In this paper, we investigate the moving target tracking problem in multi-robot systems. Each robot has a camera to observe the target and a processor to implement the Kalman filter to estimate the target position, but has limited communication capability and can only communicate with neighbors.We propose to use a novel distributed Kalman filter to estimate the target position. We deduce the distributed Kalman filter based on the standard Kalman filter. The robot group needs to track the moving target, i.e. each robot needs to control its motors to move towards the detected target based on the estimated position. Meanwhile each of them needs to avoid collisions with neighbors. A flocking algorithm is developed to track the moving target. A group of wifibots is used to test the proposed algorithm. The experiment results show that the target tracking algorithm provides a satisfactory position estimation and the flocking group can smoothly track the target.

View full abstract

In this paper, we propose a multi-agent system to support group idea generation by e-mail. In our prior study, as a result of an experiment using a system which provides functions of idea evaluation and chat by synchronous communication, sometimes group communication messages were not uniform in quantity and quality, depending on whether the group communication is effective or not. In order to reduce the quantitative and qualitative unevenness of messages, we propose a new group idea generation method, taking the advantage of asynchronous group communication, and we realize a multi-agent system to support its work. We construct three kinds of agent: Admin agent that manages user access and composes a group; Mail agent that sends and receives e-mail messages; User agent that monitors each member's state and generates instruction messages according to the method. As a result of an applied experiment, the unevenness of messages could be reduced.

View full abstract

Recently, robots are requested to operate in a complex and difficult to predict variable environment like living space. For this request, many researchers study in the field of individual intelligence or multi agents.But there are not many researchers about growth of individual intelligence of robot.In this paper,we will target improvement of individual learning.To realize this, we propose a mechanism which agents exchange information on the crowd about learning method as combination of action selection method and action evaluation method on reinforcement learning and learn learning method adapting to each situations.We apply proposed system to n-armed bandit problem and show a validity with computer simulation.

View full abstract

Agent technologies and agent-oriented design have caught a lot of attention in the software engineering community. However, there has not been a single methodology emerged to become a mainstream approach widely adopted by practitioners. A multi-agent system architecture based on semiotics was proposed in this paper. This approach used a uniform multi-agent model throughout the software development life cycle, from analysis up to implementation without the use of special agent environment nor toolkits. Examples and a case study were used to illustrate the building of multi-agent systems using of the architecture after a brief introduction is presented about how semiotics could be applied in information systems engineering.

View full abstract

Gesture motion has different means according to circumstances. Human recognize other person's intention or attending points from gesture, face direction, situation and so on. In this research, a consistent interface to give the necessary environment phase division and necessary information is constructed according to the scale on which the environment is called the house, the store, and a shopping street. The user's understanding easily from the use of three dimension visualization space as Virtual-City to the interface in the sight, and using it easily become possible. In addition, the installation of the market etc. Achieves the instruction by the gesture in natural shape without the necessity for the interface by using the vision board. Moreover, it introduces the experiment on the system for which the object indicated by pointing provides with each services as one service done by using the constructed system example.

View full abstract

Reliable predictability, which is tightly connected to consistency of environmental changes, is one of the main factors that determine human behaviors. As a constructive approach to understanding this mechanism, the authors have developed a method to generate autonomous object pushing motions based on consistency of object motions using a humanoid robot. The method consists of constructing a dynamics prediction model using Recurrent Neural Network with Parametric Bias (RNNPB), and motion searching based on an object consistency evaluation function using Steepest Descent Method. The model was analyzed through two experiments, pushing cylindrical objects with a humanoid robot. The analysis has shown the method's effectivity and limitations to generate consistent object motions.

View full abstract

This paper discusses how information of an object gives an influence on the handling force prediction applied to the object. Information of the object includes a short-term operation memory and prior information, that is experience on handling a similar object and an observed visual information. We especially deal with a grasping-object motion, where surface EMG (sEMG) signals are measured and the handling force prediction is estimated and defined using integrated EMG (iEMG), that is integrated immediately prior to the grasping-motion. Experimental results show that the predicted grasping force is closely related to the short-term operation memory and the prior information or knowledge of the object.

