제어로봇시스템학회:학술대회논문집
Institute of Control, Robotics and Systems (ICROS)
 기타
Domain
 Machinery ＞ Robot/Automated Machinery
1993.10

This report proposes a novel method of identification of continuous timedelay systems from sampled inputoutput data. By the aid of a digital prefilter, an approximated discretetime 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 th sampling period. Then an identification method combining the common linear least squares(LS) method or the instrumental variable(IV) method with the genetic algorithm(GA) is proposed. That is, the timedelay is selected by the GA, and the system parameters are estimated by the LS or IV method. Furthermore, the proposed method is extended to the case of multiinput multioutput systems where the timedelays in the individual input channels may differ each other. Simulation resutls show that our method yields consistent estimates even in the presence of high measurement noises.

This paper presents a generalized truncated least, squares adaptive algorithm and a twostage design method. The proposed algorithm is directly derived from the normal equation of the generalized truncated least squares method (GTLSM). The special case of the GTLSM, the truncated least squares (TLS) adaptive algorithm, has a distinct features which includes the case of minimum steps estimator. This algorithm seemed to be best in the deterministic case. For real applications in the presence of disturbances, the GTLS adaptive algorithm is more effective. The twostage design method proposed here combines the adaptive control system design with a conventional control design method and each can be treated independently. Using this method, the validity of the presented algorithms are examined by the simulation studies of an indirect adaptive control.

Freeform surface machining is usually performed with an NC milling machine and a ball end milling cutter. Since this conventional method is basically sculpting on a plane, it is not suitable for three dimensional body machining. This article will introduce a new machining method for three dimensional body with freeform surface and newly developed machine tool suitable for such machining.

In this paper, a design method of controller which incorporates pole restriction into implicit self tuning algorithm is proposed. The idea behind pole restriction is that the closed loop poles of the system are restricted to a userchosen circle in the region to meet maximum percentage overshoot and settling time specification. Most algorithm based on pole restriction are explicit schemes involving a parameter estimation and synthesis stage to obtain controller parameters. The object of this paper is to have an algorithm that has the idea of pole restriction and the simplicity of the implicit approach.

In this paper, we attempt to estimate the state of a finite state system. In such system, we can observe time series data which has some significant behaviors corresponding to its system states. The behavior is characterized by feature parameters extracted from time series. Our thought is that the system output time series data is expressed as a sequence of behavior patterns which are represented by clusters in feature parameters space. An algorithm jointing fuzzy clustering to fuzzy finite state transition model is suggested.

In this paper, the estimation of the structure of a logic circuit under test is made from the observation of the inputoutput correlation function by use of Msequence, from which we can estimate whether or not any fault exist in the logic circuit. Especially, investigation was made in case of the 2_stage logic circuit. We checked theoretically the sequence of correlation function, and we have shown that the correlation function is a function of period of Msequence only, land the appearing number of correlation function in a period is a constant value depending on the logic circuit only. And by computer simulations we have shown that the structure of the circuit under test can be estimated from the observation of sequence of correlation function.

Simple recurrence relations for calculating completion times of various storage polices (unlimited, intermediate storages(FIS), finite intermediate storages(FIS), no intermediate storage(NIS), zero wait(ZW) for serial multiproduct multiunit processes are suggested. Not only processing times but also transfer times, setup (cleanup) times of units and setup times of storages are considered. Optimal scheduling strategies with zero transfer times and zero setup times had been developed as a mixed integer linear programniing(MILP) formulation for several intermediate storage policies. In this paper those with nonzero transfer times, nonzero setup times of units and setup times of storages are newly proposed as a mixed integer nonlinear programming(MINLP) formulation for various storage polices (UIS, NIS, FIS, and ZW). Several examples are tested to evaluate the robustness of this strategy and reasonable computation times.

The adiabatic slirntype autoclave reactor for free radical polymerization of LDPE is represented by a twocompartment fourcell model, which is proven effective to predict the reactor behavior as well as the polymer properties. Since the temperature distribution along the reactor axis plays the central role for the properties of the polymer product, it is important in practice to regulate the temperature in each compartment. The present study aims for the application of the adaptive control algorithm not only in the period of startup but also during the steady state operation. It is shown that the temperature control is significantly improved over the conventional PIDcontrol and this also brings about a reduction of variations in the polymer properties. This study demonstrates the potential application of the adaptive controller for the control of the polymerization reactor operated under the adiabatic condition.

Importance of separation of a nonlinear dynamical system into nonlinear static part and linear dynamical part was insisted in designing a controller for the nonlinear system. We further proposed compensation techniques for oscillation of controlled variables caused by system time delay and compensation of steady state errors caused by modelling errors of the systems. The proposed principle of designing procedure and the compensation methods were discussed by applying them for temperature and level control of an actual tanked water system.

In this paper, it is shown that chaotic nonlinear chemical process can be controlled based on the Poincare map based control algorithm. An isothermal autocatalytic CSTR, which has chaotic dynamics, is successfully controlled and period 2 orbit is generated in a normal chaotic region with small perturbation of the control parameter.

This paper presents a method of estimating 3D surface geometrical features that are necessary for 3D object recognition and image interpretation. The features, such as surface needle maps and curvatures, are computed from range or intensity images. In general, the range and intensity images are prone to noises, and hence the features computed by differentiation calculi on such a noisy image are hardly applicable to industrial recognition tasks. In our approach, we try to obtain a more accurate estimate of the features by using a leastsquares minimization procedure subject to local curvature consistency constraints. The algorithm is robust with respect to noises and is completely independent of the viewpoint at which the image is taken. The performance of the ajgoritlim is evaluated using both synthetic data and real intensity images.

A system to measure 3dimensional coordinates of large structures is presented in this paper. In order to enable accurate measurements calibration of the measuring system is important. Feature of our system are that by introducting one simplified calibration procedures setting of the measuring system is simplified and also that tracking performance is increased. An experimental result reveals the applicability of our systems.

The aim of the present paper is to show the effectiveness of Genetic Algorithm for data classification problems in which the classification criteria are not the Euclidean distance. In particular, in order to improve a search performance of Genetic Algorithm, we introduce a concept of the degree of population diversity, and propose construction of genetic operators and the method of calculation for the fitness of an individual using the degree of population diversity. Then, we investigate their performances through numerical simulations.

We represent a method of 3D object recognition using multi images in this paper. The recognition process is executed as follows. Object models as prior knowledgement are generated and stored on a computer. To extract features of a recognized object, three CCD cameras are set at vertices of a regular triangle and take images of an object to be recognized. By comparing extracted features with generated models, the object is recognized. In general, it is difficult to recognize 3D objects because there are the following problems such as how to make the correspondence to both stereo images, generate and store an object model according to a recognition process, and effectively collate information gotten from input images. We resolve these problems using the method that the collation on the basis of features independent on the viewpoint, the generation of object models as enumerating some candidate models in an early recognition level, the execution a tight cooperative process among results gained by analyzing each image. We have made experiments based on real images in which polyhedral objects are used as objects to be recognized. Some of results reveal the usefulness of the proposed method.

