Although studies of driving assistance system are proceeding, a driving assistance is sometimes recognized as an unnecessary support to the driver. Therefore it is necessary to consider perceptual characteristics of human for an improvement of such driving assistance. In order to obtain the perceptual properties of drivers, the present authors focus on a haptic pedal system and measured the JND (just noticeable difference) to the reaction force on the pedal by utilizing driving simulator with a haptic pedal mechanism. In this paper, dynamic JND against no-static different force conditions was measured and the psychometric function was estimated through approximating on of the measured data.
This paper proposes a speed control method to move an autonomous mobile robot safely under the environment where blind areas exist. Sensors like a LRF or an optical camera cannot detect the areas in the backside of walls, pillars or objects. When a person suddenly appears from these blind areas, the mobile robot is very likely to collide with him/her, if its speed reduction performance is limited. In this paper the safety speed of robot is carefully controlled using the distance information between the robot and the blind areas as well as obstacles. Meanwhile, the optimum path is generated to minimize the cost of moving time, considering the lengthened time cost caused by speed control process. The effectiveness of the proposed method is confirmed by both the computer simulation and the experiment using a real robot.
In surface mount technology, many kinds of image processing methods are used in inspection and positioning for mounting surface mount devices (SMDs) onto PCBs (Printed-Circuit Boards). Currently, such methods require part shape data to detect and inspect the position of parts. These shape data are generated by manual operations which require costs (e.g. operators, times). Automation of the shape data generation has been performed by hierarchical classification method proposed in previous research. Our research focuses on SMDs of complex shape, which was not supported in previous research. We proposed a lead detection method combining frequency analysis and AdaBoost to generate more accurate part shape data. We evaluated our proposed method with 70 parts that were not supported by previous research. The result of evaluation showed that the proposed method reduced false detection rate of leads detection from 24.5[%] to 2.4[%].
We propose quantized feature with angular displacement for pose-based activity recognition. We calculate a 3D joint angle from three postural coordinates. The angular displacement should be quantized since joint angle includes errors due to system noises and similar posture. To investigate appropriate features, we propose four kinds of quantization levels; binarization, ternarization, quaternarization, and quinarization. We apply quantized features in order to improve pose-based activity recognition with the UTKinect-Action Dataset. In the experiment, we show the appropriate feature for activity recognition. As the result, the ternarized feature achieves the highest recognition rate in average. The recognition rate of trials with ternarized feature is improved 2.4% to one with no-quantized feature, and 1.8% to conventional method.
Near-Infrared Spectroscopy has become one of the most important measurement tools of the brain's function. In this study we used a Near-Infrared Spectroscopy while performing a category fluency task to diagnose dementia. 50 dementia patients (age 83.3±6.7, 12 male, 38 female, Mini Mental State Examination: 0-22) and 16 health elderly controls (age 80.5±4.6, 2 male, 14 female, Mini Mental State Examination: 24-30) were instructed to produce a noun belonging to the category of vegetables while being scanned by the Near-Infrared Spectroscopy. Our results showed better test results and also a significant larger prefrontal blood volume among the healthy elderly control subjects compared to those subjects previously diagnosed with dementia. This leads us to conclude that the measurement of the prefrontal blood volume using a Near-Infrared Spectroscopy, usually used to diagnose other psychiatric disorders, could also be used as another tool to diagnose dementia.
Hematocrit monitoring in an extracorporeal circulation apparatus is required for detection of adverse events such as hypotension and bleeding. Nevertheless we have researched non-invasive and continuous hematocrit measurements using optical methods, it remains a challenge to measure absolute levels of hematocrit without calibration. The purpose of this study is to develop an optical hematocrit measurement method without calibration, and to confirm the effectiveness in in-vitro blood experiments. We have developed an improvement method for dual wavelength and detector model by maximizing the difference of optical path lengths. Based on the proposed method, a measurement system has been developed. In the basic blood experiment, the correlation of hematocrit and calculated values by the proposed method has been confirmed. Then, we have conducted hematocrit measurement experiments with attachment errors and change of blood flow. The data indicate that the accuracy of the proposed method without calibration has been equivalent to that of an invasive method, which is employed in clinical practice. In conclusion, we have confirmed that the proposed method is effective for non-invasive and continuous hematocrit measurements.
Solar power business has been expanding rapidly in Japan since new feed-in tariff began in July 2012. Generally, monitoring for failure detection and total electric generating capacity is done at power conditioning system (PCS) in large scale photovoltaic (PV) power plant. Although PCS monitoring can detect its own failure or reduction in power generation amount, it is difficult to monitor in PV string/panel. In this paper, new failure detection method available to detect by PV string, by collecting current value of each string measured at each terminal and by analyzing these data is proposed. The data collection is done utilizing DECT (Digital Enhanced Cordless Telecommunications) wireless communication method realizing easy installation to both existing and new PV power plants. Finally, the effectiveness of this proposed method is shown in the demonstration experiments using actual F Onomichi PV power plant.
Mass consumption of energy is a big problem. Energy-saving activities which are a solution to this problem have been attracting a lot of attention. Nevertheless, in order to engage in thorough energy-saving activities, consumers need support; on the other hand, even with support, the effects seem to be only temporary. In this paper, in order to obtain a long-term effect, we focus on developing a method to raise consumers’ awareness about energy-saving behavior. To this end, we designed an interface which offers consumers the cue to think about energy-saving activities and conducted a 2week-expriment using the interface with 10 participants. The results suggest three things. First, the participants became aware of the importance of thinking about energy. Second, the participants clearly understood which energy-saving activities are important, and which are not. Third, the more times they perform an activity, the less likely it is that they will change their way of thinking about the activity as time passes.