View full abstract

In this research, we develop automatic driving systems using for a compact electric car "i-mobility". Vehicles assume important roles in modern society. They make our living easier, but also cause many traffic accidents. Most traffic accidents are caused by driver's carelessness and lack of surrounding information. So we proposed robot car systems which are the network intelligence. One of the systems assists in the situation that an ordinary person goes out with an elderly person or a handicapped person who rides on i-mobility. By using the system, on wide street i-mobility runs next to a nursing person automatically. On narrow street i-mobility runs at the back of a nursing person and if there are obstacles in front of i-mobility, i-mobility avoids them by using Laser-Range-Finder and environment information. The other system carry driver to the destination automatically by using the individual attribute and the environment information.

View full abstract

This paper proposes a Environmental Map Building in unknown environments for mobile robots. One of the important intelligent capabilities is to build an environmental map and to estimate and correct the self-location because the robot cannot know the environment beforehand. Map building by mobile robots has a long history. The measured distance is used for matching with an environmental map, but this map should be also generated by the robot itself. This problem is well known as a simultaneous localization and mapping(SLAM). Environmental map should shows various kind of information in environment such as temperature, humidity, barometric pressure, etc.. Basic distance map is made by laser range finder. When the difference between the measured distance and its corresponding map data is large, the robot updates the self-location by using the steady-state genetic algorithm, and updates the map by using topological approach. We propose map building methods based on a topological map based on growing topological neural network. Finally we discuss the effectiveness of the proposed methods through several experimental results and comparison results.

View full abstract

We can estimate the intentions of others and their internal states such as plans by observing their actions. We can also decide our actions based on the estimated internal states of others. On the other hand, others can also observe our behavior and estimate our intentions to influence their actions. When we demonstrate a specific action to others, we can induce their intentions and actions as we intended. However, there are multiple strategies for action decisions when we consider the intentions of others. To achieve efficient interaction between ourselves and others, we must select a strategy from a set of possivle ones of the moment. On the strategy of intention estimation-based action decisions, passive methods that decide self-action based on the other's intention have been studied so far. Incontrast, in this paper, we propose an active method that induces the other's intention to a desired direction and evaluate its effectiveness by computer simulations and advocate the necessity of a "meta strategy" in action decisions and the existence of interaction patterns in such interactive situations.

View full abstract

The author is trying to achieve a breakthrough in the elucidation of information processing of the cerebral cortex, in order to realize highly intelligent robots like human beings. At present, very promising result is obtained. The author designed a neural network model of the cerebral cortex, called BESOM model. BESOM elegantly combines four machine learning technologies, a self-organizing map, a Bayesian network, independent component analysis, and reinforcement learning. Surprisingly, the neural network that executes the algorithm of the model is in good agreement with six-layer and column structures that represent the anatomical characteristics of a cerebral cortex.

View full abstract

A novel computational paradigm with high-dimensional analog chaotic neuro-dynamics is explored. A hardware prototype, which combines analog computations with chaotic neuro-dynamics and digital computation through algorithms, is constructed using analog chaotic-neuron integrated-circuits. In the system, 300-dimensional analog chaotic neuro-dynamics drive a tabu-search algorithm. We demonstrate experimentally that the prototype system efficiently solves quadratic assignment problems (QAPs) as benchmarks through physical chaotic dynamics. We also qualitatively analyze the underlying mechanism of the highly parallel and collective analog computations by observing global and local dynamics. Furthermore, we introduce temporal mutual information to quantitatively evaluate the system dynamics. In addition, we propose a synchronous update algorithm with modified neuron model suitable for a massively parallel hardware system in order to effectively solve large-size QAPs. The experimental and simulation results confirm the validity and efficiency of the proposed computational paradigm with the physical analog chaotic neuro-dynamics. We may assume that this hybrid system would be analogous to the information processing in the brain based on the interaction between conscious and subconscious processes. That is, the heuristic algorithm and the analog chaotic neuro-dynamics correspond to the conscious and subconscious processes, respectively.

