제어로봇시스템학회:학술대회논문집
Institute of Control, Robotics and Systems (ICROS)
 기타
Domain
 Machinery ＞ Robot/Automated Machinery
1988.10

In this paper, a new procedure for selecting weighting matrices in linear discrete time quadratic optimal control problems (LQproblem) is proposed. In LQ problems, the quadratic weighting matrices are usually decided on trial and error in order to get a good response. But using the proposed method, the quadratic weights are decided in such a way that all poles of the closed loop system are located in a desired area for good responses as well as for stability and values of the quadratic cost functional are kept less then a specified value. The closed loop systems constructed by this method have merits of LQ problems as well as those of pole assignment problems. Taking into consideration that little is known about the relationship among the quadratic weights, the poles and the values of cost functional, this procedure is also interesting from the theoretical point of view.

For the trajectory control of dynamic systems with unidentified parameters a secondorder iterative learning control method is presented. In contrast to other known methods, the proposed learning control scheme can utilize more than one error history contained in the trajectories generated at prior iterations. A convergency proof is given and it is also shown that the convergence speed can be improved in compared to conventional methods. Examples are provided to show effectiveness of the algorithm, and, via simulation, it is demonstrated that the method yields a good performance even in the presence of distubances.

New design methods for constructing nonlinear adaptive control system are considered. The proposed adaptive controllers are applicable to the case where the degree of the controlled process is unknown. It is shown that the degree of the controller is determined independently of the degree of the process. Several types of nonlinear functions are introduced to deal with uncertainties of the degree of the process. Finally, some simulation results show the effectiveness and simplicity of the proposed methods.

A way of improving the transient performance is suggested for a class of model reference adaptive control systems. To increase the convergence rate of a model following error, an error feedback term is incorporated into the control law.

In this paper, the delay margins of the LQ and the LQG regulators are obtained in the time domain. These margins are represented in terms of the singular values of system matrices and the solutions of a Riccati equation and a Lyapunov one. And their asymptotical properties when gains tend to infinity are investigated.

In this paper, we propose a new method of designing a reducedorder controller for a linear discretetime system. Firstly, we study a design problem for a twodegreeoffreedom control system with a feedforward controller. Secondly, in order to obtain a reducedorder controller, frequencyweighted least squares approximation problems are considered. Thirdly, we propose a synthesis procedure of a reducedorder controller. Finally, an example is given to illustrate the effectiveness of this proposed method.

In this paper the failure possibility and the error possibility are used to represent reliability of a technical component and that of a human operator, respectively. The failure possibility and the error possibility are fuzzy sets on the interval [0,1]. In a manmachine system, reliability of the technical component and that of the human operator are usually affected by many factors, e.g., the environment in which a machine is operated, psychological stress of the human operator, etc. The possibility is derived from not only the failure or the error rate but also estimates of these factors. The fuzzy reasoning plays an important role in the derivation. The reliability analysis is performed by the use of the possibility obtained by the present method. Moreover this paper discusses the sensitivity analysis which evaluates what extent the change of the estimation of each factor has an influence on reliability of a manmachine system. The important factors to be ameliorated are shown through the sensitivity analysis.

Some properties of a matrix containing at least one pair of purely imaginary eigenvalues in the matrix sign function algorithm are explicated. It is shown that such a nonsingular matrix can be end up a singular matrix in the matrix sign function algorithm independently of the matrix condition. The result can be utilized to identify and locate all the eigenvalues theoretically.

In quantized control systems, the control values can take only given discrete (e.g. integer) values. In case of dealing with the control problem on the discretetime, finalstage fixed, quantized control systems with multidimensional performance functions, the first thing, new definition on noninferior solutions in these systems is necessary because of their discreteness in state variables, and the efficient search for those solutions at finalstage is unavoidable for seeking their discretetime optimal controls to these systems. In this paper, to the quantized control problem given by the formulation of Halkintyped linear control systems with two performance functions, a new definition on noninferior solutions of this system control problem and a heuristic effective search on these noninferior solutions are stated. By use of these concepts, two definitions on noninferior solutions and the algorithm consisted of 8 steps and attained by geometric approaches are given. And a numerical example using the present algorithm is shown.