In this paper we discuss optimization of neural network parameters, such as inclination of the sigmoid function, the numbers of the input layer's units and the hidden layer's units, considering application to recognition of hand written Hangul. Hangul characters are composed of vowels and consonants, and basically classified to six patterns by their positions. Using these characteristics of Hangul, the pattern of a given character is determined by its peripheral distribution and the other features. After then, the vowels and the consonants are recognized by the optimized neural network. The constructed recognition system including a neural network is applied to nonlearning Hangul written by some Korean people, which are the names randomly taken from Korean spiritual and cultural research institute.

In this paper robust stability conditions are obtained for linear dynamical systems under structured nonlinear timevarying perturbations, using absolute stability theory and the concept of dissipative systems. The conditions are expressed in terms of solutions to linear matrix inequality(LMI). Based on this result, a synthesis methodology is developed for robust feedback controllers with worstcase H
$_{2}$ perforrmance via convex optimization and LMI formulation. 
In this paper we analyze the RHLQG/FIRF optimal.contol law presented in [4,5] in order to stabilizes a stochastic linear time varying systems with modeling uncertainty. It is shown by the frequency domain analysis that the RHC is robuster than the LQ control law. Explicit LTR procedures are given to improve the robust performance of RHLQC/FIRF cotrol law. Using the mismatching function technique [8], we propose an LTR method which makes the RHLQG/FIRF controller recover the feedback properties of the R.HC law. Also we compare the LTR performance of the RHLQC/FIRF via simulation with conventional LTR methods.

In this paper, we discuss H
$_{\infty}$ robust performance problem for uncertain system described in a descriptor form. We show that the method based on Riccati equation can be extended to solve this problem. First, such a sufficient condition is given that the system described in a descriptor form is quadratic stable and H$_{\infty}$ norm of a specified transfer function is less than a given level. Using this result, a state feedback law which ensures H$_{\infty}$ robust performance of closed loop system is derived based on a positive definite solution of a Riccati equation. This result shows that a solution of the problem can be also obtained by solving H$_{\infty}$ standard problem for an extended plant. Finally, a design example and simulation results will be given.ven. 
In Part 1, we derived robust stability conditions for an LTI interconnected to timevarying nonlinear perturbations belonging to several classes of nonlinearities. These conditions were presented in terms of positive definite solutions to LMI. In this paper we address a problem of synthesizing feedback controllers for linear timeinvariant systems under structured timevarying uncertainties, combined with a worstcase H
$_{2}$ performance. This problem is introduced in [7, 8, 15, 35] in case of timeinvariant uncertainties, where the necessary conditions involve highly coupled linear and nonlinear matrix equations. Such coupled equations are in general difficult to solve. A convex optimization approach will be employed in this synthesis problem in order to avoid solving highly coupled nonlinear matrix equations that commonly arises in multiobjective synthesis problem. Using LMI formulation, this convex optimization problem can in turn be cast as generalized eigenvalue minimization problem, where an attractive algorithm based on the method of centers has been recently introduced to find its solution [30, 361. In the present paper we will restrict our discussion to state feedback case with Popov multipliers. A more general case of output feedback and other types of multipliers will be addressed in a future paper. 
This paper is concerned with a twodegreeoffreedom control system design for a flexible arm, a twodegreeoffrecdom control system can achieve a robust stability specification and a control performance specification independently. By this property we improve the control performance with maintaining the same robust stability level as that of the oncdcgreeoffreedom control system. At First we design a twodegreeoffteedom control system which includes a feedforward controller and a feedback controller. The feedforward controller can be given by specifying a transfer function of a dcsired closedloop model. We obtain a feedback controller by solving a mixed sensitivity problem. Several numerical results show that twodegreeoffreedom control systems acheive a better control performance than that of onedegreeoffreedom control systems.

It is well known that neural networks can be used to diagnose multiple faults to some limited extent. In this work we present a Multiple Fault Diagnosis Method (MFDM) via neural network which can effectively diagnose multiple faults. To diagnose multiple fault, the proposed method finds the maximum value in the output nodes of the neural network and decreases the node value by changing the hidden node values. This method can find the other faults by computing again with the changed hidden node values. The effectiveness of this method is explored through a neuralnetworkbased fault diagnosis case study of a fluidized catalytic cracking unit (FCCU).

It is difficult to establish a mathematical furnace model and automaticfumace control because of the difficulty in direct measurement of the inner condition of a blast furnace. To solve this problem, we has developed and actually operated a blast furnace operation control system using artificial intelligence tool to be applied to the blast furnace process computer system. Since this system has a function of automatic Treat level control, higher practicality has been proved than the previous guidancelevel expert system. This paper introduces an outline of the system and the result of application.

Artificial muscle actuators are used in various fields. Especially, they are applied to the power assist circuit to make use of their characteristics. The purpose of this paper is to and analyze the power assist circuit using an artificial muscle actuator. As a result, it is found that the operating feeling of the power assist circuit depends mainly on the flow gain of the pneumatic servo valve. The required flow gain is calculated from the proposed model, and the experimental results agreed with the calculated results.

A new learning method of both NNI and NNC by which the NNI identifies precisely the dynamic characteristics of the plant is proposed. For control of ihe nonlinear plant we use two neural networks, one for identification and the other for control. We define a closed loop enor which depends on identification and control error. In the proposed learning method, the closed loop enor is utilized to train the NNI and the NNC. Computer simulation results reveal that the NNC based on proposed method is insensitive to variations of the plant parameters.

In this paper, a fish drying process control system is proposed, which predicts the proper change with time in weight of the material fish and the drying conditions in advance, based on the performance of skilled worker. In order to implement a human expertise into an automated fish drying process control system, an experimental analysis is made and a model for the process is built. The proposed system divided into two procedures: The procedure before drying and the one during drying. The procedure before drying is for the prediction of necessary drying time. To estimate the necessary drying time, first, the proper change in weight for the product is obtained by using fuzzy reasoning. The condition part of the production rule consists of the factors of fish body and the expected degree of dryness. Kext, the necessary drying time is obtained by regression models. The variables employed in the models are the factors, inferred change in weight and drying conditions. The model for the procedure during drying is also proposed for more accurate estimation, which is described by a system of lineardifferential equations.

A welldefined relative degree, which is one of the basic assumptions in adaptive control or nonlinear synthesis problems, is addressed. It is shown that this is essentially a necessary condition for asymptotic tracking in discretetime nonlinear systems. To show this, tracking problems are defined, and a local linear inputoutput behavior of a discretetime system is introduced in relation to a welldefined relative degree. It is then shown that if a plant is invertible and accessible from the origin and a compensator solves the local asymptotic tracking problem, then the plant necessarily has a welldefined relative degree at the origin.