View full abstract

In this paper, we propose a vector conversion method of words and a language processing neural network (LPNN) with additional learning. A vector conversion method is a technique of generating word vector by clustering words in thesaurus. The proposed system is able to handle a large number of words converted to vector. The LPNN receives Japanese texts, and outputs or infers the knowledge that related to the inputs. First, the proposed system extracts a relation between words and converts them into triple representations. Then words are converted into vector expression and the word vectors are input into the network. The LPNN is able to learn inputs, even if the first learning process is completed. We carried out several kinds of computer simulations to confirm validity and effectiveness of the network.

View full abstract

This paper presents a new theoretical data mining framework that adapts the existing data mining systems with the architecture of the Knowledge Grid, the mechanism of the ontologies, and the factor of the human-driven knowledge. Aiming at much of the research to date focusing on the technique and algorithms, the new framework describes the essential factors from systemic and technical viewpoints respectively in order to balance the effect between the two aspects.

View full abstract

In many cases, atomic Web services are not able to provide requested functionality. Therefore, the composition task is required on available atomic services. In addition, Context awareness in Web services is gaining momentum as the opportunity to benefit the interactions between human, applications and the environment. So, it seems reasonable to use available context information in composition procedure to provide a composite Semantic Web service which is adapted to the context of participant Web services and user. In this paper, we introduce a novel approach for semi-automatic context-aware composition of stateless context-aware Semantic Web Services. One of the main characteristics of proposed approach is the ability of considering the degree of context-awareness of the service in the composition procedure. Moreover, we use a new schema for Semantic Web service description which is able to represent both functional and non-functional information about context-aware Semantic Web services.

View full abstract

As typified by World Wide Web, a lot of information became accumulated on the Internet. However, most of the currently distributed information is occupied by written document. Compared with it, spoken document is hardly distributed. Therefore, if the mechanism for distributing them can be built, our human society will be able to share much more information. This paper proposes a technique for editing a sentence in spoken document for the purpose of converting it into the Internet contents equipped with the accessibility and readability. By aligning the recorded video data or speech data with the edited text on a fine level, it can be utilized as the multimedia contents equipped with the accessibility. Our technique consists of the following three sentence technologies: (1)paraphrase, (2) division, and (3) structuration. We implemented a spoken document edit system based on our techniques. We conducted an edit experiment by using lecture speech data and our technique could achieve high accuracy. From the results, we confirmed the availability of our technique.

View full abstract

In public spaces, there are a number of different contents such as visitors, physical information resources, and virtual information resources via their embedded output devices (e.g., displays and speakers). Therefore, we might unexpectedly enter the public spaces that have our unfavorable characteristics (e.g., dismal and dangerous) and/or our unwanted information. For this problem, we have proposed a model and architecture of "Secure Spaces", which provide access control over public output devices and entry control over electrical lockable doors to prevent visitors from entering the public spaces that have their unfavorable characteristics and/or their unwanted information according to their access policies for spaces or information, that is, what spaces or information they do not want to access. This paper tackles how to extract information for making access or entry decisions in Secure Spaces from very large text corpora such as the Web to enables users to more flexibly specify their access policies by keyword-based expressions in practice.

View full abstract

Recently, the sites in internet on which users can write private ideas and opinions are increasing. In addition, the number of people who want to know other's opinions about the interested products is also increasing. However, it is very difficult for people to read whole reviews on internet. This study tries to develop a new review analysis system which shows evaluation information about products using graph structure of evaluation keywords. This paper focuses on the part of extraction of evaluation keywords from reviews on internet. This paper shows the extraction method for evaluation keywords and the result of keyword graph structure with extracted evaluation words. It employs HK Graph (Hierarchical Keyword Graph) which can visualize the relationship among words with hierarchical network structure based on the co-occurrence information for the keyword graph.

