Image registration is an important problem and a fundamental task in computer vision and image processing field. Especially, CT and MR imaging of the head for diagnosis and surgical planning indicate that physicians and surgeons gain important information from these modalities. In this paper, we develop a new method for automatic registration of head CT and MR images by using ICP (iterative closest point) algorithm in several extracted data and maximization of mutual information.

View full abstract

ACL (Anterior Cruciate Ligament) reconstruction is performed to recover the injured knee which often happens in sports activity. The ACL reconstruction makes two bone tunnels in the femur and tibia. Then, the harvested graft passes the bone tunnels so that the femur and tibia are connected. However, the bone tunnel is enlarged with time. The bone tunnel enlargement can cause instability of the knee. To examine relationship of the bone tunnel enlargement and clinical results, the bone tunnel should be evaluated quantitatively and three-dimensionally. Therefore, this study proposes an automated extraction method of bone tunnel from MDCT image by using fuzzy inference. Furthermore, the volume of the bone tunnel is automatic and suggests a method to extract. The experiments compared extracted volume between manual operation and the proposed method. As a result, the proposed method could automatically detect the volume of bone tunnel.

View full abstract

We propose feature extraction algorithm for ultrasonic images of thyroid gland based on neighbor organ regions. It estimates the target region using mark of search unit and detected region in air passage and carotid artery. The detection experiments for real ultrasonic images taken in Kochi medical school hospital show that error in diameter of thyroid gland is 0.11 cm. Using the proposed algorithm, it is possible to reuse medical treatment information for analysis and study.

View full abstract

This paper describes a method for heartbeat rate and respiratory rate monitoring system using an air pressure sensor. Many people suffer from sleep disorders such as Sleep Apnea Syndrome (SAS), and these population increases recently. In generally, polysomnography (PSG) is used for diagnosis of sleep assessment. However, this monitoring system constrains patients during the measurement. The air pressure sensor can detect unconstraintly the vital information in the bed. We propose a detection method of heartbeat point and respiratory point using fuzzy membership functions. The experimental results show that the best error ratio of heartbeat rate was 1.71 % and the best error ratio of respiratory rate was 2.08 %. Consequently, this system can noninvasively detect vital signals by using an unconstrained device.

View full abstract

The control method to organically unite two or more state vector fuzzy control devices who are two dimension fuzzy controllers is developed in 1991, and announced in the first time FAN symposium. This is a control method to raise Pole of the cart-pole system from the state of the pendency, and to do the headstand stabilization and the cart positioning control at the same time. Afterwards, this control scheme is applied to the Arm-Pole system etc. Moreover, these computer simulations are demonstrated in *technological exchange in Tsukuba 2000*. In this paper, this control scheme is applied to the Segway type coaxial two-wheeled vehicle running control, and running on a curved surface road etc. have been achieved. In addition, this control scheme is applied to two pole headstand stabilization control of the cart-pole system that has two parallel poles simultaneously. As a result of one device, the purpose of two pole headstand in the cart starting point simultaneously is achieved.

View full abstract

We have developed an underactuated unicycle robot which consists of a body, a disk driven by a dc-motor to stabilize its posture, and a wheel driven by another dc-motor to move the body forward. The robot has only two motors and there is no direct force to change the yaw direction of the robot. The present paper proposes a control method for changing the yaw direction by using sinusoidal inputs, where the sinusoidal inputs for the roll and for the roll have the same angular velocity. At first, the yaw rate is shown to depend on the phase difference between these two sinusoidal inputs. And then a regulator is proposed, which changes the yaw direction to zero from any initial value by modifying the phase difference. Next, a tracking controller is proposed, which consists of the relation between the rate and the phase difference and the Kanayama's tracking control method for a car-like vehicle. The performance of the two controllers is confirmed by simulation results.

View full abstract

The inverted pendulum is a system that has the state of the multi parameters. The inverted pendulum is a SIMO(Single-Input Multiple-Output) system. In the SIMO system, it is difficult to construct by conventional controller which considering multi parameters at the same time. On the other hand, it is easy to construct by conventional controller that considering one parameter. Moreover, human can judge the state operated by priority each situation of the inverted pendulum. In this study, we propose intelligent hierarchical controller. This intelligent hierarchical controller consists of two parts. One is the controller group which has controllers of each parameter. Another is the judgment part where human knowledge was built in. The proposed controller is applied to swing-up and stabilizing control of the inverted pendulum. The effectiveness of the proposed controller was confirmed experimentally.