This paper discusses a design of a nonlinear control for a class of singleinput doubleoutput nonlinear mechanical systems. When conventional linearization methods are applied to the mechanical systems, some problems of oscillation and unstable phenomena arise. The proposed nonlinear control system resolves these problems. In this design the eigenvalues of the closedloop nonlinear system are assigned to desired locations and local asymptotic stability of the closedloop system. is guaranteed. The design method is applied to an inverted pendulum system with a moving weight mechanism. Experimental results show that the proposed nonlinear controller is more effective for stability than the usual linear controller.

A new method is proposed for obtaining Volterra kernals of a nonlinear system by use of a nonlinear systems by use of pseudorandom Msequences and correlation technique. Msequence is applied to a nonlinear technique. Msequence is applied to a nonlinear system and the crosscorrelation function between the input and the output displays not only the linear impulse response of the linear part of the system, but also crosssections of the Volterra kernals of nonlinear system. Simulations are carried out for up to 3rd order Volterra kernal, and the results show a good agreement with the theoretical considerations.

In this paper, we consider the problem of a stochastic optimal switching control, which can be applied to the control of a system with uncertain demand such as a control problem of a power plant. The dynamic programming method is applied for the formulation of the optimal control problem. We solve the system of QuasiVariational Inequalities(QVI) using an algoritlim which involves the finite difference approximation and contraction mapping method. A mathematical example of the optimal switching control is constructed. The actual performance of the algorithm is also tested through the solution of the constructed example.

In this paper, design of an automatic grinding robot system for car brazing bead is introduced. Car roof and side panels are joined using brazing, and then the brazing bead is processed so that the bead is invisible after painting. Up to now the grinding process is accomplished manually. The difficulties in automation of the grinding process are induced by variation of position and shape of the bead and nonuniformity of the grinding area due to surface deformation. For each car, the grinding area including the brazing bead is sensed and then modeled using a 2D optical sensor system. Using these model data, the position and the direction of discrete points on the car, body surface are obtained to produce grinding path for a 6 degrees of freedom grinding robot. During the process, it is necessary to sense the reaction forces continuously to prepare for the unexpected circumstances. In addition, to meet the line cycle time it is necessary to reduce the required time in sensing, signal processing, modeling, path planning and data transfer by utilizing realtime communication of the information. The key technique in the communication and integration of the complex information is obtaining infield reliability. This automatic grinding robot system may be regarded as a jump in the intelligent robot processing technique.

Telemanipulator became one of refocused fields with the help of various technology integration concerned. As one of key technologies much simpler opeartion of manipulator by operators and feedback method to operator can be counted. For such purpose virtual reality techniques are being implemented and in this paper tactile feeling among human 5 feelings will be discussed. Movement manipulation in free space and gripper manipulation methods are discussed in combination with operator's organs. Tactile signal of telemanipulator's gripper is also integrated into tactile VR control system. Such tactile VR technology is regarded as one of important roles for advanced robot technology.

In this paper, we design a time optimal controller of a dual arm robot to handle the object. Differently from the masterslave type robot, same priority is imposed on the both of the arms for effective handling the specifed object. For finding a time optimal collisionfree trajectory, a graphical method is applied for the robot with two degree of freedom. Some simulation results show the effectiveness of the proposed method.

This paper deals with the modeling and experiment of a robot system for force/impact control performance. The basic model is composed of a direct drive motor, servo amplifier, link, force sensor and environments. Based on the developed model, the stability of the whole system was analyzed via root locus method. For the force control, integral force compensation with velocity feedback method shows the best performance of all the explicit force control strategies. In dealing with impact, PID position control and the explicit force control method were implemented. Instead of add more damping to the robot system by velocity feedback, we developed a new passive damping method and it was also applied to enhance the damping characteristic of the system.

In this paper, an optimal motion control scheme is proposed for robot manipulators. A simple explicit solution to the HamiltonJacobi equation is presented. The optimization of motion control is based on the mininization of the torque term affecting the kinetic energy and the augmented error which has the firstorder stable dynamics for the position and velocity tracking error. In the presence of parametric uncertainty, an adaptive control scheme using the optimal principle is proposed. The global stability of the closedloop system is guaranteed by the Lyapunov stability approach, The effectiveness and feasibility of the proposed control schemes are shown by simulation results.

The POSTECH Hand adopting coupled tendon driven technique with planar two fingers is developed. The hand is designed to emulate principal motions of the human hand which has two and three joints respectively. Its kinematic parameters are determined through a parameter optimizing technique to aim at improving the isotropy of fingertip motions with new criterion functions of design. For the control of the hand, tension and torque control algorithms are developed. Based on the virtual stiffness concept, we develop the stiffness control method of a grasped object with redundant finger mechnism and investigate experimentally.

Since the welding process is complex and highly nonlinear, it is very difficult to accurately model the process for realtime control. In this paper, a discretetime transfer function matrix model for gas metal arc welding process is proposed. Although this linearized model is valid only around the operating point of interest, the adaptation mechanism employed in the control system render this model useful over a wide operating range. A multivariable onestepahead adaptive control strategy combined with a recursive leastsquares method for online parameter estimation is implemented in order to achieve the desired weld bead geometries. Command following and disturbance rejection properties of the adaptive control system for both SISO and MIMO cases are investigated by simulation and experiment.

A method for the closedloop control of the torsional tip motion of a piezopolymer actuator is presented. The application of Lyapunov's direct, method to the problem is explored. A feedback control of the torsional tip motion of tile piezopolymer actuator is derived by considering tile time rate of change of the total energy of the system. If the angular velocity of the tip of the actuator is known, all the modes of tile actuator can be controlled simultaneously. This approach has tile advantage over the conventional methods in the respect that it allows one to directly with tile system's partial differential equations without resorting to approximations.

A theoretical and experimental study is presented for the force holding control of a miniature robotic ringer which is driven by a pair of piezoelectric unimorph cells. In the theoretical analysis, one finger is modeled as a flexible cantilever with a tactile force sensor at the tip and the mate of the finger is a solid beam supposed with sufficient stiffness. Further, the force sensor is modeled by a onedegreeoffreedom, massspring system and the output of sensor is then described by the sensor stiffness multiplied by the relative displacement. The problem investigated in this paper is that two typical holding tasks of the human finger are picked up and applied to the robotic finger. One is the work holding a stationary object with a prescribed, timevarying force and the other one is to keep the contacted force constant even if the object is in motion. The simple PID feedback control scheme is used to control the minute gripping force of order 0.01 Newton. It is shown both experimentally and theoretically that the artificial finger with the piezoelectric actuator works well in the minute force holding of the tiny object.

This paper presents the parameter design method for the desired time response of hydraulic track motor system of an industrial excavator. The dynamic response depends upon many component parameters such as motor displacement, spring constant and various valve coefficients. Most of them are to be determined to obtain the desired response while some parameters are fixed, or discrete for the offtheshelf type components. The parameters might be selected through repeated simulations of the system once the system is mathematically represented. This paper, however, presents optimization technique to select two parameters using a parameter optimization technique. The variational approach is applied to the system equations which are represented as state equations and from those system equations derived are the adjoint equations. The gradients for each parameter also are formed for the iterations.

