Dye-sensitized solar cells (DSSC) with design as decoration can be fabricated by screen printing of TiO2 film. The power generation amount of the designed solar cell decreases by the decrease in area of TiO2 film, which contributes to the power generation. In addition, it depends on not only the area but also the pattern of TiO2 film. In this study, we investigated the influence of the pattern on the solar cell performance. As a result of measurement of solar cells with various pattern, it was found that the position and division of TiO2 film in the solar cell to suppress reduction in the power generation amount of the solar cell.
The risk of an illegal attack against the sensor devices used by the sensor network has been recently reported. Therefore, it is important that a security countermeasure by using an encryption technique. Then, lightweight block ciphers, which can realize low power, small area, and low latency, have been attracted attention as security countermeasures. Midori is a lightweight block cipher designed for the low power. Regarding the hardware security, the threat of fault analysis for a cryptographic circuit is pointed out. It is important that the tamper resistance verification of lightweight block ciphers against the fault analysis for the security of the future sensor devices. However, the fault analysis for lightweight block ciphers has barely been studied. Therefore, this study proposes a new fault analysis method for Midori. The proposed method performs the statistical analysis for the estimation of the key. Moreover, the proposed method performs two stages statistical fault analysis to analyze the all secret keys of Midori. Simulation results demonstrated the validity of the proposed method and the vulnerability of Midori against the proposed method.
This paper develops a wireless LAN base station with QoS control based on Web-QoE by bandwidth control and priority control; it evaluates performance of the developed system by experiments with subjects. By using the relationship between importance of data of a Web Service and the distribution of the length of IP packets, the base station can reduce the bandwidth consumption over a wireless LAN without worsening of QoE. The authors have implemented by adopting two technologies for priority control: OpenFlow and Wi-Fi Multimedia. In the experiment, the authors treat three actual Web services and assess QoE for them. From the experimental results, the authors show the effectiveness of the developed base station.
NTMobile (Network Traversal with Mobility) has been proposed to achieve end-to-end encryption communication supporting IP mobility in environments where IPv4/IPv6 networks coexist. However, since NTMobile unconditionally establishes an encrypted UDP tunnel between NTMobile-ready nodes (NTM nodes), a malicious NTM node can attack a target NTM node through the encrypted UDP tunnel without being detected by a firewall. Moreover, since communication with a general server always passes through a relay server, the route becomes redundant even when IP mobility is not needed, and the communication delay increases. In order to solve these problems, this paper proposes an access control function using the name of the correspondent node and a “Route option” which can select whether the relay server is used or not. As a result of implemention of the prototype system and evaluation of its performance, it was confirmed that the increase of the start-up time and that of the overhead at the beginning of the communication were quite small, and there was little influence on practical use.
A controller with a function of estimating unmeasurable inputs based on a minimal-order observer has been proposed previously. However, it seems very difficult to synthesize the disturbance attenuation controller based on the observer, because the synthesis method of the controller gain, which can simultaneously perform disturbance attenuation and internal stability of a closed-loop system, has not been investigated. In this paper, a novel synthesis method of the controller with a minimal-order observer, which can remove mismatched disturbances (i.e., disturbances do not meet the matching condition), is investigated. The basic concept is based on two considerations: 1) the system matrix of the measurement output estimator can be arbitrarily given and 2) synthesis of the compensator gain can be formulated as a synthesis problem of the high-gain feedback control system. Consequently, systematic synthesis of a controller that can simultaneously perform disturbance attenuation and provide internal stability can be achieved.
Reinforcement learning is generally performed in the Markov decision processes (MDP). However, there is a possibility that the agent can not correctly observe the environment due to the perception ability of the sensor. This is called partially observable Markov decision processes (POMDP). In a POMDP environment, an agent may observe the same information at more than one state. HQ-learning and Episode-based Profit Sharing (EPS) are well known methods for this problem. HQ-learning divides a POMDP environment into subtasks. EPS distributes same reward to state-action pairs in the episode when an agent achieves a goal. However, these methods have disadvantages in learning efficiency and localized solutions. In this paper, we propose a hybrid learning method combining PS and genetic algorithm. We also report the effectiveness of our method by some experiments with partially observable mazes.
This paper improves the linear displacement of an inchworm using piezoelectric actuators and electromagnets. The deformation of the piezoelectric actuators and excitation of the electromagnets are synchronized so that the inchworm can move. Since the inchworm does not use any guide nor preload, a closed loop feedback position control system is employed. In this paper, the displacement of the inchworm for one control cycle is measured, and the deformation of the piezoelectric actuators is adjusted in order to retain a designed trajectory. The position and angle of the inchworm is compensated by the use of appropriate feedback constants. This method is simple and is applied to camera vision position control system.
In this paper, we propose a new vibrato feature in order to measure temporal characteristics of vibrato in singing voice. The data to evaluate the vibrato feature includes non-imitative singing voice and imitative singing voice of target singer, whose voice and singing style have strong individuality. The data also includes the target singer's voice extracted from CD. The proposed vibrato feature has an advantage that the feature properly reflects the magnitude of fluctuations of fundamental frequency. The experimental results indicated that fluctuations of fundamental frequency of the imitative singing are larger than those of the non-imitative singing, and the magnitude of the proposed vibrato feature of the imitative singing is close to that of CD. We also performed objective evaluation by using the database that measures a distance between features calculated from imitative voice or non-imitative voice and CD's voice. The evaluation showed that the proposed vibrato feature correctly indicates smaller distance between the imitative voice and the CD's voice.