View full abstract

With the rapid growth of Web-based social networks, it is necessary to provide a way to compute the trust among neighbors in network. Several works have addressed the problem of modeling the trust and its propagation in a social network. But, in this paper, we'll present a method to infer trust from a social network which does not contain any explicit trust information. Actually, we use a game theoretic approach to compute the trust values beyond links in Blogsphere. We'll show the intuitions behind our methods which will lead to the Nash equilibrium. Experimental results show that Bloggers should not fully trust their neighbors, rather, the average recommended amount of trust is a few more than 50 percents.

View full abstract

In this paper we present a new robust approach for eye localization adopted into lip reading in mobile environments, where the input image is assumed to contain a single face. Firstly, we segment eye candidates regions using intensity information in YCbCr color space. Then coupled regions of eye candidates including assumed eye brows are extracted with the restrictions of morphology and geometry characteristics of eyes in frontal face. GMM is applied next to validate exact eye couple.

View full abstract

This paper proposes an advanced techniques for exciting and receiving the torsional guided wave to detect flaws in pipe systems. It is known that there are some difficulties in selecting and exciting of modes by using the nickel strip attached on pipe systems, such as qualification of residual magnetic field and excitation of the unwanted modes etc. In order to overcome these difficulties we propose a new sensor, so called Crossed-coils sensor. We prove that it is possible to select the modes to be excited and to find an optimal excitation condition for torsional modes by using the proposed sensor.

View full abstract

In this paper, a different internal fault modeling of transformer based on the physical information of the transformer (equivalent circuit parameter) using transmission line method (TLM) and an Fault Identification algorithm method using FIS based on PCA in Matlab is presented. A pilot transformer is modeled by considering the non-linearities as hysteresis and saturation and an internal fault short circuit as internal fault portion is introduced into the model. Then, the transformer internal fault currents are discriminated from the rated currents and the degree and priority of transformer internal faults are obtained by the proposed techniques. Results were compared to real time measurement to investigate the accuracy of the proposed method.

View full abstract

We obtain some characterizations of continuous, open and closed functions in intuitionistic topological spaces. Moreover we reveal that the category of topological spaces is a bireflective full subcategory of the category of intuitionistic topological spaces.

View full abstract

There have been various proposals on cluster validity functions but they have not yet been compared using many numerical examples. In this study we compare performances among different cluster validity functions. That is, five measures of the sum of determinants and sum of traces of fuzzy covariances of clusters, Xie-Beni index, Davies-Bouldin index, and Fukuyama-Sugeno index as well as their kernelized versions are considered. In particular, algorithms for calculating the kernelized measures are shown. Effectiveness of these indices are compared using thousands of automatically generated clusters. It will be shown that no single measure outperforms others, and the sum of traces performs as good as that of determinants, contrary to the common understanding. Moreover, kernelized measures perform as well as non-kernelized ones.

View full abstract

Since different types of classifiers work best for different types of data, our approach is to parameterize the classifier and tailor them to individual data set. In the proposed fuzzy c-means (FCM) classifier, parameter values are chosen by evolutionary algorithms such as the particle swarm optimization (PSO). The free parameters of membership functions are optimized in order to increase true positive rate or true positive count. The golden section search is applied to select a cut-off point, which realizes a preselected positive count or false positive count. This makes classification decisions more pragmatic than those by the standard methods with missclassification rates or loss functions.

View full abstract

In the field of marketing, companies often carry out a questionnaire to consumers for grasping their impressions of products. Analyzing the evaluation data obtained from consumers enables us to grasp the tendency of the market and to find problems and/or to make hypotheses that are useful for the development of products. Semantic Differential (SD) method is one of the most useful methods for quantifying human-impressions to the objects. The purpose of this study is to develop a method for visualization of individual features in data. This paper proposes the Clustering method based on Orthogonal Procrustes Analysis (COPA). The proposed method can cluster subjects among whom the distributed structures of the SD evaluation data are similar. The analysis by this method leads to discovery of majority/minority groups and/or groups which have unique features. In addition, it enables us to analyze the similarity/difference of objects and impression words among clusters and/or subjects by comparing the cluster centers and/or transformation matrices. This paper applies the proposed method to an actual SD evaluation data. It shows that this method can investigate the similar relationships among the objects in each group and compare the similarity/difference of impression words used for the evaluation of objects among subjects in the same cluster.