In this paper, characteristics of the motions of a human arm are investigated experimentally. When the conditions of the target point are restricted, human adjusts its trajectory and velocity pattern of the arm to fit the conditions skillfully. The purpose of this work is to examine the characteristics of the trajectory, velocity pattern, and the size of the duration in the following cases. First, we examine the case of pointtopoint motion. The results are consistent with the minimum jerk theory. However, individual differences in the length of the duration can be observed in the experiment. Second, we examine the case which requires accuracy of positioning at the target point. It is found that the velocity pattern differs from the bell shaped pattern explained by the minimum jerk theory, and has its peak in the first half of the duration. When higher accuracy of the positioning is required, learning effects can be observed. Finally, to examine the case which requires constraint of the arm posture at the target point, we conduct experiments of a human trying to grasp a cup. It is considered that this motion consists of two steps : one is the positioning motion of the person in order to start the grasping motion, the other is the grasping motion of the human's hand approaching toward the cup and grasping it. In addition, two representative velocity patterns are observed : one is the similar velocity pattern explained in the above experiment, the other is the velocity pattern which has its relative maximum in the latter half of the duration.

In this paper, we present a full digital control scheme which controls currents and speed of the permanent magnet AC servo motor with large range of bandwidth and high performance. The current equations of the permanent magnet AC servo motor are linearized by feedback linearization technique. Both acceleration feedforward terms and IP controllers, whose gains are functions of motor speed, are used in order to control motor currents. In addition the phase delays in current control loops are compensated by placing phase leadlag compensators after current commands, which make it possible to avoid high gains in the current controllers. Unity power factor can be achieved by the proposed current controller. Pulsewidth modulation is performed by way of the wellknown comparison with a triangular carrier signals. The velocity controller is designed on the basis of the linearized model of the permanent magnet AC servo motor by the proposed current controller. The performance of the entire control system is analyzed in the presence of uncertainty in the motor parameters. The proposed control scheme is implemented using the digital signal processorbased controller composed of an Analog Device ADSP 2111 and a NEC78310. The pulsewidth modulation (PWM) signals are generated through a custom IC, SAMSUNGPWM1, which has the outputs of current controllers as input. The experimental results show that the permanent magnet AC servo motor can be always driven with high dynamic performance by the proposed full digital control scheme of motor speed and motor current.

The robot has spread remarkably, is used not only in manufacturing but also in various other fields, and is becoming more popular in everyday life. At the same time, the functional demands for all manner of robots have been diversified. Education regarding robots has been developing in the computer, mechanism, sensor and artificial intelligence fields. Technical education which integrates all of the above is necessary and in great demand. We have developed an educational robot so that it can be used in education in fields including structure, sensory and brain function and can also organically integrate those.

This paper described a relationship between motion speed and working accuracy of industrial articulated robot arms. Working accuracy of the robot arm deteriorates at high speed operation caused by a nonlinear transformation of the kinematics and the time delay of the robot arm dynamic. The deterioration of the following trajectory was expressed as a linear function of the squares of the robot arm motion speed, depending upon a posture of the robot arm and division interval of the objective trajectory.

Most of the control problem is for the redundant manipulators use the pseudoinverse control, thit is, the redundancy is resolved by the pseudoinverse of the Jacobian matrix and then the controller is designed based on this resolution. However, this pseudoinverse control has some problems when the redundant robot repeats the cyclic tasks. This is because the pseudoinverse resolution is a local solution that generates the different configurations of the robot arm for the same hand position. Therefore it is necessary to find the global solution that maintains the optimal configuration of the robot for the repetitive tasks. In this paper, we want to propose a redundancy resolution method by the optimal theory that uses the calculus of variation. The problem formulations are : first to convert the optimal resolution problem to an optimal control problem and then to resolve the redundancy using the necessary conditions of optimal control.

There are so many types of sensors which have been developed in order to construct intelligence robots. This paper presents the study of the movement of a vehicle robot using a CCDCamera. The CCDCamera is used as a sensor to control a vehicle robot in a stable movement. This vehicle robot is called CVR. The system is the combination of the CCDCamera, the vehicle robot and a dedicated software controller. The stability of CVR is proven by studying the movement methodology. The performance of the movement is experimented.

In this paper we propose two feedfroward unbalance compensation algorithms, they accommodate changes of rotor dynamics including rotating speed. The first one determine the compensating signals by identifying system dynamics successively. Whereas, the second one is more primitive like PID algorithm without identifying system dynamics.

In KACC'91 and '92 conference, we proposed a method of automatically detecting the shape of the faulty holes in a shadow mask by use of CCD ca.mera and image data processing technic. In this method, two adjoining test areas from one image data. of the shadow mask are taken and comparing the shape of holes in these two areas, we can detect the faults in the shadow mask. In this paper, a method is described by use of spatial filtering of effectively finding the faulty holes from the difference image data between the two tested image data. The main role of the filter is to remove sampling errors occurring at the edge of the holes. And the second role is not only to find the existence of faulty holes but also exactly express the shape of faulty holes. Computer simulations and actual experiments with shadow masks have shown that this method of fault detection is very effective for practical use.

A new method of spindle waveform detection is presented for the automated analysis of sleep EEG. The method is based on the combined application of signal conditioning in the timedomain and local spectrum analyzing in the frequencydomain. The overall detection system is implemented and, tested in realtime with a total of 24 hour data obtained from four subjects. The result shows an average agreement of 86.7% with the visually inspected result.

To reconstruct the real 3D shape from the 3D measurement data from the multiple directions, the rconstruction of the object on the basis of the mosaic processing of the 3D measurement data are proposed. In this method, to conduct the reconstruction, the connection points have to be identified among the overlap area between adjacent 3D data. In this study, the simple image matching method is adopted for the identification of connection points, and this method is verified from numerical experiments.

In an intelligent manmachine interface, it is very effective to support human thinking and to be in communication in some intuitive fashion. For this, sharing experience between the party concerned, human operators(s) and the interface is essential. It is also necessary to keep mutual understanding in some conceptual levels. Here in the present paper, figures which are an aspect of concepts and form a basis of mental image are discussed.

Here in the present paper. A methodology for understanding scenes which includes moving objects in it, in the framework of notion of concepts. First by conceptualizing, understanding an object which is an element of a scene will be described. Then how to know the direction to which that object is heading will be discussed. Further, the methodology proposed, for understanding conceptually the motion of an object will be described utilizing the above knowledge of direction.

In understanding concepts, there are two aspects; image and language. The point discussed in this paper is things fundamental in finding proper relations between objects in a scene to represent the meaning of the that whole scene properly through experiencing in image and language. It is assumed that one of the objects in a scene has letters as objects inside its contour. As the present system can deal with both figures and letters in a scene, the above assumption makes it easy for the system to infer the context of a scene. Several personal computers on the LAN network are used and they process items in parallel.

