A new electroabsorption modulator operating at a parallel electric-field to the quantum wells is proposed and its modulation characteristics have been calculated. Transverse p-i-n structure on semi-insulating substrate enables us to apply a parallel field for MQW layers and results in low device capacitance and high-speed operation. The exciton resonance can easily broaden and disappear at field-intensity of 104 V/cm, one tenth of the perpendicular field previously used. A 3-dB bandwidth and operation voltage required for 20 dB extinction ratio are estimated to be over 250 GHz and less than 1 V, respectively, with transmission loss of less than 3 dB. High allowability of high incidental optical power and negative chirp operation are also discussed for this structure.
The main purpose of the present studies is to analyze the transmission characteristics using graded index waveguides without image distortion. For the studies on transmission characteristics, we considered a model with Ti-diffused LiNbO3 waveguides. Through theoretical calculations in the space domain, we derived the point spread function (PSF) when a spatial impulse is used as input. For the electric field distribution inside the waveguide, we assumed Hermite-Gaussian functions and derived the first order perturbation solution, and performed the theoretical calculations. We evaluated the propagation characteristics of an optimized GRIN type optical waveguide with respect to propagation distance. The periodic focusing points in the propagation direction are confirmed. In order to remove the noise component during the transmission, we consider spatial filters. The image transmission and processing using spatial filters will be discussed.
The electrostatic discharge (ESD) between charged metal objects produces electromagnetic (EM) fields having broad-band frequency spectra over a microwave region, which cause serious EM interference (EMI) to high-tech information equipment. For this kind of ESD fields, focussing on EM fields due to the spark discharge between metal spheres in free space, we previously demonstrated that the presence of metal spheres enhances the field level according to their sizes. For the EMI due to the above ESD, however, its mechanism is not still elucidated. This may be because even the fundamental propagation characteristics of ESD fields above a ground plane or inside a room remains unknown in conjunction with the EM coupling to equipment. In this paper, we analyzed the electric field due to an ESD event occurring in a corridor using the current dipole inside a rectangular waveguide. The distance attenuation of the field peak for the ESD occuring horizontally to the floor was examined numerically. As a result, we found that the far-field peak dopes not always decrease with increasing the distance and that there exists a region in which the field peak is enhanced with the distance. This finding was also confirmed by the spark experiment of metal spheres inside a corridor.
The ESD (electrostatic discharge) between charged metals generates transient electromagnetic fields having broadband frequency spectra, which often gives fatal EMI (electromagnetic interference) to high-tech infor-mation equipment. For the EMI of this kind, propagation characteristics of ESD fields in conjunction with electromagnetic coupling between the ESD fields and equipment still remain unknown. From this perspective, we previously proposed an image-dipole method and FDTD methods to analyze the ESD fields due to charged metals. The former method, however, is limited for calculation to the sphere sizes that can keep the equi-potential on the metal surface, and the tatters are unable to analyze the propagation characteristics of the ESD fields over the broad region because of the limited analysis space. In this paper, the feasibility of the above image dipole method was investigated together with the FDTD method by comparing measurement and calculation of the magnetic near-field waveform due to the spark discharge between metal spheres with diameters of 3cm to 10cm. The magnetic far-field peak and its distance attenuation were numerically examined for the spark discharge between the metal spheres.
A compact and simple distance meter based on self-coupling effect of semiconductor laser has been devel-oped. Using the interference in the laser resonator with the light scattered at the object surface, the distance sensor is composed of only a laser diode and a lens. But the operating wavelength of semiconductor laser varies with the surrounding temperature. Then the method of self-correction using Michelson interferom-eter is studied. This sensor is capable of measuring a distance stably even if the temperature varies. The measurement range is from 20cm to 120cm, and the measurement error is below 0.6 percent.