Various association memory models have been proposed with artificial neural networks. For example, an interconnected network model proposed by Hopfield is able to recall stored patterns stably, a chaotic neural network (CNN) proposed by Adachi et al. is able to recall stored patterns dynamically. Kuremoto et al. proposed a multi-layer chaotic neural network (MCNN) with CNNs, which is able to recall multiple time series patterns orderly and dynamically. However, conventional association memory models used to be examined their association ability by experiments with simple binary patterns. In this paper, a novel association system is proposed to realize chaotic recollection of time series for video images using MCNN. In the proposed system, features of video images are extracted and clustered by Kohonen's self-organizing map (SOM), and those clustered feature maps are transformed to be binary images which are stored by MCNN with Hebbian Learning rule. In the recalling process, MCNN outputs time series patterns of different video images in the sense different features, and typical frame of images is able to be reproduced by the median feature vector. Dynamical and temporal association of the proposed system for the video images was confirmed by the experiment results.
This paper proposes novel Artificial Bee Colony (ABC) algorithms for solving problems including interdependence among variables. ABC algorithms are one method of solving multi-variable real number space optimization problems, in which the search space is a set of vectors constructed of variables. The main search process in the ordinary ABC algorithm creates a new solution vector by changing only one variable of the current solution vector. Therefore, the new solution vector is created along only one coordinate axis. This procedure, however, is not appropriate for solving problems including interdependence among variables. For such problems, a method that is able to change more than one variable of a solution vector at the same time is required. In our proposed methods, the original coordinate axes are transformed to linearly uncorrelated axes by using principal component analysis (PCA) in the searching process. Our ABC algorithms create a new solution vector along one of the axes transformed by PCA. Hence, from the viewpoint of the original coordinate axes, the new algorithms are able to change more than one variable. The proposed algorithms have been compared with the ordinary ABC algorithm by solving five benchmark problems. Through the computer simulation results, our algorithms were shown to have better performance for solving problems including interdependence among variables than the ordinary ABC algorithm.
Function localization neural networks (FLNNs) are neural networks that have not only the capability of learning but also the capability of function localization. Function localization in the FLNNs improves the efficiency of individual neurons, and the FLNNs have better representation ability. However, a conventional backpropagation (BP) algorithm for a FLNN training is very easy to get stuck at a local minimum. The reason may be that an error function used for the training becomes complicated because the overlapping modules are switched according to input patterns. By statistical analysis of numerical simulation results, it has been found that there is a strong relation between local minimum problem and the variance of errors calculated for different modules. Based on the analysis result, this paper proposes an evaluation function combining the ordinary sum of squared errors (SSE) and the variance of module SSE, and applies it to a BP training. In this way, the BP training tries to reduce both the error of FLNN and the variance of module errors so as to avoid getting stuck at a local minimum. Numerical simulations are used to show the effectiveness of the proposed evaluation function.
Critical Dimension Scanning Electron Microscope (CD-SEM) is widely used as a measurement tool of semiconductor patterns. It is necessary to set imaging sequence including corrections of imaging position and focusing of electron beam for the reliable measurement. Conventionally, addressing point (AP) and auto-focus point (AF) suitable for these processing are selected by the hand and this is a drop factor of SEM operation rates. We propose a technique to generate imaging sequence from design data of the pattern layout. Proposed method calculates selection indices for pattern complexity, uniqueness, and so on from design data and selects AP and AF templates automatically based on these indices. For 901 evaluation points, success rate of SEM imaging by proposed method was 100% and generation time of imaging sequence was 18 minutes. Compared with manual selection (conventionally it takes ten several hours by using SEM), large reduction of operation cost can be realized.
In automatic defect classification of semiconductor wafers using SEM images, we propose a technique of tuning decision parameters for rule-based defect classifier. The proposed method adopts a coarse-to-fine search for reduction in processing time. However, due to a search leakage, there is no guarantee that the same solutions as a full parameter search can be obtained. In order to prevent the leakage in the coarse search theoretically, the proposed method evaluates a candidate of parameter set based on estimated range of classification accuracy attained by not only the candidate but the surrounding solutions eliminated by the coarse search. The experiments on real image data demonstrate the effectiveness of the proposed method. The proposed method can extract the same solutions as the full parameter search within almost the same processing time as the conventional coarse-to-fine search. When the sampling step of the coarse search is three to six, while the tuning time of the conventional coarse-to-fine search is 1 to 21 seconds, that of the proposed method is 5 to 35 seconds.
In order to establish a novel healthcare monitoring system for elderly people, this paper investigated temporal change of intraoral temperature during drinking cold/hot water and breathing. The temperatures were measured by four sensors located extensively on the palate. As a result, it was shown that from the temperature changes during drinking and breathing, there exist the possibility of these feeding behavior detection.
In this paper, we propose a cluster-structured spiral optimization method that aims to enhance diversification capability to search a solution space more globally than the original spiral optimization. The effectiveness of the proposed method is confirmed through some numerical experiments for five types of typical benchmark problems.
While proximate optimality of a solution space is evaluated by a correlation coefficient of objective function values and distances among solutions, a Proximate Optimality Principle (POP) based new meta-heuristics for combinatorial optimization problems is proposed in this letter. The performance of the proposed combinatorial optimization method is verified through simulations using two types of typical benchmark problems.
In this paper, we propose a computational method of border-collision bifurcation point for piecewise nonlinear discrete-time dynamical systems. First, an n-dimensional piecewise nonlinear discrete-time dynamical system is defined. Next, we show the conditional equation of border-collision bifurcation and propose the derivation method of the bifurcation points in specific terms. Finally, we confirm the validity of the method by applying it to a two-dimensional piecewise nonlinear discrete-time dynamical system.