We have proposed a synthesis method of the linear quadratic regulator (LQR) with the desired low-pass property achieved by the second-order lag elements. However, a synthesis method of the LQR with the property attained by the higher order lag elements has never been proposed. In this letter, the LQR with the desired low-pass property achieved by the third-order lag elements is studied. The basic design idea is based on the fact that feedback control law can be reduced to the problem which solves a quartic algebraic equation.
In order to work effectively, a robot must be able to adapt to different environments by deciding the correct course of action in a given situation, using determinants other than pre-registered commands. That said, when considering an action decision based on a future prediction, it is necessary to determine the properties of a disturbance signal from the outside environment. Furthermore, determining the properties of a disturbance signal is dependent on the specifications of the target processor, in particular, its sensor resolution or processing ability. As humans, we often make a guess that face up to situations. In this case, we will explore rough adaptation to given situations. To apply this method to a robot, the ability to predict the future state of the robot is necessary. In this paper, we focus on control of a robot in an attempt to produce rough adaptation similar to that observed in humans. In particular, we consider the situation where a periodic disturbance signal occurs.
Humans tend to change their lexical expressions to resemble those used by their interlocutors to achieve smooth conversations. Such phenomena, called “lexical alignments,” are affected by gender. Even though lexical alignment is observed not only in human-human interaction but also in human-robot interaction, the gender effects on it in human-robot interaction haven't been investigated yet. Identifying whether gender affects lexical alignment in human-robot interaction would contribute to the design of conversational strategies for interactive robots for more natural interaction. This paper reveals that gender affects lexical alignment in human-robot interaction. We conducted an experiment with twenty participants who interacted with a robot in object reference conversations and referred to an object whose identity was confirmed by a robot. We developed a robotic system that engaged in object reference conversations with two interaction strategies and measured the gender effects on lexical alignment in human-robot interaction. Our experimental results showed that female participants were lexically more aligned with the robot than males; female participants used more references that were useful to uniquely identify objects in environments than males.
In this paper, a synthesis of a complex filter with reduced transmission zeros is proposed. A new characteristic function is proposed. It is described that the element values are obtained from the proposed characteristic function. The proposed prototype complex filter has finite transmission zeros at negative frequency area only. The required elements of the proposed prototype complex filters are fewer than that of the conventional ones with finite transmission zeros. Algorithms for determining the element values is described. The resulting circuit can be simulated without bilinear complex coefficient transfer functions. As an example, a third-order filter with equiripple passband response is designed. The image rejection ratio, sensitivity and the number of required elements are compared with those of the complex filters based on the frequency shifting method. The proposed circuit has the best image rejection ratio in the complex filters based on the frequency shifting method. The validity of the proposed method is confirmed through both of computer simulation and experiment. The experimental circuit exhibits complex bandpass characteristics(1-5kHz) and an image rejection ratio of 40.3 dB.
This paper describes a novel method of probabilistic self-modeling based on learning of operating space from exploratory actions of a multi-DOF manipulator, which is designed for mounting it onto the wheelchair. The developed anthropomorphic multi-DOF manipulator is able to learn both of the operating range in each joint and the probabilistic operating space based on Gaussian Mixture Model and Variational Bayesian learning algorithm. We introduce an acquisition method of the operating space by using the historical data of irregular overload, which is detected by using analogue current signals measured by solely internal sensor of joint motors. In addition, online behavior learning with a simple probabilistic path planning is also presented based on the obtained probabilistic operating space. We will conduct several experiments with a real multi-DOF manipulator arm. After the basic characteristics of the obtained operating space are shown, the performance of interaction with different situations such as different load given to the arm and obstacles placed in the surrounding environment will also be demonstrated.
In the semiconductor device manufacturing there has been a pressing need for inspection and measurement by using high-resolution images of scanning electron microscopes (SEMs). The SEM spatial resolution is mainly determined by the beam spot size at the specimen's surface and the spreading of the beam electrons in the specimen. Image restoration sharpens an observed image by the process of removing blur, which is represented by a point spread function (PSF), from the image. In this paper we address the problem of identifying the parameter of the PSF, the so-called blind deconvolution problem. The proposed method estimates the parameter from both the Fourier spectrum of the observed image and the diameter of the electron beam calculated by electro-optics simulator. Extensive experiments on real image data indicate significant improvements in image quality compared with conventional methods. The estimation error of the PSF parameter by the proposed method was about half that of the conventional one.
The reinforcement learning is a method of training for an agent for accomplishing task by selecting suitable action from the current state. Deep Q network is combining convolutional network with Q-learning. By using the Convolutional Neural Network, Deep Q Network can apply to large dimentional input state tasks without special pre-processing. However Deep Q Network needs a large iteration for getting excellent outputs. The reason of that the Deep Q Network is using ε-greedy for action selection, and the ε is set to high value (close to one) in initial stage in learning. High ε value means that the agent selects action randomly in the learning. Hence, the agent needs large number of iteration of learning for accomplishing a task. In this paper adopts the Boltzmann selection to Deep Q Network. Finally, our algorithm has been applied to 2 kinds of arcade learning environment tasks, and results showed that our algorithm is better than ordinary Deep Q Network.