Nishida, Shigeto;Nakamura, Masatoshi;Miyazaki, Masahito;Suwazono, Shugo;Honda, Manabu;Nagamine, Takashi;Shibasaki, Hiroshi 283
We constructed morphological filter for single sweep records of eventrelated potential (ERP), especially P300 waveform. By combining 4 basic operations; erosion, dilation, opening and closing, we can derive any desired filters whose property fits the current objectives. The morphological filter for single sweep records of ERP was constructed by taking account of the features of the signal and noise components. The morphological filter has superior properties of separating the signal ancl the noise even existing within a same frequency band. The constructed morphological filter was tested by using simulation data of ERP and then applied to actual ERP data of a normal subject. The results proved that the constructed morphological filter was an appropriate tool for single sweep records of ERP. 
This paper summerizes a prototype central control system of distribution automation system, which is a result of project carried out by KERI and 6 companies. Of the many possible systems, feeder automation oriented system is described. Control hardware system is composed of one supermini and 2 workstations. Two kinds of commercial softwares, DBMS and graphic tools, are adopted. Three component system of the hardware has its own role, host system for DB managent, F.A. program running, communication scheduling, and etc., two workstations for communication node and graphic interface node. System management program, feeder automation program including load forcasting, communication scheduling and supervisory contorl functions are developed on the basis of above hardware and properly designed protocol, communication system and terminals.

In this paper, a new design method for Assembly Line Control System(ALCS) is presented. This system consists of five independent modules having their own specific functions such as production management, facility management material management, quality management, and remote control. To implement the ALCS, we propose design of the common data management module(CDMM). This module has the roles of integrating the above five modules and of communicating the common data between them. Using this method, we realize an information management method in the view of CIM. In addition, we standardize the intercommunication of common data between machines having different interface protocols.

In pneumatic systems the temperature changes during operation owing to air compression or expansion, friction of air movement and friction between solid interface. The temperature change usually ha undesirable influence on process. To attain higher quality of pneumatics, studies in thermofluid dynamics is needed. This paper presents experimental results anti theoretical analysis on the temperature change by air charge and discharge to cylinders, which has no piston yet. The temperature increase by charge shows a strong, dependence on axial location along the cylinder, which is proved in theoretical analysis. The temperature decreases by discharge shows rather uniform in the cylinder, which is also proved by theory.

In this study, a new model for flexible disk grinding process will be proposed. A grinding mechanism with a grinding disk attached to the rubber platen has been introduced. Since the spinning axis is fixed and only the disk is deflected with respect to this axis, earlier model is not adequate to represent this proces. A new dynamic process model includes an assumption that the disk is deflected locally around the middle of its radial span between the spinning axis and the disk tip instead of several continuous deflection points along the radial span of the disk. Detailed kinematic analysis is proposed as for the removed portion during the process. Cutting force comonent and depth of cut profile trend is compared with the measured result.

Research concerning gears included in rotating machines has been reported using the acoustic emission (AE) method, however, almost no research has been conducted using the AE method in regard to running gears in a bending fatigue processor spur gear teeth. Therefore, in this report, a power circulatingtype gear testing machine was used and AE signals and crack length were measured in the bending fatigue process of casehardened spur gear. Furthermore, the envelope of the AE signal was detected and various analysis were carried out in this data. In the course of the experiments, the following results were observed : the AE signal envelope consists mainly of contact frequency component and twice as many as this;two peaks of AE appear in each tooth contact by the tip corner contact ; as a result of the severe tip corner contact ; as a result of the severe tip corner contact with the sudden increase of crack length, AE signal becomes large.

Optimal control theories based on the maximum principles have been evolved and applied to distributed parameter systems(DPSs) represented by partial differential equations (PDEs) and integral equations (IEs). This paper intends to show that an optimal control of a tubular reactor described by a onedimensional partial differential equation was obtained using the distributed parameter control method for parabolic PDEs. In develping an algorithm which implements the calculation, the method of lines (MOL) was adopted through using a package called the DSS/2. For the tubular reactor system chosen for this paper, the optimal control method based on PDEs with the numerical MOL showed to be more efficient than the one based on IEs.

Electromagnetic suspension (E.M.S) type levitation system is studied in the control system design viewpoint. Dynamic characteristics in theoretical analysis as well as hardware implementation is considered. Open loop unstable, nonlinear and timevarying characteristics are reviewed in the theoretcal section, while levitation control system for multivehicle train as well as magnet drive system is reviewed in the practical section. This paper suggests not only some wellknown problem appearing in levitation control system design but also a subtle problem and solution candidates. But there exist many unmentioned problems wating for a smart problem solver.

A gain scheduling approach for the suspension control of a nonlinear MAGLEV System is presented. We show that this technique is ver useful for improving not only performance to the operational disturbances originating aerodynamic force but also robustness to the uncertainty of payload. As a scheduling variable, even though the external disturbance need to be estimated in real time, but the additive measurement is not required to do it. Some simulations show that the gain scheduling control system performs very well comparing with other method using a nonlinear feedback linearization or a fixed gain linear feedback.

One maglev vehicle is composed of 6 r 8 modules. Each module is composed of 4 staggered magnets attached to an aluminum bogie. In the view point of levitation control except propulsion by LIM, 5 is the maximum degrees of freedo to be controlled. But rolling control of the vehicle depends on the bogie structure. We describe just antiroll type out of bogie structures and 4 degrees of freedom control is sufficient for levitation quality improvement. Multivariable poleplacement concept and the decentralized control concept are used for controller design. Computer simulation and control experiment are performed on a specially designed test module.

A feedback linearization controller for EMS system is implemented using DSP. In this paper, we show that given EMS system is inputstate linearizable and satisfies some robustness condition. Also we derive feedback linearization controller for given system. Finally, some experiments are performed to demonstrate the performance of the proposed controllerespecially, comparing this with the classical state feedback controller using linear perturbation.

An H
$_{\infty}$ control system design for a magnetic levitation system is presented. In the control system design, we consider the influence of both disturbances and uncertainties in the model. The main disturbances stem from the position sensors.The uncertainties are divided into electromagnetic and mechanical ones: the former are due to the gain change in the current amplifier, the influence of leakage flux and modelling error in the magnetic circuit and the latter are due to the changes of the mass and the moments of inertia of the vehicle. Therefore, the designed controller is indispensable to guarantee the robustness of this system for both stability and performance. The controller design is based on the standard H$_{\infty}$ optimal control problem. As the novel features in this paper :(1) there are two poles on j.omega.axis in the control model;(2) an integrator is included in the controller so that equivalently there are three poles on j.omega.axis in the model. Finally, several experiments and simulations are carried out to verify the high performance and robustness of the designed control system.m. 
This paper presents an online feedback error learning control algorithm for a magnetic levitation system. It will be shown that even in the case of abrupt changes of the system parameters and disturbanes, the control performance is still very satisfactory.