This paper deals with the simulation of big problems using FDTD method. FDTD consumes a lot of memory and computing time. Most research focuses on processing time, but memory is the first obstacle in simulating big problems. The common FDTD equations are reformulated into a big system of linear equa-tions. A new recursive algorithm is proposed for dealing with this problem, and the parallel version of the algorithm is also introduced. Some numerical comparasions are done, and conclusions about the advantages of the algorithm and the required improvements to make it practical.
This paper presents a computer system that automatically generates camera works for a scene with moving objects. Most realistic and attractive shots follow the common sense of camera control. Thus we implement each rule of the camera control as a virtual director. A virtual director proposes several shots for a scene based on his rule from three-dimensional shapes, locations, and properties of objects in the scene. He adds his idea to the shots by other directors if possible. The directors evaluate all shots, and then shots with the high score are selected. In this processes, shots for satisfying many rules are created. By jointing the shots in the order with the high score, our system creates several series of shots that represent the whole scene. The series are evaluated again based on the rules of camera control, and finally most valuable one is selected. The experimental results confirmed that the system can generate camera control sequences which are comparable to that made by human.
Spray Modeling System is a real-time interactive 3D modeling system that shaves and swells objects by spraying virtual particles onto their surfaces. Users can make holes and fill them with the same spraying operation. This system achieves the simple but realistic modeling environment by following the real method called Sand Blast. More sensuous operations are available with a data tablet. So many people, especially beginners of 3D modeling, can use the system easily. Object surfaces are described with triangles. The triangles are subdivided or reduced automatically. The hole is displayed with the stencil buffer during a continuous operation and surface topology is changed just af-ter the operation. Front-list and Polygon-list, which are introduced in the system, subdivide the object space in order to accelerate shaving surfaces and making holes. These techniques achieve real-time 3D modeling on standard PCs. The performance of the system is confirmed with computer experiments.
This paper proposes a new method for the pitch estimation of sung songs for transcription. The previous pitch estimation methods that are considered most for musical instrument sounds but few for sung songs are based on the extraction of the pitch frequencies. But the extraction of the pitches is rather difficult and sophisticated signal processing is necessary since the musical sound has many harmonic components. The principle of our method is the elimination of the pitch and its harmonic frequencies. This can be performed simply by a comb filter. To adapt the fluctuation of song pitches, we use double comb filters. Using two cascade connected twelve double comb filters that are connected in parllel and are corresponding to each tone in one octave, we can estimate the pitches of solo and duet songs.
There are a number of approaches which apply the DCT (Discrete Cosine Transform) on image coding, however higher compression rate causes blocky noise by quantizing the low band and causes mosquito noise by quantizing the high band when the image is being reconstructed. To overcome the defects, we propose the DCT - Wavelet Transform for improving the quality on reconstruction of the compressed image. By implementing the proposed method, the examples demonstrated that a better quality is obtained for image reconstruction.
We propose a new method for extracting separately each of the sounds from the mixture of two speech sounds, which are uttered concurrently. First the mixture is transformed into a sound spectrogram which is thereafter treated as an image. Exploiting image processing techniques, the onsets and offsets of the fre-quency components of each speech sound are detected. Then the harmonic structure of each speech sound is extracted by tracing each onset through the corresponding offset and relating each of them to one another in the frequency domain. A set of band-pass filters are designed reflecting the extracted harmonic structure. Each speech sound is extracted by applying the set of band-pass filters to the mixture. Experiments were conducted with the mixture of a male speech sound and a female speech sound both consisting of Japanese vowels. The evaluation results demonstrated that the separation was done reasonably well with the proposed method.
Recently, we are concerned with a photovoltaic (PV) power generation system because of environmental problems. The output power of the photovoltaic generation system is affected by weather conditions. Espe-cially, clouds have great effects on fluctuations of the output power. This paper describes a new method to distinguish two types of clouds, thick clouds and thin clouds, for predicting quantity of drops of PV output power due to clouds. The method uses the image processing technique. First, cloud images are extracted from the sky images according to the color of images. Then, the changes of the gray level in the cloud images are examined. A large change of the gray level means a thick cloud and a little change does a thin cloud. The correlation between classification results and quantity of output power drops has also been investigated. Thick clouds and thin clouds decrease the output power by more than 70% and about 50% respectively. The experimental results show that it is possible to predict rough amount of the PV output power drops by the present method.
