In this study, we numerically investigate the oscillatory behavior of the cantilever probe in a dynamic AFM, which is approximated by a one-dimensional continuous model. In particular, we investigate the amplitude characteristics of the fundamental oscillation mode and two higher oscillation modes.

View full abstract

An atomic force microscope (AFM) is a microscope that measures the information of a sample surface when the probe tip approaches the surface. In our previous work, we investigated a mass-spring model of the driven probe tip influenced by a homogeneous sample. In this study, we investigate the influence of a heterogeneous sample surface on the amplitude characteristic of the tip oscillation by comparing the calculated amplitude and the theoretically obtained amplitude using a bifurcation theory.

View full abstract

A selective desensitization neural network (SDNN) has a high function-approximation ability, low hyperparameter dependence, and suitability for online incremental learning. These properties suggest that an SDNN can deal well with temporal changes in the characteristics of data, or concept drift, although this has not been verified. In this study, we conducted experiments on online learning using an artificial dataset generated using a time-varying function and a real-world dataset of a stock prices index, and evaluated the effectiveness of an SDNN to solve concept drift problems. The results show that using the SDNN exhibited superior performance over the existing methods for both datasets, suggesting that an SDNN is highly suitable for certain types of concept drift problems.

View full abstract

The most successful generative tasks, such as image completion, generally rely upon generative adversarial networks (GANs). The hardware implementation of GANs has important requirements of low power and acceleration, but differ from a usual neural network by requiring a training phase. We developed a hardware-oriented training algorithm using a quantized stochastic gradient descent method without the use of a hardware-oriented training algorithm. From this result, we devised a GANs architecture requiring 7 bits for inference and 26 bits for the training phase, when using a resized MNIST dataset with a three layer perceptron for each network. We can achieve real-time processing when the architecture functions ideally; however, the bit width and process speed are affected by the network model.

View full abstract

A reported technique for cochlear delay (CD) based audio information hiding achieved imperceptibility in non-blind approaches. However, the phase shift keying (PSK) technique was utilized with a blind method, causing drastically phase changes that imply perceptible information was inserted. This paper presents an investigation on the feasibility of hiding information in the linear time variant (LTV) system. We adapted a CD filter design for the LTV system in the embedding scheme. The detection scheme was conducted using instantaneous chirp-z transformation (CZT). Objective tests for checking the imperceptibility (PEAQ and LSD) and for checking data payload (bit detection rate (BDR)) were conducted to evaluate our method. Our experimental results supported the feasibility of utilizing the CD-based audio information hiding in the LTV system. ln addition, detection regarding both imperceptibility and data payload was better in our method than in the previous blind method.

View full abstract

In this study, we investigate directivity control using two circular loudspeaker arrays and compare it with those of single circular and linear loudspeaker arrays. The design of the directivity control filter is based on the least-squares method in the frequency domain. Circular loudspeaker arrays are considered to be rigid to avoid the forbidden frequency and for practical use. We compare the beam patterns with different shapes of arrays with a single frequency via computer simulation. These are evaluated by comparing the width of the main lobe and the directivity index. The two circular loudspeaker arrays can form a high-directivity beam with a narrower main lobe than that of the single circular loudspeaker array. The two circular loudspeaker arrays have more robust performances than that of the linear loudspeaker array.

View full abstract

In optical-wireless communications, an advanced framed-differential on-off keying (AF-DOOK) system has been proposed by the authors. The AF-DOOK system can solve an optimum decision level adjustment problem and bit-slip problem in the conventional on-off keying (OOK) system. Moreover, the AF-DOOK system can detect frame errors by frame error detecting schemes, named the parity check method and the polarity check method. In this study, the polarity check method for the AF-DOOK system is investigated. In particular, the performance of frame error detection is evaluated by theoretical analysis. As a result, for the AF-DOOK system, the frame error detection probability of the polarity check method is found to be better than that of the parity check method. Moreover, it is clarified that there is a minimum frame error misdetection probability.

View full abstract

Indoor visible-light communications using lighting devices is attracting attention. Using a communication function, a lighting function and a positioning function simultaneously is expected to upgrade the previous optical wireless system. We previously proposed the variable N-parallel code shift keying (VN-CSK) system with a positioning function. Triangulation location scheme using pseudonoise (PN) codes was proposed for the VN-CSK system. In this paper, a VN-CSK communication-based optical fingerprint scheme is proposed. The most suitable placement pattern of LEDs in the optical fingerprint scheme differs from that in the triangulation location scheme in that the optical fingerprint uses the signals received from all LEDs. The positioning accuracy of the optical fingerprint scheme is the same as that of the triangulation location scheme.