In this paper, reheatfan engine is described as class of models constructed from nominal and uncertainty model for robust control. In this class of models, uncertainty model consists of structured and unstructured uncertainty, and each model is identified from nonlinear simulation using FFT and ML technique. Then, control requirements and augmented plant are specified. H
$_{\infty}$ controller satisfying the control requirements is designed by using constant scaling matrix. Finally, efficacy of the H$_{\infty}$ controller is showed by computer simulation.n. 
This paper deals with the problem of robust TDF(Two Degree of Freedom) H
$_{\infty}$ control design for a linear system with parameter uncertainty in the state space model. The uncertain system considered here is with the timeinvariant normbounded parameter uncertainty in the state matrix. A TDF H$_{\infty}$ control design is presented which robustly stabilizes the plant, guarantees the robust H$_{\infty}$ performance and improves the tracking performance for the closedloop system in the face of parameter uncertainty. It is shwon that a suitable stabilizing control law can be constructed in terms of a positive definite solution to a certain parameterdependent algebraic Riccati equation and a good tracking performance can be constructed in terms of suitable feedforward control law.aw. 
In this paper, we propose a new design procedure of the Robust Model Matching(RM) using the Normalized Left Coprime Factorization (NLCF) approach. The RMM aims at reducing the sensitivity of a given control system, but standard design procedures are not for robust stability. Therefore we try applying the robust stability condition based on NLCF to RMM procedure. We first formulate the RMM using the robust stability condition of NLCF approach, then we propose the new procedure of the RMM. The point is that the condition includes the measure of sensitivity of the RMM. In the proposed procedure, a cost function is determined through the condition and solved by H
$_{\infty}$ contro technique. Finally we show a design example and check the performance.. 
This paper gives a convenient parameterization for the clas of all stabilizing controllers in the presence of plant perturbation. The perturbations are constrained in such class as plants are stabilizable by a nominal controller. By using the controller stabilizing a given plant with perturbation, we can obtain a parametrization of all stable closedloop transfer functions, which are affine in the free parameter of the controller. It is easy to extend the controller to the case of a twodegreeoffreedom controller.

In this paper, we investigate a setmembership identification approach to the quantification of an upper bound of model uncertainty in frequency domain, which is required in the H
$_{\infty}$ robust control system design. First we formulate this problem as a setmembership identification of a nominal model error in the presence f unknown noise input with unknown bound, while the ordinary setmembership approaches assume that an upper bound of the uncertain input is known. For this purpose, the proposed algorithm includes the estimation of the bound of the uncertain input. thus the proposed method can obtain the hard bound of the model error in frequency domain as well as a parametric lowerorder nominal model. Finally numerical simulation results are shown to confirm the validity of the presented algorithm.. 
In this paper, an intelligent adaptive controller is proposed for the process with unmodelled dynamics. The intelligent adaptive controller consists of the numeric adaptive controller and the intelligent tuning part. The continuous scheme is used for the numeric adaptive controller to avoid the problems occurred in the discrete time schemes. The adaptive controller is adopted to the process with time delay. It is an implicit adaptive algorithm based on GMV using the emulator. The tuning part changes the design parameters in the control algorithm. It is a multilayer neural network trained by robustness analysis data. The proposed method can improve the robustness of the adaptive control system because the design parameters are tuned according to the operating points of the process. Through the simulation, robustnesses are shown for intelligent adaptive controller. Finally, the proposed algorithms are implemented on the electric furnace temperature control system. The effectiveness of the proposed algorithm is shown from experiments.

It is the purpose of this paper to present a dialogical tuning method of the sampling period in fuzzy control systems. Last year, the authors gave a dialogical tuning technique of fuzzy control system under the fixed sampling period in this symposium. In the case where sampling period is chosen larger, the response of the control system is unsatisfactory, and in the case where the sampling period is smaller, ineffective control actions are repeated. The appropriate sampling period is chosen through the step response of the closed loop fuzzy control process. As the tuning technique depends on the controlled plant, it is necessary to estimate the rough characteristics of it. The authors propose a method to decide th appropriate sampling period, by inspecting the characteristics of the plant.

A new approach using a Hopfield type neural network to solve line balancing problems for manufacturing planning is proposed. The energy function of the network to evaluate solutions is composed of three terms;(a) an operation should be processed at one and only one workstation, (b) the precedencerelationship between two operations shoud be satisfied, and (c) the cycletime of operations should be minimized. It is shown that the network can solve the line balancing problems but not always because of the difficulty to keep the precedencerelationship. Therefore, a method to keep the precedencerelationship by software logic is proposed and it is verified that the linebalancing prblems can be solved with high probability.

This paper presents an application of back propagation neural network to the tracking control of line of sight stabilization system. We design a neurocontrol system having two neural networks one for learning system dynamics and the other for control. We use a learning method which adjusts learning rate and momentem as a function of plant output error and error change.

This paper presents a neuro adaptive control method for nonlinear dynamical systems based on artificial neural network systems. The proposed neuro adaptive controller consists of 3 layers artificial neural network system and parallel PD controller. At the early stage in learning or identification process of the system characteristics the PD controller works mainly in order to compensate for the inadequacy of the learning process and then gradually the neuro contrller begins to work instead of the PD controller after the learning process has proceeded. From the simulation studies the neuro adaptive controller is seen to be robust and works effectively for nonlinear dynamical systems from a practical applicational points of view.

A variable structure control scheme that can be applied to the process with input/output delays are proposed and its control performances are evaluated. The proposed VSCS, which is an output fedback scheme, comprises an integrator for tracking the setpoint and the Smith predictor for compensating the effects of time delay. With The VSCS, the robustness against the parameter variations and external disturbances can be achieved even when the controlled process includes I/O delays. And the desired transient response is obtained by simple adjustment of the coefficients of the switching surface equation.

The transient responses of a linear system having undesired disturbances are dominantly governed by the system's left eigenstructure(eigenvalues/left eigenvectors). In control system design problem of altering the transient response of the system, both the controllability and the disturbance suppressibility, should be considered simultaneously to obtain a robust, effective controller. The controllability of the system may be degraded if the left eigenstructure is chosen to suppress the disturbance, or vice versa. In this paper, first, proposed are a modal disturbance suppressibility measure and an improved version of the modal controllability measure suggested by Hamdan and Nayfeh. Second, a simple and general left eigenstructure assignment scheme, considering both the proposed modal disturbance suppressibility measure and the improved version of modal controllability measure, is suggested. When the previous works are applied to assign the left eigenvectors may differ from the desired ones. But the proposed left eigenstructure assignment scheme makes it possible to achieve the desired colsedloop eigenvalues exactly, provided the desired left eigenvectors reside in the achievable subspace. In case the desired left eigenvectors do not reside in the achievable subspace, the closedloop eigenvalues are achieved exactly and the left eigenvectors are assigned to the best possible set of eigenvectors in the least square sense. Finally, a numerical example is included to confirm and demonstrate the usefulness of our propositions and to illustrate the proposed design scheme.