This paper presents a control method of flexible master-slave manipulators (FMSM) with a rigid master arm and a flexible slave arm. The proposed control system is constructed on the basis of the conventional force reflecting servo type bilateral control in consideration of elasticity of the slave arm. The disturbance observer is used to compensate for friction torque of motors. In order to reduce the effect of elastic vibrations on the control performance, we design the tip angle control and the vibration control on the basis of the concept of the decomposition of elastic vibrations in the strain gage signal. When applying this concept to FMSM, the input shaping technique is used to eliminate the frequency component exciting elastic vibration modes from the master angle signal that is a reference input to the slave system. Some experimental results show the effectiveness of the proposed controller.
In recent years, an electric power system becomes more and more complication and large-scale. Therefore, transient stability problems and voltage ones of power system are taken seriously. In this paper, the switch-ing technique of the serial and parallel resistors for power system stabilization is proposed and the following were executed; Computer simulation and experiment using transmission simulator. The effectiveness for the stabilization was shown by that a system swing was damped by the on-off control of the series and parallel resistors.
In real control systems, there are some constraints on the value of inputs, states and outputs. If such constraints are violated, systems may become unstable in the worst case. One of the promising approaches to solve this problem is to switch the controller according to the state of the closed loop system. No useful design strategy for each controller, however, has been developed yet. In addition, the decision of controller state at the instance of controller switching has not been investigated. First of all, this paper presents a new design methodology which merges the switching control and H∞ controller design into one design algorithm. Secondly, this paper presents one of the decision method for controller states at the instance of controller switching. Finally, the proposed method is applied to a force control of one DOF manipulator to demonstrate the usefulness of the proposed method.
In this paper, we propose a computer aided writing in a markup method. The user doesn't need to pay much attention to sentence connectivity when he/she newly inserts a sentence or changes the order of sentences, because the system generates appropriate connective expressions. At first, we present our expression of text structure. Secondly, we propose a markup language called TSML (Text Structure Markup Language), which designates text structure. Thirdly, we discuss about a method inserting appropriate connective expressions. We implemented the TSML Editor on Windows NT machine by Visual C++ To evaluate the implemented system, we asked each of eleven subjects to write a document on this system and to fill out a questionnaire which consists of twelve items. Finally, we discuss the advantages and disadvantages of the system based on the result of these questionnaires.
Peculiar characteristics of a Fabry-Perot resonator with a dielectric plate inserted between the mirrors are reported. A simple theoretical consideration shows that, when the two mirrors are moved with the resonance condition maintained, the magnitudes of the two displacements are generally not equal; if the frequency of the electromagnetic wave is changed so as to recover the resonance condition that is broken by the change in the mirror separation, the required frequency shift generally depends on the position of the dielectric plate.
This paper presents a design of a continuous-time CMOS low-pass filter using floating-gate MOSFET (FG-MOSFET) integrators. In this design, FG-MOSFETs were used to realize fully-differential current-mode integrators which can operate at a low supply voltage of 1.5V. The analysis has been carried out on the filter performances as well as on integrator's bias conditions, operating frequency range, gain, harmonic distor-tion, and Q-enliancement technique. HSPICE simulations showed that the proposed method can realize a low-voltage/low-power continuous-time fully-differential current-mode CMOS filter usable in the frequency range up to several tens of megahertz. The proposed filter is suitable to low-voltage mixed-signal VLSIs, because it can be fabricated by a standard double-poly CMOS process.