View full abstract

Content-oriented networks are proposed as a novel network architecture in which routing is conducted using the content ID instead of an IP address. In content-oriented networks, there is a problem that users cannot discover contents when the mobile communication device that has the contents moves about. Therefore, many studies have been undertaken to solve this problem. In conventional research, by applying ant colony optimization (ACO), Manome and Asaka tracked the movement of contents by adding a pheromone to the route of the mobile device itself. This method improved the discovery rate after movement compared with the case without the pheromone, indicating the effectiveness of their method. However, since the path searching by ACO depends on the behavior of movement of the mobile device, there is a problem of detouring along the route to the contents in some cases. To solve these problems, we used the true slime mold algorithm instead of ACO. In the simulation, we compared our results with those of the conventional method of using ACO to evaluate the performance of our proposed method.

View full abstract

In our previous studies, we showed that the estimation of the rock-scissors-paper (RSP, janken) game strategy is effective for the prediction of a player's hand sign sequences. The purpose of this study is to propose a method to estimate the RSP game strategy in the basis of human personality in an RSP game. To estimate a player's strategy in the RSP game, it is effective to compare the player's hand sign sequence and the hand sign sequences given by various typical RSP strategies on the basis of similarity. In this study, we propose the method of using a homology search to calculate the similarity between sequences. The results show that our proposed method is effective for strategy estimation.

View full abstract

In this study, for the case of evacuation in a disaster situation, we propose a method for automatically selecting installation spots for evacuation signs on a map of the local area. The proposed method is a genetic-algorithm-based method. In our method, the individuals represent the evacuation sign arrangement on the local map, and the objective function to be optimized is defined by the evacuation completion time and the number of evacuation signs. The evacuation completion time is calculated by multi-agent simulation, which is a method that can simulate social phenomena by defining simple agent rules.

View full abstract

Resolution-scalable lossless image compression methods are indispensable for medical imaging and digital archives for example. Therefore, we have developed hierarchical lossless image compression methods using cellular neural network (CNN) predictors. In this method, CNN predictors are assigned for each pixel and their shapes and assignment are optimized to maximize the compression ratio of a given image. In this research, we propose a recompressible hierarchical lossless image compression scheme using CNN predictors optimized by the particle swarm optimization with a refractory period (PSO-RP). The main advantage of the proposed method is that it can compress images quickly using CNN predictors given in an earlier phase of optimization and that recompression is possible via the proposed compression framework. The results of a compression experiment clarified that the compression performance of the proposed method is gradually improved in proportion to the number of repetitions of optimization, and very high compression performance is archived with little performance degradation due to recompression.

View full abstract

A novel method for edge component detection based on image power spectrum sparsity is presented. The edge size can be varied by changing the block size and threshold parameter to obtain the desired edge component. The image is first divided into sub-blocks and the power spectrum sparsity for each sub-block is calculated. On the basis of the image power spectrum sparsity value, each block is verified by the threshold value to determine the edge component. The experimental results show that the proposed method is suitable for object tracking because of the novel feature of the flexible edge size, which can dramatically reduce the amount of data to be stored.

View full abstract

In this paper, we propose a novel method to recognize an H-type landing target accurately based on information only from an image for an unmanned aerial vehicle (UAV). A UAV generally includes an ultrasonic sensor, an infrared sensor, and GPS to obtain its current location. However, there is a possibility that the GPS cannot estimate the correct position depending on the operating environment. By acquiring the height from an image taken with a camera, the UAV can be operated such an environment. In this paper, we focus on the recognition of landing targets, especially for autonomous landing.

View full abstract

We describe the mechanism and attitude control of a biped robot using a spring-damper mechanism for jumping movement. To improve the movement performances of the biped robot, it is necessary to incorporate the leap period into movements such as jumping, which is also expected to further improve surmountability on irregular terrain. To obtain the instantaneous power required for jumping, the proposed robot has a spring damper on its legs. We also designed an attitude controller that stabilizes the attitude of the robot after landing from the jumping movement. The usefulness of the designed controller is demonstrated by physical simulations.

View full abstract

We describe traveling control of an agricultural mobile robot using a depth camera in a greenhouse. In our previous research, we also developed an agricultural mobile robot using a 2D laser range finder (2D-LRF). However, since the detection area of the 2D-LRF is only a 2D plane at the installation height, it is difficult to classify the type of obstacles using the 2D-LRF. On the other hand, the depth camera can detect distances in 3D space and capture 2D color images. For that reason, we propose an autonomous traveling control algorithm for the robot using the depth camera. Depth information output from the depth camera is converted to 2D data and the robot is controlled on the basis of the 2D-LRF algorithm. Several experimental results using our developed mobile robot in a mock-up greenhouse demonstrate the usefulness of the proposed control algorithm using the depth camera.

View full abstract