A method for weighing fruits without separating them from stem is proposed. The base of stem is fixed and a fruit or a cluster of fruits is forced to vibrate. The approximated vibration model is constructed by the use of Transfer Matrix Method. The natural frequency (w) in this model can be represented as a function of weight elements, and the length and stiffness of branch elements of stem. With this function, only w is possible to measure. However, several small weights whose weights are known are attached to weight elements in various combinations. From these equations, unknown parameters are determined so that the weight of each fruit can be obtained by a nondestructive method.

A twodimensional positioning system by use of an Marray is proposed. An Marray pattern, which is known as one of the twodimensional pseudorandom array, is attached on an object to be positioned. The Marray pattern is observed by a TV camera and crosscorrelated with the reference Marray. The maximum of the twodimensional crosscorrelation function is sought by twodimensional servo system. This method of positioning can be used in very noisy circumstances.

Partially occluded objects are recognized from a 2D image through the use of maximum curvature points on the image boundary. The vertices of high curvature on an occluded object are classified by the objects which are hypothesized to be involved in the occlusion. A heuristic method is developed for computational speed. Two typical examples are given to illustrate the accuracy as well as the simplicity of the heuristic method.

It has become clear that through the use of force/torque sensing many previously uneconomical or unsuccessful applications can now be performed successfully. The earlieranticipated boom of robots dominating factory production has not yet occured. Many robot installations have not met the initial expectations of users or have failed completely, souring the prospects of future robotic applications. Yet the reason many of these earlier applications have failed is very basic, that is that the robot was expected to blindly perform a task perfectly in an imperfect environment. There must be an additional level of feedback so that the robot may adapt to these imperfections, thus ensuring the ultimate success of the application. This additional level of realworld sensing is provided by force/torque sensors and their continued use will ensure the eventual proliferation of factory robots.

G. Binnig and H. Rohrer introduced the Scanning Tunneling Microscope (STM) in 1982 and developed it into a powerful and not to be missed physical tool. Scanning tunneling Microscopy is a real space surface imaging method with the atomic or subatomic resolution in all three dimensions. The tip is scanned over the surface by two piezo translators mounted parallel (Xpiezo and Ypiezo) to the surface and perpendicular to each other. The voltage applied to the third piezo (Zpiezo) translator mounted perpendicular to the surface to maintain the tunneling current through the gap at a constant level reflects then the topography of the surface. The feed back control loop for the constant gap current is designed using the automatic control technique. In the designing process of the feed back loop, the identification of the gap dynamics is very complex and has difficulty. In this research, using some suitable test signals, the system dynamics of the gap including the Zpiezo are investigated. Especially, in this paper, a system model is proposed for the gap and Zpiezo series system. Indicial response is used to find out the model. The driving voltage of the Zpiezo and the tunneling current are considered as input and output signals respectively.

This paper is concerned with a computeraided supporting system to a systematic check of mechanical drawings. The problems treated in this paper are limited to the checking of the omissions and miswritings of dimensioning in the mechanical drawings made by a CAD system. A drawing check system has been made up on a personal computer with Basic and Pascal. The feasibility of the proposed drawing check system is confirmed for omissions and miswritings of dimensioning in mechanical drawings.

PRNM(Power Range Neutron Monitoring) of BWR (Boiling Water Reactor) is a system that processes signals from about two hundred LPRM (Local Power Range Monitor) sensors in the nuclear reactor and this system monitors the neutron flux level during the plant operating region. Development has been made by employing a special technique for multiplexing neutron sensor signals and the recent advanced microelectronics technology. It is applicable to the total plant digital control system for a nuclear power plant.

This paper presents a mobile robot that travels employing visual information. The mobile robot is equipped solely with a TV camera as a sensor, and views from the TV camera are transferred to a separately installed micro computer through an image acquisition device. An acquired image of a view is processed there and the information necessary for travel is yielded. Instructions based on the information are then sent from the micro computer to the mobile robot, which causes the mobile robot next action. Among several application programs that have already been developed for the mobile robot other than the entire control program, this paper focuses its attention on the travelling control of the mobile robot in a model environment with obstructions as well as an overview of the whole system. The behaviour the present mobile robot takes when it travels among obstructions was investigated by an experiment, and satisfactory results were obtained.

This paper describes a system which enables a fast 3dimensional measurement of a human face using a sliray projection method. One distinctive feature of our system is that a realtime video signal processor is employed in order to reduce the amount of image data to be processed and enable a fast measurement. Another feature of our system is that a skillful calibration software is developed. Due to this calibration software, opetators can be free from cumbersome settings of the measuring system.