View full abstract

Each data on a real space is generally transformed to a point in a pattern space and analyzed in clustering. Actually, the data should be often represented not by a point but by a set because of uncertainty of the data. From such a viewpoint, some clustering algorithms which can handle the data with uncertainty have been proposed. In the conventional algorithms, uncertainty of data is estimated by the given range. However, the algorithms cannot handle the data of which the uncertainty is not regarded as range. In this paper, we consider new optimization problems in which the uncertainty without the range and construct new clustering algorithms based on fuzzy c-means. First, the uncertainty of the data is introduced into optimization problems using spring modulus. Next, the problems are solved and some algorithms are constructed based on the results. Finally, usefulness of the proposed algorithms is verified through numerical examples.

View full abstract

In this paper, we propose a new clustering algorithm based on probabilistic similarity. The probabilistic similarity is formed by introducing the concept of probability into conventional similarity. First, we define the probabilistic similarity. Next, we show some probabilistic similarity functions which satisfy the definition of the probabilistic similarity. Third, we consider a objective function with the probabilistic similarity. Furthermore, we construct a new clustering algorithm based on the probabilistic similarity by using the optimal solutions which minimize the objective function. Last, we show some numerical examples to verify the effectiveness of the proposed algorithm.

View full abstract

The purpose of this paper is to propose a method for automatically evaluating category names and category structures in the Web navigation system. The evaluation system requires only category names with a link structure and content articles. This approach consists of three stages. First, the system classifies the articles into the categories based on the similarity between the category name and the article. Second, the system identifies usability problems. Third, the system calculates an index that indicates the appropriateness of the category names and category structures. The system would be useful in designing a category menu interactively. As applications, a category name selection problem and a link structure evaluation problem are discussed. This system was also applied to real data such as Encarta's and Wikipedia's category menus. The results confirmed the usefulness of the system.

View full abstract

An increasing number of social networks can be found in the internet with the latter's widely increasing popularity. From web blogging to photo and video sharing, social networks have provided everyone equal opportunity to make them self heard in the virtual world and establish and maintain connections with other people. Despite the increasing popularity of social networks, not so much focus have been done on creating a social network system specifically targeted for migrant workers. Mailing lists and forums of migrant workers can be found anywhere in the net, but there is no specific discussion for social problems of migrant workers. Hoping that we can help migrant workers with their problems, we came up with a research project that will focus on creating a multi-agent based social security system for migrant workers.

View full abstract

The paper first describes a model of clinicians' diagnostic process and discusses how the clinicians diagnose their patients. It is important to detect possible critical diseases at early stage of medical diagnosis regardless of their frequency or probability. Therefore, we propose a system that applies inverse causal reasoning to diagnose possible diseases from observed clinical information, which is based on possibilistic causal relations between diseases and clinical information. Most of clinical diagnosis systems proposed so far adopt single disease assumption. However, multiple diseases often occur simultaneously in clinical reasoning. The paper also discusses the assumption of multiple-diseases to support the clinical reasoning.

View full abstract

Each people might have different way of walking which correspond to his/her habit. Automated human identification from their walking behavior is challenging problem and has gotten much interest from researchers in the field of machine vision system. However, the systems which are able to detect the specific pedestrian attributes based on their walking behavior have not been developed. Thus, a soft computing approach for understanding human behavior based on motion imagery are studied as the basis development of pedestrian safety information system. In this stage, classification of human gender and age are developed based on pattern of human motion through the experiments. At the front end, image and video processing was performed to separate the foreground from the background images. The widths of human shape were analyzed by 2D Fourier transform to extract human motion features. Feature sub-set selection methods were then performed to find the salient and the effective features for the classification process. Finally, Choquet integral agent networks (CHIAN) with competitive learning algorithm were employed to classify human gender and age among several classes. The experimental results demonstrated capability of the proposed system to classify the gender and the age in high accuracy rate.