As the development of the fine fabrication technology increases a gate density, the wiring delay by the number of the entry pins and the inner wiring is becoming a serious problem. On the other hand, because it is possible to make a signal correspond to the many levels in the multiple valued logic system, the amount of information per line increases. Then, the inner wiring area and the number of the transistors can be decreased. Especially, the electric current mode has been often used in the multiple valued logic system, because it is possible to utilize the advantage that the addition and subtraction calculation can be executed by using only the wiring. However, in the electric current mode, the consumption increase of the electric power can not be ignored. Because the neuron MOSFET is manufactured in the 2 layer poly silicon CMOS process, a special manufacturing process isn't needed. When using a neuron MOSFET at the binary logic circuit, it is possible that the number of the transistors and the number of the wiring are substantially reduced. In the floating gate part of the neuron MOSFET, the electric charges are piled up and the electric charges are balanced. By using neuron MOSFET source follower for the output part in adopting this function, the multiple valued logic circuits can be composed in the voltage mode. In this paper, by using a neuron MOSFET as the element of the multiple valued logic circuits, it is designing the layout of the quaternary full adder, which is due to the voltage mode, with the signed digit number representation. However, as for the layout figure of this full adder, the rate with the area of the floating gate part accounts for becomes big. And, the layout area gets widely. Therefore, by attempting to share a floating gate, it is reducing layout area. It is shown in the experiment that the occupation percentage of the floating gate is reduced to about 61%, and also to about 41% as the whole layout area, after the sharing of the floating gate.
In a dynamic environment, maps become outmoded unless they are updated from time to time. While map making has received considerable attention in the literature of robotics research, less attention has been paid to map updating. This paper will present an algorithm for map updating by autonomous mobile robots based on prediction of changes in the environment. Map updating problem can be characterized as a sequential decision problem. At each step, the robot searches the environment by using range sensor, decides next observation location and goes to the location. It is not practical to plan an entire observation route. since the robot will encounter with so many changes in the work space. The method predicts the work space and segments it into some subregions so that each subregion will contain less changes. Then, the solution is obtained from subproblems of finding shorter routes in subregions. To search candidates of next robot action efficiently, the method employs path graph. Consequently, suitable solutions are found very quickly.
Articulated robot arms are employed in various applications such as welding, cutting, grinding in industry. With advancement in industrial robotics, high-speed and precise performance are required for the operation. Some of the requirements are not easy to realize by use of one robot, due to the limits in speed and operating range. Then, it is suggested to employ two robot arms with co-operative control. In this research, co-operative control by use of synchronous positioning control technique is proposed to realize co-operation of robot arms. Precise co-operative control of performer MK2 and performer MK3s robot arms (miniature type of industrial robot arms) are realized by use of the proposed method.
This paper presents an overlapped niulti-neural-network (OMNN). An ODINN consists of two parts: main part and partitioning part. The main part. structurally, is the same as an ordinary feedforward neural net-work, but it is considered as one consisting of several subsets. All subnets have the same input-output units. but some different hidden units. The partitioning part divides input space into several parts. each of which is associated with one subnet. An improved random search algorithm called RasID is introduced to train the OMNN. Numerical simulations show that such an OMNN has superior performance in that it has better presentation ability than an ordinary neural network and better generalization ability than a non-overlapped multi-neural-network.
In this paper, the I/O (Input-Output) Analysis method of LCA (Life Cycle Assessment) is compared with the Process Analysis method, in order to analyze differences between their performances such as CO2 emission intensities for iron steel products, plastics and energy. The new I/O Analysis method, where environmental loads of imported goods burdened in their export countries are added to environmental loads derived by I/O Analysis of the own country, is proposed. This I/O Analysis method derives values almost same or slightly different from values given by the Process Analysis method, for most materials on CO2, emissions on every life cycle stages. There are some exceptionally large differences between the CO2 emissions derived for some materials such as galvanized steel sheets by the two methods. The reasons, why the large and/or small differences would be occurred by the two methods, are also discussed.