In this paper, we present a system which enables a realtime measurement of the number and also the body length of the fry (baby fish) using a line image sensor. Here, we consider a situation that fry are transported from a pond to another, pond through a pipe. At one position of the pipe a transparent rectanglar channel is mounted. The images of the fry, which run through this rectanglar channel, are detected by a line image sensor. The image signals are digitized to binary ones and the contour of the fry are detected. After that, a realtime image analysis is executed with a digital signal processor. Labeling program analyses the connection of every pixel. The results are transfered to a personal computer and displayed on the online monitor graphically.

This paper presents an approach to segment an image into areas of surfaces, and to compute the surface properties from a grayscale image in order to describe the surfaces for reconstruction of the 3D shape of the objects. In general, an rigid body has several surfaces and many edges. But if it is not polyhedoron, it is necessary not only to describe the relation between surfaces, i.e. its line drawings but also to represent the surfaces' equations itself. In order to compute the surfaces' equation we use a probability of edge distribution. At first it is extracted edges from a graylevel image as much as possible. These are not only the points that maximize the change of an image intensuty but candidates which can be seemed to be edges. Next, other character of a surface (color, coordinates and image intensity) are extracted. In our study, we call the all feature of a surface as "texture", for example color, intensity level, orientation of an edge, shape of a surface and so on. These features of a surface on a pixel of an image plane are mapped to a point of the feature space, and segmented to each groups by cluster analysis on this space. These groups are considered to represent object surface in an image plane. Finally, the states of object surface in 3D space are computed from distributional probability of local and overall statistical features of a surface, and from shape of a surface.a surface.

"Motion Stereo" is quite useful for visual guidance of the robot, but most range finding algorithms of motion stereo have suffered from poor accuracy due to the quantization noise and measurement error. In this paper, 3D position estimation and refinement scheme is proposed, and its performance is discussed. The main concept of the approach is to consider the entire frame sequence at the same time rather than to consider the sequence as a pair of images. The experiments using real images have been performed under following conditions : handheld camera, static object. The result demonstrate that the proposed nonlinear leastsquare estimation scheme provides reliable and fairly accurate 3D position information for visionbased position control of robot. of robot.

We propose a new algorithm to obtain the output feedback controller for power system generators. The performance criterion of this controller is the integral of quadratic form of output differences between reference model and controlled system. With this criterion, we can easily compute the output feedback gains using Astrom's algorithm for the integral calculation of quadratic form. Simulations on a one machine infinite bus system shows the effectiveness of this approach.

A new control method for position control of D.C. servo motor based on the variable structure control is presented. The desired trajectory satisfying the given performance requirement is used as the sliding curve. And the control input forcing the system to follow the desired model system is applied. As a result the method is robust to disturbance. The performance of the proposed controller is compared with that of the conventional state feedback controller through digital computer simulation.

A simple procedure is proposed for the design of VSS controller to stabilize the single inverted pendulum. The controller is implemented by using a 16 bits personal computer.

In induction motor control, power efficiency as well as high dynamic performance is important. We attempt to achieve both of them by decoupled control of rotor speed and flux. A nonlinear feedback controller with a wellknown rotor flux observer is proposed with its stability analysis. Experimental results demonstrate that the proposed control method based on recently developed nonlinear feedback control theory is of practical use.

A microprocessor based control system for a brushless DC motor used in the motordriven electromechanical total artificial heart was developed. Functionally, the control system is composed of two parts. The first part is the velocity and position controller to assure that the motor follows a predetermined optimal velocity profile with minimal energy consumption, and to guarantee the full stroke length. This part also utilize the passive adaptive control method to be robust against the load disturbance, system parameter variation, and uncertainty which is the environment of artificial heart system. The pump output control is the second part, and this part provides the required responses of the artificial heart to the timevarying physiologic demands. The basic requirements of these responses are preload sensitivity, afterload insensitivity, and the balanced ventricular outputs. The performance and reliability of this control system was evaluated through a series of mock circulation tests and animal implantation, and the results are very encouraging.