View full abstract

This paper proposes an approach to human motion understanding by computational information fusion intermediating words. The procedure consists of three stages, i.e., (1) feature extraction from sequential pattern data of physical body motions, (2) motion modality recognition with respect to basic and modal phases, and (3) semantic understanding of behavior, intention or emotion classes. For these three respective stages, "Eigen-Space Methods (ESM)", "Hidden Markov Models (HMM)", and "Choquet Integral Agent Networks (CHIAN)" are employed. Based on the subjects experiments for expressing their emotion by human gesture in everyday life, the computational information fusion mechanism was constructed, and the effectiveness of the proposed computation procedure for understanding emotion emerged from gesture was investigated.

View full abstract

The dissociated culture is a suitable system to observe behavior of functional molecules and an intercellular interaction in detail. The cultured neuronal network was confirmed to have a mechanism to modify own functional structure automatically. We found characteristic transient high frequency burst activity(HFB) during culture days. The neuronal network generated such particular spontaneous activity, suggesting that the HFB autonomously constructed a heterogeneous network structure suitable for neurotransmission. After the HFB, some neurons became to have many inputs while others became to have little. We confirmed that the HFB terminated within 10days, without a drastic change of the number of neurons. It suggests that the termination of the HFB activity was not caused by the decrease of the neurons, but was performed by the modification of the intercellular interaction.

View full abstract

Saccade-related electroencephalogram (EEG) signals have been the subject of application oriented research by our group toward developing a brain computer interface (BCI). Our goal is to develop novel BCI based on eye movements system employing EEG signals on-line. Most of the analysis of the saccade-related EEG data has been performed using ensemble averaging approaches. In signal processing method for BCI, raw EEG signals are analyzed. In ensemble averaging method which is major EEG analysis is not suitable for processing raw EEG signals. In order to process raw EEG data, we use independent component analysis. This paper presents extraction rate of saccade-related EEG signals by four ICA algorithms and six window size. As results of extracting rate focused on ICA algorithm, The JADE and Fast ICA have good results. As you know, calucuration time in Fast ICA is fater than calucuration time in JADE. Therefore, in this case, Fast ICA is best in order to extract saccade-related ICs. Next, we focus on extracting rate in each windows. The windows not including EEG signals after saccade has and the windows which has small window size is good extracting rate.

View full abstract

An easy-to-use generic platform for realizing online brain-computer interface (BCI) named xBCI developed by authors is introduced and described. The platform consists of several functional components such as data acquisition, storage, mathematical operation, signal processing, network communication, data visualization, and feedback presentation. Users can build their own BCI system easily by combining components on GUI-based diagram editor. The platform works on multiple operating systems, and supports parallel (multi-thread) data processing and data transfer to other PCs through network(UDP). In this paper, the performance of the platform was evaluated, and then two types of online BCI systems which were built on this platform are introduced. First, a "Brain Switch" system which detects motor imagery through the changes in band power of electroencephalography (EEG) was built. The system consisted of components for acquisition of EEG signals, band power extraction, feedback (a white bar of which length was proportional to band power) presented to subject, data transfer through network, and data visualization. The second system is a steady-state evoked potential (SSVEP)-based BCI system. On this system, subject was instructed to gaze at LEDs flickering at different frequencies. EEGs were recorded and discrete Fourier transform (DFT) calculation was performed to extract frequency components at flickering frequencies and their harmonics. Both systems were built easily using the components available on the platform. On-line experiments were conducted to verify the usefulness of the platform.

View full abstract

Implementing acupuncture stimulation, we have proposed a new technique to reduce muscle fatigue in functional electrical stimulation (FES), one of neural interfaces for restoration of motor function. For the analysis of mechanical stimulation, we have recorded torque and EMG during leg extension exercises. According to our experimental results, press needle stimulation over the posterior root of the lumbar nerve L2 and L3 inhibits the rectus femoris muscle. We have speculated that the dermatome might be important to link the stimulation point and the target muscle.