Hong, KeumShik 420
New conditions for the exponential stability for both linear nonautnomous finite and a class of infinite dimensional systems described by parabolic partial differential equations (PDE's) are derived. The results for the parabolic systems are derived via semigroup approach. 
Some upper bounds for the solution of the continuous algebraic Riccati equation are presented. These consist of bounds for summations of eigenvalues, products of eigenvalues, individual eigenvalues and the minimum eigenvalue of the solution matrix. Among these bounds, the first three are the first results for the upper bound of each case, while bounds for the minimum eigenvalue supplement the existing ones and require no side conditions for their validities.

First, we propose a transparent and efficient design of discretetime integral controllers accounting sampling skew. Based on the proposed controller, we derive a statespace representation of doubly coprime factorization including integral action. The representation is then used to obtain a convenient statespace parametrization of discretetime twodegreeoffreedom integral controllers acconting sampling skew.

Shinohara, shigenobu;Michiwaki, Motohiko;Ikeda, Hiroaki;Yoshida, Hirofumi;Sawaki, Toshiko;Sumi, Masao 434
Described is a stabilizing circuit of the Doppler beat signal obtained by the coherencedependent fiberoptic laser Doppler velocimeter (LDV), which employs both a selfmixing laser diode (SMLD) and a 10m100m long optical fiber. The stabilizing circuit maintains the SMLD drive current at an optimum value, which gives a maximum Doppler signal during long hours. 
Fujimoto, Ikumatsu 440
A two dimensional large scale laser digitizer with a cordless cursor was developed. The coordinate detecting scheme of this digitizer is fundamentally based on the triangulation method, in which two laserrays are scanned by the rotating plane mirros, reflected backward by the cursor, reflected again by the rotating mirrors, and detected by optical sensors. From angles in which the cursor reflections are detected, we can determine the position of the cursor. But this method involves several problems about optical alignment and its calibration especially when it is applied to a large scale digitizer. In this paper, especially we propose simulation for error analysis with connection to angles measured at five control points which are needed to decide an appropriate model for calculating coordinates and optimal simulation for deciding the position of five control points to give the better coordinate accuracy. In this way, we realized the onsite calibration and onsite insurance of measurement accuracy with our appropriate model for calculating coordinates. The time required for onsite calibration is within 5 minutes and the average accuracy of 4m * 3m digitizer is about .+.0.12mm. 
In this paper we propose an arc welding robot system, where two robots works coordinately and employ the vision sensor. In this system one robot arm holds a welding target as a positioning device, and the other robot moves the welding torch. The vision sensor consists of two laser slitray projectors and one CCD TV camera, and is mounted on the top of one robot. The vision sensor detects the 3dimensional shape of the groove on the target work which needs to be weld. And two robots are moved coordinately to trace the grooves with accuracy. In order to realize fast image processing, totally five sets of highspeed parallel processing units (Transputer) are employed. The teaching tasks of the coordinated motions are simplified considerably due to this vision sensor. Experimental results reveal the applicability of our system.

Interaction between system and disturbance results in system with timedependent parameter. Parameter variation due to interaction has random characteristics. Most of the randomly varying parameters in control problem is regarded as white noise random process which is not a realistic model. In real situation those random variation is colored noise random process. Modified FPK equation is proposed to get the response of the random parametric system using some correction factor. Proposed technique is employed to obtain the colored noise parametric system response and confirmed via MonteCarlo Simulation.

In this paper, a new guidance law for a shortrange airtoair missile with constant thrust is presented. It is essentially based on the concept of proportional navigation. First, the theoretical guidance law is derived. Then, we show the technique for practical implementation of the guidance law. By a computer simulation, it is shown that the new guidance law gives better performance than the conventional proportional navigation.

Even nowadays ships on the sea are important strategic base for aircraft and missiles. Thus we have been studying an unmanned attack system against ships recently. We experienced severa problems when this system was simulated on the computer. In this paper, problems and solutions of an AirToSea Guided Bomb for this system are presented.

Augmented proportional navigation requires the information of a target acceleration, which must be estimated by a filtering logic. The process necessary accompanies a time lag, which degrades the guidance performance. A tradeoff study between augmented and conventional proportional navigation is conducted with the time lag taken into consideration. The result shows that the conventional proportional navigation has better performance against a target maneuver in the missiletarget coplane, while the augmented has better performance against outofcoplane maneuver.

A study about twodimensional pursuitevasion dynamic games is presented and discussed. A pursuer tries to intercept an evader by a strategy based on proportional navigation guidance, while the evader tries to maximize a miss distance by the optimal control. The study is applied to a ball game and an aircombat game. The results show the same features exist in both games, therefore the study will be able to apply for general two dimensional dynamic games. In the ball game, the study is extended to cases where a goal exists, while in the aircombat game, some threedimensional problems are solved and the results are also shown.

A method for obtaining optimal orbital maneuvers of a space vehicle has been developed by combining feedback linearization method with the elegance of the Lambert's theorem. To obtain solutions to nonlinear orbital maneuver problems. The full nonlinear equations of motion for space vehicle in polar coordinate system are transformed exactly into a controllable linear set in Brunovsky canonical form by using feedback linearization by choosing position vector as fully observable output vector. These equations are used to pose a linear optimal tracking problem with a solutions to Lambert's problem and a linear analytical solution of continuous low thrust problem as reference trajectories.

In addtion to a basic motion task, redundant manipulators can achieve some additional tasks by optimizing proper performance criteria. Some of performance criteria can be transformed to inequality constraints. So the redundancy resolving problem can be reformulated as a local optimization problem with equality constraints for the end effector and inequality constraints for some performance criteria. In this article, we propose a method for solving the inverse kinematics of a manipulator with redundancy using the KuhnTucker theorem to incorporate inequality constraints. With proper choice of inequality constraints, the proposed method gives a way of optimizing multiple criteria in redundant manipulators.

In this paper, we present a methodology to model an indoor environment of a mobile robot using fuzzy numbers and to make a global map of the robot environment using graph theory. We describe any geometric primitive of robot environment as a parameter vector in parameter space and represent the illknown values of the prameterized geometric primitive by means of fuzzy numbers restricted to appropriate membership functions. Also we describe the spatial relations between geometric prinitives using graph theory for local maps. For making the global map of the mobile robot environment, the correspondence problem between local maps is solved using a fuzzy similarity measure and a Bipartite graph matching technique.

This paper presents an endpoint position control of 1link flexible robot arm by the PID selftuning fuzzy algorithm. The governing equation is derived by the extended Hamilton's principle and based on the BernoullieEuler beam theory. The governing equation is solved by applying the Laplace transform and the numerical inversion method. The arm is mounted on the translational mechanism driven by a ballscrew whose rotation is controlled by dcservomotor. Tip position is controlled by the PID selftuning fuzzy algorithm so that it follows a desired position. This paper shows the experimental and theoretical results of tip dispalcement, and also shows the good effects reducing the residual vibration of the endpoint.