During the 1980s, the use of highspeed printing and copying machines which will not tolerate much variation in paper weight has increased. As a result, there is a growing demand for improved basis weight control systems for paper machines. Basis weight, the weight of a square meter sheet of paper, is the most important measure of paper consistency. Until quite recently, basis weight could only be controlled manually; this required experienced operators. The reasons for the difficulty in automating control of basis weight is due to the complexity and variability of process characteristics, the "noisiness" of the process (process disturbances), and actuator inadequacies. This paper describes a method of automatic control that we have developed; it incorporates the experience of expert operators, and the results of simulation, in confidence factors.e factors.

The operational control of the rotary snowplow is considered to improve its working efficiency. The speed of the rotary snowplow is controlled, so that the load to the rotary snowplow is kept constant. As the load can not directly be detected, some items considered for the controlled variable are, for example, the engine revolution, the load pressure and etc. In order to examine these, the working simulation of the rotary snowplow was considered by introducing the experimental equation of the load. The control methods were examined by means of the simple digital control using the personal computer. These control methods were compared with simulations and experiments. Consequently, the working efficiency is improved about 20% than the manual operation.

In this work it is shown how the methods used in chemical engineering for the analysis and simulation of processes may be applied to the actual phosphoric acid plant. Attention has been focused on the dihydrate process for which the necessary fundamental experimental data and plant operation data are available. The results of the simulation have shown that a reasonable description of the process at hand is possible by the proposed method. However, because of the complexity of the process, of the limited basic experimental data reported in literature, and the limitations of mathematics, the model was somewhat idealized and gave a reliable representation of the influence of only a few of the variables that affect the performance of the plant.

For more economical operation of chemical plants, optimal operating conditions are to be set and maintained as far as possible. For this purpose, optimizing control is applied to chemical plants. In this study, a process optimizer composed of a process simulator and an optimization routine using Successive Quadratic Programming as optimization technique is developed and the effect of optimizing control is tested on an example process, and a new process optimization strategy based on modified Jacobian matrix is developed.

This is a successive study of the development and application of an electronic safeguard system for elderly men (senile dementia patients) who wander without purpose because of declined mental capability, while retaining their physical ability. The new safeguard system is designed with some additional functions on the basis of the previously developed system. Firstly, alarms are designed not to disturb other patients at night, so that informations about doors from which the patients go out may be transmitted to helpers individually by radio paging system. Secondly, the system hardware can be set up anywhere without laying particular signal transmission cables, provided that there exist AC power distribution lines for the utilization as a transmission line of signals to alarm indicators. Thirdly, it is possible to have a grasp of the whole states of the safeguard systems at the center of operation by monitoring the operational state of each system with a necessary data acquisition according to its instruction through telecommunication network. Thus, each safeguard system can be economically supplied to the special nursing homes and the helpers are ensured more released from physical and psychological burdens so that they can devote themselves to the care of senile elderly men, thereby improving their patients' comfort and human dignity.

We discuss what kind of scale is suitable for analog signal recording paper from the viewpoint of human engineering. We show a method for automatically generating a suitable scale from the higher and lower limits of the recording range.

Tree search can be applied to a control problem if the system is a discretetime one and if the control input takes only discrete values. This paper considers the application of heuristic tree search to a simple control problem. The results of simulation studies show the good possibility of this approach to a control problem.

In machining die cavities or mechanical parts, we often encounter the needs to remove a material within a given boundary. Even though this pocket cutting capability has been implemented in many NC packages, most of them can handle convex shaped pockets bounded by curves of limited types and numbers. In this work, a procedure has been developed to machine pockets of a free surface bounded by lines, circular arcs and free curves. Also, the cutter location data is computed directly without using iterative method for better computational efficiency.

The structure of a digital controller based on modern control theory is more complex than that of a PID controller. In implementing the digital control of an actual system by using the digital controller, we often encounter gaps between theory and practice e.g. quantization error, sampling error, modeling error, contaminated noise etc. In such cases, simulator plays an important role in detecting difficulties. This paper demonstrates the importance of the computer simulator for designing a digital controller. The controller and the simulator are constructed by different computer respectively, with a link between the blocks by analogue signals through the A/D, D/A converters. Through the simulator test, we can evaluate the digital controller; identify and solve difficulties in the digital control. The controller, which pasted the simulator test, is used identically in the actual system. This was a successful procedure for designing the controller. As an example, we successfully constructed the digital controller using the computer simulator for inverted pendulum control. We then compared the control results of simulator and actual equipment. Furthermore we commented on the construction of the computer simulator which exactly expressed the actual system.