View full abstract

Recently, researches on interactive art regarding computer graphics and computer music are actively performed. In the computer music, genetic algorithm(GA) and interactive genetic algorithm(IGA) are usually used for the sound generation process, but it is difficult to just handle these methods directly for the music composition. In this research we propose a music composition system based on Twelve-tone Technique that is a music composition technique of the contemporary music. Because it is a technique has an advantage to compose the music step by step, the Twelve-tone Technique has an advantage that it is suitable for the composition of computer music. Making of twelve-tone Technique is an important process to determine the subject and atmosphere of music. However, it is difficult for a human without the knowledge of music to evaluate twelve-tone rows. In this paper, we design the fitness function based on the relation between the consonance interval and dissonant interval that are a part of general musical theory and perform the automatic generation of twelve-tone row by searching the consonance tone row with GA. We confirmed the effectiveness of this system by constructing the simulator that generates twelve-tone rows and executing the comparison questionnaire of twelve-tone rows made by the simulator and human.

View full abstract

The development of a successful consumer product line often depends on the development of a marketable selection of colors. Although previous studies have elucidated color preference and sensibility for individual product colors, few studies have investigated the relationship between different types of products and their associated color preferences. The objective of this research is to begin build the basis of models for understanding the relationship between the function of a product and the preference for color when they are applied to that product. This paper describes the development of a model relating product function to color preference. Selection of a color from a palette allowed a subject to see how various products would appear in the selected color, which was applied immediately to a product image. At this stage, subjects were instructed to select colors from the palette and evaluate each selection. This data was used to develop a model correlating product function to product color preference. Additional experiments were conducted to evaluate the model's ability to determine color preference from function. This paper provides a method for product designers to quickly assess their own products color selection requirements, and for marketers to determine color selection demands for a variety of products.

View full abstract

The paper proposes person identification system using clothes color. Clothes color information is important for searching a specific person. Authors developed a person identification system using image matching. General image matching algorithm is far from human sensibility. Therefore, authors implement color clustering which is suitable human sensibility for the identification system. In the experiments, effectiveness of proposed system is shown.

View full abstract

Various colors are used for things that we use in daily life. A color combination is important for design and has a great influence on image or human kansei. The aim of this study is to construct a color combination support system considering personal preference and human kansei. In this paper, we propose a IGAs based color combination support system that shows color combinations that the user likes more. In this system, the user can makes a favorite color in the beginning of the system first. This color is one color composed of three color combination, and the other two colors combined with this color are generated by IGA. Also, the sort function are introduced in order to support user's relative estimation. Evaluation values that user given to candidates were higher than that of an initial generation in later generations, color combinations of user's favorite was obtained. Also, effectiveness of functions of changing the position of the favorite color and sorting candidates were shown.

View full abstract

Soft actuators are expected to be applied for human assistive devices by making multiple actuators cooperate. However, it is difficult to cooperative control them using conventional linear controllers because soft actuators have nonlinear characteristics. On the other hand, humans are able to control own muscle which is nonlinear actuator and work together. In this paper, an intelligent cooperative controller that controls cooperative movement of multiple nonlinear actuators is proposed. The controller was constructed referring to a human's cooperative work, and consists of multiple control parts, each control part handles each actuator. The effectiveness of the proposed controller was confirmed experimentally using a pneumatic lift device.

View full abstract

In the field of human-machine interaction mechanical impedance has become well implement. However, in human-vehicle interaction there is a variant environment when drivin a vehicle, thus apart from cooperating with human, an impedance is needed to adapt to the variant environment. But as being a second order mass-viscosity-stiffness model with fixed parameters, for mechanical impedance it is difficult to adapt to variational constraints. In this paper, a fuzzy virtual impedance of vehicle is studied. Fuzzy virtual impedance is based on fuzzy instruction that is a fuzzy set of control instruction candidates with reasoning of TS fuzzy model. The characteristics of achieved impedance is as a spring part with variable parameter so as to adapt to the variant road situation. A human-vehicle cooperation experiment was preformed to demonstrate the practical application of the proposed impedance and verify the adaptability to variational constraints.