An efficient solution algorithm of the optimal load distribution problem with joint torque constraints is presented. Multiple robot system where each robot is rigidly grasping a common object is considered. The optimality criteria used is the sum of weighted norm of the joint torque vectors. The maximum and minimum bounds of each joint torque in arbitrary form are considered as constraints, and the solution that reduces the internal force to zero is obtained. The optimal load distribution problem is formulated as a quadratic optimization problem in R, where I is the number of robots. The general solution can be obtained using any efficient numerial method for quadratic programming, and for dual robot case, the optimal solution is given in a simple analytical form.

In this paper we discuss the nonlinear motion of a conservative twolink arm using first integrals, which includes one integral constant. In the analysis of the motion, the constant plays important role. First, we give some discussion on the free motion by focusing on the integral constant. As the result, the free motion can be classified into two typesthe one is oscillation and the other is rotation. Second, we discuss the forced motion of the arm actuated only at the second joint. We take the first integral in a more general form, and show that the forced motion of the second link can be expressed as a variation of the integral constant. Also, the characteristic of the forced motion actuated by arbitrary constant torques is discussed.

Simultaneous position and vibration control of the flexible object while using dualarm manipulatorsIn this paper, we consider the handling f a flexible object using dualarm manipulators. We choose both the side arms as rigid, and the objects to be manipulated as flexible. Our purpose is to realize position control for the flexible object while suppressing its vibration. In particular, the problem taken up here is the stability of the control system while manipulating the object. We propose that the traditional approach to investigate the robot system be expanded to include the object's characteristics (thus transferring the stability of the robot system into the full assembly system). We define a handling characteristic while manipulating the object. Finally, the relationship between the handling characteristic and the positional constraint condition in the hold position of the arms is studied while considering the stability of the control system.

In this paper, the variable structure model following control scheme is proposed for the nonlinear robot manipulator system. The proposed control system guarantees that the system state is in the sliding mode for all time. Therefore, error transient can be prescribed in advance for all time. Furthermore, overall system is globally exponentially stable. Chattering problem is reduced by the introduction of a boundary layer. Simulation results are given to show the usefulness of the proposed control scheme.

Component Cost Analysis considers any given system driven by a white noise process as an interconnection of different components, and assigns a metric called "component cost" to each component. These component costs measure the contribution of each component to a predefined quadratic cost function. One possible use of component costs is for model reduction by deleting those components that have the smallest component cost. The theory of Component Cost Analysis is extended to include finitebandwidth colored noises. The results also apply when actuators have dynamics of their own. When the dynamics of this input are added to the plant, which is to be reduced by CCA, the algorithm for model reduction process will be called Weighted Component Cost Analysis (WCCA). Closedform analytical expressions of component costs for continuous time case, are also derived for a mechanical system described by its modal data. This is very useful to compute the modal costs of very high order systems beyond Lyapunov solvable dimension. A numerical example for NASA's MINIMAST system is presented.presented.

This paper describes a new method of dynamic error compensation, using a digital convolution integrator and two digital low pass filters. In this method, the process of compensation consists of three steps. First, sampling and digitizing of input signal, second, removing the noise in sampled data by the low pass filter and third, making a convolution integral using the output data of low pass filters. This method showed a good experimental result of reducing dynamic error even if there was a slight noise in the input signal. As a result, the detecting time constant of resistance thermobulb was improved to about 1/10th.

In this paper, we present an iterative learning method of compensating for position sensor error. The previously known compensation algrithms need a special perfect position sensor or a priori information about error sources, while ours does not. To our best knowledge, any iterative learning approach has not been taken for sensor error compensation. Furthermore, our iterative learning algorithm does not have the drawbacks of the existing iterative learning control theories. To be more specific, our algorithm learns a uncertain function inself rather than its special timetrajectory and does not request the derivatives of measurement signals. Moreover, it does not require the learning system to start with the same initial condition for all iterations. To illuminate the generality and practical use of our algorithm, we give the rigorous proof for its convergence and some experimental results.

This paper attempts to propose new procedures to evaluate roughness of ground metallic surface in the range of 110.mu.m from data gained by an optical, inprocess measurement of the surfaces. Studies are made to process the data of reflected lights pointed at the surface to be measured. Results obtained are compared with those of measurement by stylus roughness meter. Correlations between the two types of roughness measurement are well. The proposed method can be used as a sensor for a polishing robot.

an alignment system in the Xand Ydirections an XY.theta. stage driven by piezoelectric actuators is presented. A pair of quadruple gratings and a quadruple photodetector are used. The difference between the two 0th order moire signals in reflection with a relative spatial phase of 180.deg. is used in each direction to control the alignment of the XY.theta. stage. The stage is aligned at the position where the difference is zero. The quadruple gratings are 10 mm * 10 mm, and of a binary squaretype with a 1/2 duty cycle. Their pitches are 16 .mu.m. Alignment accuracy of .+.20nm was obtained in this system.

Methods of Alternating Noise Canceling were previously developed for the optical instrumentation;one using a dual photo sensor and another using a single photo sensor that could cancel normal mode noise on a transmission line, even if the noise was of equal status noise. But the methods could not remove noise on sensor line. This paper discusses a new method of using a photo sensor with a resistance sensordummy, effectively canceling equal status normal mode noise not only on a sensor line but also on a transmission line of an optical instrumentation. The accuracy of this method has been verified by experiments using sinusoidal wave as an equal status noise on a sensor line and/or rectangular wave as an equal status wide band noise on a transmission line respectively.

The error analysis for the TimeinFlight Laser Range Finder with Multiple Tone Amplitude Modulation relevant to the phase detection error is made. The distance can be estimated to solve the formulate which express the relationship between the absolute distance from the range finder to the object and the wavenumbers and the phases of the modulated waves by the optimization technique. The main cause of the estimation error can be considered as the phase detection error induced from the amplitude modulator and the phase detector. To clarify the phase detection error and the optimal amplitude frequency set, the numerical analysis are made.

The main difference between a linear system and a nonlinear system is the existence of direct interactions between input signals. These interactions will be classified into three types, (1) selfinteraction among different order terms of control signals, (2) static mutual interactions between the control signals, and (3) dynamic interactions through the coefficient venctor fields of the control variables. In this paper, we will show that interactions of type (2) and (3) can be avoided by applying an appropriate dynamic compensator, while the interaction of type (1) is fatal.

It is well known that H
$_{\infty}$ control problem involves solving an algebraic Riccati equation which includes a pair of parameters (.gamma., .epsilon.). Focusing on .epsilon. the maximum of .epsilon.. We discuss in this paper about the properties between the H$_{\infty}$ norm of a trnsfer function matrix and the parameters(.gamma., .epsilon.). We can change the algebraic relattion between .gamma. and .epsilon. by the similarity transformation of a considered system and we can find a proper transformation to get a simple quadratic algebraic equation between .gamma. and .epsilon.. This relation provide the H$_{\infty}$ norm of a transfer function.on.