To improve the safety and productivity of continuous processes, it is becoming increasingly necessary to have simulators to train operators. This paper describes a realtime simulator developed for this purpose by Yokogawa in cooperation with the Tokyo Gas Company. This simulator  based on the YEWCOM computer  not only trains operators, but also evaluates their proficiency.

The process and results of computer simulation using bond graphs for a hydraulic system driving large rotational inertia are presented in this paper. As the large rotational inertia and its application characteristics, control criteria of this system is not positioncontrol nor velocitycontrol but appropriate acceleration, deceleration and inching ability. All the components' nonlinear characteristics are modelled using bond graphs. The equationing and solution process is carried out by a package. Finally it is concluded that modelling of this kind of system by bond graphs and using a software as its solver shows good approximated results to actual experimental data, and that the proposed modelling may be useful to actual design process for this kind of hydraulic system.

Control technology has been making remarkable progress in electrical control (E), instrumentation & control (I) and computer control (C) field respectively. In order to rationalize system engineering and simplify system configuration and operation work, so called, EIC are now to be integrated into one system. FUJI has developed it's own E, I&C integrated system, to meet above mentioned market requirements of variety. This paper describes basic concept of FUJI's Integrated ManMachine Interface, hereinafter called SINGLE CONSOLE, from some view points of E, I, C integration technology.

In this paper we present a method of determining the unknown degree of any 2D discrete linear shiftinvariant system which is characterized only by the coefficients of the double power series of a transfer function, i.e. a 2D impulse response array. Our method is based on a 2D extension of BerlekampMassey algorithm for synthesis of linear feedback shift registers, and it gives a novel approach to identification and approximation of 2D linear systems, which can be distinguished in its simplicity and potential of applicability from the other 2D Levinsontype algorithms. Furthermore, we can solve problems of 2D Pade approximation and 2D system reduction on a reasonable assumption in the context of 2D linear systems theory.

In order to construct a nonlinear observer, change of coordinate system is necessary. However, as in the case of feedback linearizable system it is not easy to obtain a coordinate transformation map. ln this paper, a canonical structure is proposed for observable systems with an objective of finding a vector field which is necessary for the generation of a new coordinate system.

A simple method is proposed for determining the frequency response function G(j.omega.) of a system using a pair of characteristic Msequences (maximum length linear feed back shift register sequence). When a characteristic Msequence is sampled with q
$_{1}$ and q$_{2}$ both of which are coprime with N, where N is the period of the Msequence, the obtained pair of sequences have conjugate complex frequency spectrum. Making use of this fact, two charcteristic Msequences having conjugate complex frequency spectrum are applied to a system to be measured. Since the magnitude of spectrium of Msequence is known, the gain of G(j.omega.) is directly obtained from the Fourier transform of the system output. The phase of G(j.omega.) is obtained simply by taking the average of the two phases of output spectrum. 
A New type of controller for stabilization of ordinary system in which delayed states are included in feedback loop, is presented. Simple conditions are proposed for the stabilization of ordinary systems with the delayed state feedback controller. Under these conditions, controller gains can be chosen such that desired system performances are satisfied. It is shown that by using this controller the performance and robustness of the resulting closed loop system are much improved compared to the conventional memoryless state feedback controllers.

A collisionfree trajectory planning algorithm is proposed to optimally coordinate two robots working in a common 3D workspace. Each link of the two robots is modeled as a line segment and by their motion priority, one of the two robots is chosen as the master and the other the slave. And the onestepahead minimum distance between the two robots is computed by moving the master to the next location on its specified trajectory. Then the nominal trajectory of the slave is modified such that the distance between the next locations of the master and the slave must be larger than a prespecified allowable minimum distance. Here the weighted sum of the trajectory error and the joint motions of the slave is minimized by using the linear programming technique under the constraints that joint angle and velocity limits are not violated. To show the validity of the proposed algorithm, a numerical example is illustrated by employing a two dof's and a three dof's planar robots.

Teleoperator as a special robotic system for remote handling in hostile environment is reviewed in this article. Inherent features and prospective applications pertinent to teleoperator are indentified with particular concern for nuclear applications as major area of teleoperator technology exercise. Korean status as well as major world programs of teleoperator R&D are also reviewed with perspectives.