View full abstract

The self-localization of mobile robots by using GPS and INS with extended Kalman filter has attracted significant research attention in recent years. Although the extended Kalman filter(EKF) has been extensively employed to solve these problems in mobile robots, the performance of the EKF can degrade significantly, if the correct a priori knowledge of sensor/measurement noise covariance matrices is not available since depending on available GPS satellite positions. In this paper, we propose a new particle filter based self-localization method for autonomous navigation in outdoor unknown environments, and compare the conventional GPS with extended Kalman filter and the GPS with particle filter.

View full abstract

Research and development of an unmanned system for various uses is underway around the world. However, in spite of the rapidly growing market for unmanned systems, there is not yet a standard control interface for such systems. One potential standard control interface for unmanned systems is the Joint Architecture for Unmanned Systems(JAUS) originally proposed by the United States Department of Defense to develop an open architecture for unmanned systems. It is a component-based, message-passing architecture that defines a data format and methods of communication between subsystem computing nodes. This paper describes the development of a JAUS module for the IGVC JAUS Challenge. Based on JAUS documents available on the Internet, we implement an actual unmanned ground system and demonstrate the validity of JAUS-compliant unmanned systems.

View full abstract

As there increase many senior citizens recently, robots for welfare will be needed. For such robots, the recognition of the human body and posture estimate are necessary. We proposed a new model-based recognition method which includes 3-D ellipsoidal models and their 2-D appearance models. Original images are obtained from thermograph. As 3-D models can be expanded or reduced, this method is applicable to various figures of human body.

View full abstract

We describe an artificial ecosystem consisting of five areas and evolving artificial creatures (called agents). The ecosystem is for an analysis of migrations of the Monarch butterfly. The Monarch butterfly is famous for its cross-generational migration. We report simulations on the emergence of migration biology pertaining to the Monarch butterfly. The area has two kinds of environmental changes: long-term and short-term changes. We focus on temperature as an environmental parameter. Under long-term change, temperature is gradually rising, and under short-term change temperature changes periodically as same as seasonal change. We put agents on the areas. The agent has two genetic components: an environmental adaptation scale and an action decision table. These components represent the physical features of the agent and select an action on the basis of sensory information, respectively. The agent also has a temperature sensor that functions with its environmental adaptation scale. It enables the agent to adapt dynamic temperature changes and to evolve to obtain optimal behaviors. With the ecosystem, we conduct one experiment. The result was that we observed that the range of migration expanded as the temperature rose. Also, we report the result of migration patterns obtained by the agents. These results show that the biology of the Monarch butterfly is well modeled by the ecosystem and our evolutionary method.

View full abstract

Not a day goes by without hearing music, these days. Music helps us to ease the tensions in our life in contemporary society with various stresses. Many musicians hope to compose the songs that help people to be relaxed, to get them excited, or to move. Therefore, So, clarifying the relationships between music and the brain is important and useful. In our research, we analyzed brain waves of subjects that were listening to harmony, one of the elements of modern music, and determined the correlation between listening to harmony and the relaxation index, as indicated by brain waves.

View full abstract

For creatures in the natural world, parasitism is one kind of relationship with other species of creatures. The subject of this paper is the pathogen, a kind of parasite. Pathogens evolve to one with adaptive virulence and thus survive and multiply. We also discuss the effect of how they are transmitted on their evolution.We propose an evolutionary model of the pathogen that takes vertical transmission into consideration. We constructed a spatially-structured population by cellular automata, carried out simulations, and investigated how vertical transmission affects the evolution of the pathogen. Furthermore, we discovered the direction of the evolution changes with the parameters.

View full abstract

When domestic robots with faces are used heavily, the motors controlling their faces wear out and need to be replaced. We used the communication robot ``Ifbot'' to investigate ways to reduce this motor wear by creating facial features, expressions, and transitions between expressions that reduce the amount of motor movement needed and use LEDs effectively. We created face changes for four emotions (happiness, disgust, sadness, and anger) that were distinguished by observers with a statistically significant difference, and confirmed findings of face changes for surprise and neutral emotions. However, we could not create any face changes that were recognized as fear with a significant difference.

View full abstract