View full abstract

This paper proposes a fuzzy clustering model based on the idea of self-organized dissimilarity between two objects. A conventional kernel fuzzy clustering model is defined as a model which is an improved additive fuzzy clustering. The purpose of this conventional model is to obtain a clearer result by consideration of the interaction of clusters.The purpose of this paper is to improve the performance of the kernel fuzzy clustering model by introducing a self-organized algorithm

View full abstract

This paper dicusses an online updating approach of fuzzy rules for voice instruction systems. The fuzzy reasoning outputs the 3 dimensional traveling distance of robotic forceps. The fuzzy coach-player system is one of the voice instruction approaches to control robotic systems. The effectiveness of proposed method is demonstrated by some experiments.

View full abstract

In this paper, we propose the method of stability analysis of nonlinear fuzzy systems using Multi-max/min based Lyapunov functions. Multi-max/min based Lyapunov functions have the generalized form of piecewise Lyapunov fuctions. We show that our method provides more efficient conditions for the stability analysis of nonlinear systems than the existing method using two-term quadratic piecewise Lyapunov functions do. First, we discuss about the inequations derived from Multi-max/min based Lyapunov functions in general case. Secondly, we apply these topics to fuzzy systems and derive the stability conditions in the form of BMI (Bilinear Matrix Inequarity) by using S-procedure lemma and propose the approach which combines MATLAB LMI Control Toolbox and PSO to estimate the solutions of BMI, which is difficult to solve. Finaly, we analyze a numerical example to show that our method provides more relaxed stability results than existing piecewise Lyapunov functions method.

View full abstract

The research on fuzzy control is various and performed. Fuzzy control is used for the control methods, such as automatic conveyance vehicles, such as a line trace robot and a factory, in many cases. It is verified whether fuzzy control is applicable to an object with a fast running speed like line trace car. In this research, the simplified reasoning method is used for the steering angle and the motor control of the line trace car. We tried to make smoothly controller without carrying out a zigzag run. About line tracer system, the car calculates a steering angle from the value of the sensor read during the run. Then, the line trace car calculates a motor output from the outputted steering angle.

View full abstract

It is difficult to do the evaluation and the verification on an actual road when the measure to a road traffic problem is held from the safety side and respect of time and cost. Then, evaluating it with a road traffic simulator is effective. We develop microscopic road traffic simulator MITRAM, and are doing various researches. The driving operation model is constructed by using a fuzzy method in MITRAM, and each vehicle does natural driving operation according to the traffic situation. And, a lot of this vehicles run and the road traffic is expressed. Moreover, the traffic signal control in the simulation can be made outside in MITRAM. In this study, we constructed the signal control system outside of MITRAM, and it can mount a complex algorithm on simulator. As a result, a control scheme actually done and a dynamic control scheme can be easily applied to the simulator.

View full abstract

Recently, a lot of studies in which Genetic Algorithm (GA) is applied to Multi-objective Optimization Problems (MOPs), which are called MOGA, have been reported. It is generally important to search a set of Parate solutions in MOPs, and then GA is effective to apply because of the multi-point search. The performance of MOGA, however, decreases with increasing the number of objective functions. Therefore, effective search of MOGA is the important issue in many objective optimization problems. It is one of the effective approaches to assemble some objective functions and reduce the number of them. However, it has not studied the appropriate assembling way and the number of them. The purpose of this study is to grasp the effects of assembling objective functions. This paper studies the effects of assembling when MOGA is applied to simplified Nurse Scheduling Problem (sNSP) and compares two assembling ways based on their meaning and their correlation coefficients.

View full abstract

A low voter turnout came to be a problem in Japan recently. In addition, some local governments in Japan should rebuild government finances due to their severe monetary environment. We have already considered the change of the polling place assignment in a Japanese city to increase the voter turnout and to reduce the number of the polling places using regional polling coefficients estimated from a past election result. Using those regional polling coefficients, we changed the polling place assignment using an EMO algorithm. In our previous study, we identified regional polling coefficients from a single election result. In our present study, we identify them by several election results, and analyze the difference of them. In particular, we compare the difference of regional property and results of polling place assignment by employing an EMO Algorithm, and consider the validity of the way of identifying regional polling coefficients.

View full abstract

Japanese pension system is operated by pay-as-you-go scheme now. However, the financial stability and the fairness between each generation have become serious problems due to the falling birthrate and the aging population. Generation-based funding scheme that was proposed by Banjo, demonstrates the effectiveness for the above problems using a macro simulation model. On the other hand, lately, in Japan, decentralizations are promoted and role sharing between national and local governments has been reconsidered. In this paper, we propose a regional public pension system with generation-based funding scheme which is adapted to each region, and analyze the possibility and the characteristics of the proposed using a multi-agent simulation model.

View full abstract

Research in humanoid is growing widely and hoped for application to life assistance, nursing welfare and so on. It is difficult to design every motion pattern of robot beforehand which is available for wide variety operation environment. Techniques which robot learning workers' motions by imitating their motions are remakable because of using it together with motion learning. In conventional method, each motion pattern was extracted as a frequent motion from observed time series joint angle data, such as 'Walk around' and 'Swim a breaststroke' using whole body joints. On the other hand, subclass motions are observed such as 'swing a hummer' and 'hold a box'. Desired motion are expressed with partical joints like on left hand or right hand, though previous technique extracts motions with all of joints observed.Therefore, we propose the motion extraction method with selected partical joints which is applied to synthesize new motions from observed time-series whole joints data. Redundant joints for motion is detected as non-synchronous joint motion on several motions.

