) -
有本 卓, 吉田 守夫, 裴 芝薫2006 年 42 巻 7 号 p. 791-798
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーStability of “blind grasping” of a 2-Dimensional (2-D) rigid object by a pair of multi-DOF planar robot fingers with hemispherical ends is analyzed on the basis of Lagrange's equation of motion of the overall fingers-object system subject to point contact constraints and rolling constraints. It is assumed that motion of the fingers and object is confined to a vertical plane and hence it is affected by the gravity. Blind grasping is defined as a manner of grasp when 1) the control signal is constructed on the basis of finger kinematics and measurement data on finger joints without knowing object kinematics or using external sensings such as visual or tactile sensing and 2) at the same time the closed-loop dynamics converges to a state of force/torque balance. In fact, human can grasp an object even if they close eyes, provided the hand is preshaped toward the object placed on a table. Stability is defined on a constraint manifold and also the new concept called “EP-manifold” (EP: Equilibrium Point) is introduced when the overall fingers-object system is redundant in DOF. It is shown that any solution of the closed-loop dynamics lying on the constraint manifold converges to a low-dimensional EP-manifold satisfying force/torque balance and the speed of convergence is exponential in time.抄録全体を表示PDF形式でダウンロード (2220K) -
小林 孝一, 井村 順一2006 年 42 巻 7 号 p. 799-805
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーThis paper proposes a new approach to reducing the computational time for solving the model predictive control problem of a class of hybrid systems, by representing a finite automaton with a smaller number of binary variables. First, the technical issue in the modeling on the field of real numbers is pointed out, and an implicit system expression on the binary field GF(2) is presented. Next, a systematic method to transform this system model into the state equation on the field of real numbers by way of the state equation on GF(2) is proposed, which thus enables us to reduce the original control problem into the mixed integer quadratic problem with a smaller number of binary variables. Finally, the effectiveness of the proposed method is shown by numerical examples.抄録全体を表示PDF形式でダウンロード (1637K) -
穂高 一条2006 年 42 巻 7 号 p. 806-813
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーRealization theory for linear periodic systems with real coefficients is developed in this paper. It is a well-known fact that a weighting pattern matrix is realizable as a linear periodic system if and only if the weighting pattern matrix is separable and periodic. This fact, however, can not handle a reasonable question when a weighting pattern matrix can be realized as a linear periodic system with a specific period of time. This paper attacks the question to derive a new necessary and sufficient condition for the period-specific realizability. Under the condition, a linear periodic realization for a given weighting pattern matrix is obtained. The realization procedure proposed in this paper essentially consists of integrations over a prescribed interval and computations of a real logarithm of a matrix. Thus, the procedure provides a transparent connection between weighting pattern matrices and their state-space representations.抄録全体を表示PDF形式でダウンロード (1989K) -
小林 祐一, 藤井 博基, 細江 繁幸2006 年 42 巻 7 号 p. 814-821
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーThis paper proposes a reinforcement learning method for dynamic control problems with holonomic constraints. The learning method is applicable to problems where the actual motion of the system is restricted to lower-dimensional submanifolds, so long as certain conditions are satisfied. Such dynamic control problems occur in robotic manipulation, which usually includes some holonomic constraints between the object and the robot or the environment. By introducing nonlinear mapping to one-dimensional space and approximating the boundary of a discontinuous reward function, the proposed method results in effective learning. The method is evaluated in a one degree of freedom object rotating task with contact force considerations. The effectiveness of the proposed learning method was verified by comparison to ordinal Q-learning and Dyna without the proposed mapping method.抄録全体を表示PDF形式でダウンロード (3332K) -
鈴木 桂輔, 山田 喜一2006 年 42 巻 7 号 p. 822-828
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーWe investigated the effectiveness of collision avoidance assistance system in a driving simulator. In this study, we analyzed the combined error frequency of a man-machine system, with respect to driver performance and system reliability. In the analysis of driver performance, braking timing to avoid the collision both in normal operation and malfunction of collision avoidance assistance system were analyzed. In short, not only the positive effects of the driving assistance system but also the negative effects results from driver's excessive dependence on the systems were considered to analyze the combined error frequency. It was revealed that the combined error frequency when the system was used is lower than that when they were not used, even though the drivers' dependence on the warning system (i.e. delay of braking timing to avoid the collision) was observed.抄録全体を表示PDF形式でダウンロード (2356K) -
西馬 功泰, 後藤 幸夫, 熊澤 宏之, 駒谷 喜代俊, Daniel NIKOVSKI2006 年 42 巻 7 号 p. 829-836
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーWe propose a general prediction method based on the efficient computation and online update of the Singular Value Decomposition (SVD) of historical data. The SVD is fundamental to many data modeling algorithms, but the traditional methods for computing it require large computational costs. By adopting a fast sequential SVD updating scheme, the tasks of prediction, imputation of missing values, and model updating can be performed very quickly. In this paper, an application of our method to route travel time prediction is described. Using real travel time data from short sections (links) on expressway, we evaluated prediction performance of travel time on longer section (route) including the links. Experimental comparisons with several statistical machine learning methods suggest that our linear prediction method can achieve similar prediction performance (prediction error) to other nonlinear methods at less computaional cost.抄録全体を表示PDF形式でダウンロード (4394K) -
井上 慎太郎, 川嶋 健嗣, 舩木 達也, 香川 利春2006 年 42 巻 7 号 p. 837-843
発行日: 2006/07/31
公開日: 2009/03/27
ジャーナル フリーThe measurement-integrated simulation (MI-simulation) is a numerical simulation which feedbacks information of experimental results in order to perform simulations under real conditions. Even if using rough grids, the calculation results become closer to those of experiments. Therefore, the reduction of the calculation time is expected. In this paper, we proposed a monitoring system for pipe flow including unsteady condition using the simulation. The MI-simulation was applied to airflows passed an orifice plate in the pipeline. Firstly, the measured velocities at the downstream of the orifice plate under steady condition were feedback to the simulation. Secondly, the effect of the points and number of velocities to the simulation were investigated. Then, the MI-simulation was applied to an oscillatory flow. The effectiveness of the method was confirmed compared with the experimental results. The calculation time could be much shortened compared with general simulations.抄録全体を表示PDF形式でダウンロード (2907K)
- |<
- <
- 1
- >
- >|