In mechanical systems in which the dynamics of armatures is dominated by electrostatic forces, motions will generally be unstable. This paper deals with the control problems of this kind of micro electrostatic device systems. In these systems, the mass of micro mechanical parts is so small that the inertia term in the equation of motion is negligible. However, nonlinear terms, such as friction and driving force, become dominant. The purpose of this paper is to realize the stable motion without delay and, overshoot etc. A micromechanical system used in this paper consists of a plane wafer with striped electrodes converted with an insulation layer and thin cylindrical roller is placed over on it. The performance of motions is confirmed by some simulations.

A conventional robotic manipulator is usually a very complicated system whose dynamics is too computationally time consuming for dynamic analysis and real time control. The authors have proposed the general design criteria of the robot links which greatly simplify the robot dynamic characteristics. In this paper these design guidelines are applied to a 6 degree of freedom PUMA 560 robot in order to realize actual implementation of the design concept. Based upon the design concept, the dynamic equations of the redesigned robot were derived. Dynamic characteristics of two systems, the ideally designed and conventional robot, are compared with respect to the joint input torque characteristics and degree of the coupling between the robot joints.

This twopart paper presents a control method that allows for stable interaction of a robot manipulator with the environment. In part 1, we focus on the input ouput relationships (unstructured modeling) of the robot and environment dynamics. This analysis leads to a general condition for stability of the robot and environment taken as a whole. This stability condition, for stable maneuver, prescribes a finite sensitivity for robot and environment where sensitivity of the robot(or the environment) is defined as a mapping forces into displacement. According to this stability condition, smaller sensitivity either in robot or in environment leads to narrower stability range. In the limit, when both systems have zero sensitivity, stability cannot be guaranteed. These models do not have any particular structure, yet they can model a wide variety of industrial and research robot manipulators and environment dynamic behavior. Although this approach of modeling may not lead to any design procedure, it will allow us to understand the fundamental issues in stability when a robot interacts with an environment.

A controller synthesis procedure for precise tracking of reference inputs in the sense of spheres is applied to a 3 d.o.f. robotic manipulator. This methodology applies to a class of nonlinear systems with input uncertainty and parameter uncertainty. The 3 d.o.f. manipulator to be controlled is subjected to varying payloads and is required to follow specified joint trajectories to within prespecified tolerances. The design procedure above lends itself naturally to this type of control problem. The appeal of such a design procedure lies on a special decomposition which exploits linear control theory on the one hand and facilitates a separate treatment of the effects of nonlinearities and the uncertainties on the other.

In recent years there has been much interest in using lightweight, higher performance arms for both commercial and spacebased applications, leading to the research of flexible robot manipulator. This paper is concerned with the trajectory control of a flexible arm using inverse dynamics. Inverse problems are important to robot control and programming, since they allow one to find the appropriate inputs necessary for producing the desired outputs. The input is obtained by the numerical inversion of Laplace transformation in the time domain. And we attempt the trajectory control experiment of a flexible arm using this calculated input. In this article we compare the numerical results with experimental results and can find good agreement. The results make clear that this technique has the good potential for the control of tip trajectory of flexible robot arms.

Adaptive control system has evolved as an attempt to avoid degradation of the dynamic performance of a control system when environmental variations occurs. While the feedback control system is oriented toward the elimination of the effect of state perturbations, the adaptive control system is oriented toward the elimination of the effect of structural perturbation, upon the performances of the control system. The model reference adaptive controller is utilized in velocity loop controller for positioning and tracking is designed based on the linear decoupled dynamics.

An iterative learning control scheme is presented with the aid of CMAC module. By enforcing the role of linear controller with the introduction of velocity feedback, it becomes possible to make the trajectory error equation stable. One advantage of this control scheme is that it does not require acceleration feedback. Computer simulation results shows a good performance of the scheme even in the case where the gravity is not compensated.

We have shown how unstructured modeling was used to derive a general stability condition in Part 1. In Part 2, we focus on the particular dynamics (structures modeling) of the robot manipulator and environment. Using rigid body dynamics, the stability condition for the direct drive robots has been achieved in terms of the Jacobian and robot tracking controller. Combining the structured and unstructured modeling, a stability condition for a particular application can be obtained. This approach has been used to analyze compliant motion on the University of Minnesota robot using a feedforward torque controller. We have obtained a stability condition for this application. Through both simulation and experiment, the sufficiency of this condition has been demonstrated. For a sufficient stability condition, recall that if the condition is satisfied, then the stability is guaranteed; however, if the condition is violated, no conclusion can be made.

Directdrive robots have excellent features including no backlash, small friction, and high mechanical stiffness. However, dynamic coupling among joints as well as nonlinear effects become more prominent than traditional robots with reducers. Another critical issue is that the robot becomes more sensitive to the change of load. In this paper, we develop a simple current feedback scheme for reducing the influence of dynamic coupling and load sensitivity on the directdrive robots. The method is implemented on a 2 d.o.f. planar directdrive robot. Then the validity of the method is demonstrated through experiments.

This study is concerned with modeling an elastomer constitutive relation by utilizing the truncated Volterra series. Actual experimental data from the Instron Tester are obtained for combined input, i.e. constant strain rate followed by a constant strain input. These data are then estimated for step inputs and utilized for the truncated Volterra series models. One second order and one third order truncated Volterra series models have been employed to estimated the forcedisplacement relation which is one of the prominent properities to characterize the viscoelastic material. The third order Volterra series model has better results, compared with those of the second order Volterra series model.

The author propose a method to realize a self organization in the artificial system. In self organizing system, subsystems are not constructed as functional parts of the system but cooperate with one another to realize the total system. Each subsystem obtains the local purpose from the global purpose by learning. This function is realized by using a neural network. The validity of this method is confirmed by some simulations.

The purpose of this paper is the development of a fully integrated simulation package for industrial robot. The simulation package consists of kinematics, dynamics, and control. The kinematics contains trajectory plans and inverse kinematics. The dynamics combines manipulator dynamics and actuator dynamics including the effect of payloads and viscous frictions. The control is a hardware oriented scheme which contains position controller, velocity controller, current controller, and PWM generator. Thus, the simulation package can be used not only for theoretical purposes but also for development purposes in industry. Using this package, the characteristics and performances of the SCARA robot, which has been developed in GSIS, are investigated.

This paper proposes a new method for generation of uniform random numbers using binary random sequences. These binary sequences are obtained from a de Bruijn sequence by random sampling method. Several statistical tests are carried out for the random numbers generated by the proposed method, and it is shown that the random numbers have good random properties.

In this study, we performed the recognition test on Korean vowel using peripheral auditory model. In addition, for the purpose of objective comparision, the recognition test is performed by extracting LPC cepstrum coefficients from the same data. And the same speech data are mixed with the Guaussian white noise quantitatively, then we repeated the same test, too. So we verified that this auditory model has a adaptability on noise.

This paper considers the design problem of adaptive filters based an the statespace models for linear discretetime stationary stochastic signal processes. The adaptive state estimator consists of both the predictor and the sequential prediction error estimator. The discrete Chandrasakhar filter developed by author is employed as the predictor and the nonlinear leastsquares estimator is used as the sequential prediction error estimator. Two models are presented for calculating the parameter sensitivity functions in the adaptive filter. One is the exact model called the linear innovations model and the other is the simplified model obtained by neglecting the sensitivities of the Chandrasekhar X and Y functions with respect to the unknown parameters in the exact model.

The problem of applying the linear integral filter in analysis and parameter estimation of linear continuous systems is discussed. A discretetime model, which is equivalent to that obtained using the bilinear z transformation, is derived and employed to predict system output. It is shown that the output error can be controlled through the sampling interval. In order to obtain unbiased estimates, an instrumental variable (IV) method is proposed, where the instrumental variables are constituted using adaptive filtering. Some problems on implementation of the recursive IV algorithm are discussed. Both theoretical analysis and simulation study are given to illustrate the proposed methods.

In this paper we present initial convergence properties of the Kalman filtering algorithm, we put an arbitrary small positive correlation matrix as an initial condition in the recursive algorithm. This arbitrary small initial condition perturbs the Kalman filtering algorithm and may lead to initial instability. We derive a condition which insures the stable operation of the Kalman filtering algorithm from the stochastic Lyapunov difference equation.

A systematic way of failure diagnosis in a linear system with decentralized estimators is developed. The generalized likelihood ratio (G.L.R) approach to failure detection and identification is used for designing a diagnosis system in each subsystem based on the innovation analysis. For the simplicity of the theoretical formulation, a design scheme of failure diagnosis is developed for the system decomposed into two subsystems. To demonstrate the effectiveness of our approach, several simulation studies have been carried out on a thirdorder linear system which is constructed of a secondorder damped oscillator and a firstorder lag.