View full abstract

In this paper, we focus on emotion recognition using prosodic features in speech. It can be occurred that prosodic features on the utterances with two different emotions may show similarity. Therefore, distributions of each emotion are overlapping, so that the accuracy of emotion recognition goes down. This paper proposes the emotion recognition technique which can distinguish each emotion category in overlapped distributions. In proposed technique, to recognize emotion in overlapped distributions of emotion, we put remarkable changes of prosodic features on emotion transition to use. In the experiment, we treated 4 emotion transitions; "Interest to Sympathy", "Interest to Joy", "Interest to Doubt" and "Interest to Antipathy". We calculated the differences on prosodic feature between before and after the emotion transitions, and drew histograms of differences of prosodic features. We show that there are differences among emotion transitions in points of kinds of changed features and their amount of change. Moreover, we show that there are significant differences statistically among the amount of change of prosodic features.

View full abstract

Emotion recognition becomes an increasingly important research to understand human's physical state for human-computer interfaces. Though the speech data could be obtained through communication between a human and a computer as monitoring operation words or general talk with other person, traditional emotion recognition techniques could not be applied easily because they needs utterances extracted as a single word. In order to extract each words from natural speech, another technique such as morphological analysis must be prepared which depend on user speech property. As emotion recognition be established without the properties of a certain person, natural speech is divided into not word-term but short-term utterance. When the length of utterance becomes shorter, prosodic features from the utterances shows the similarities, so that each emotion label can not be classified based on sloppy division technique.

View full abstract

To investigate the difference of recognition concerning developmental change and independence level by the age of subjects, three different aged groups were compared with fuzzy estimation methods. 120 young person subjects (18-22), 100 middle-aged (40-49) subjects, 100 elderly subjects (65+) were asked to evaluate 5 development estimation item groups to see to what degree 1-80 years old people would be able to perform independently and adequately. The subjects were also asked to evaluate independence level of 1-80 years old people and evaluate the degree of the relation between mean values of 5 item groups and the independence level in each age. Mean scores of 5 item groups (4 development evaluation items were included in each group) of 60 years old of the middle-aged and the elderly groups were higher than mean scores of the young person group. Possible reasons and influence of difference of three groups about recognition of functional changes and independence levels of each age were discussed.

View full abstract

In many schools for the visually impaired, there is no department to produce tactile teaching material. Usually, volunteers or teachers produce most tactile graphics using less intelligent computer-aided systems. So, a better computer-aided system of making tactile graphics is needed. Because of this background, we are developing a computer-aided system for automating transformation of mathematical figures into tactile graphics. A mathematical figure often includes character strings or mathematical formulas to explain lines or curves that are graphical expression of functions or relations. Furthermore, a graph is sometimes drawn by using broken lines. Then, this paper describes the following two methods, (1) the separation of a mathematical figure into the character area and the graph area, and (2) the extraction of broken line components and the classification of these components into dotted lines and the chain lines. This paper shows experimental results for our methods.

View full abstract

This paper introduces a safe exoskeleton powered hand system to augment pinching force of human fingers: an index finger and a thumb. Two items to keep system safety are proposed so that it can generate almost same grasping force as an adult male's. One is from viewpoint of a mechanical design. Most of the assistive force directly affects to a target object to avoid an excessive force affecting a human finger and then a rest affects to it through a human finger to inform magnitude of an assistive force. The other is from a controller design. The system switches two control algorithms based on bioelectrical potential. The system in a power assist control mode generates an assistive force that is proportional to an integral of bioelectrical signal if it can be measured. In following control mode it stops support if the signal is not measured. A wearer stops the support quickly and releases an object intuitively. Basic performance of the system is investigated through some experiments.

View full abstract

In this paper, we propose the Personal Software Process (PSP) Practice Support System efficiently to transmit only necessary programming capability and improvement of error. This system can transmit programming to specific human among many software processes using a Multiagent technology. The system is also synthesized to do parallel and cooperative proposing internally. Applying the proposed method to a personal process-removing task, a flexible programming for quality of software. Agent studies the best transmission programming, planning and quality according to the makes planning in the personal process.

View full abstract

We discuss a relationship between topology and reasoning.

View full abstract

We propose a generalized inference metho for constructing natural language communication. The method is used to obtain fuzzy quantifier Q' when "QA are F is tau -> Q'(mA) are mF is m''tau" is inferred (Q, Q':fuzzy quantifiers, A:fuzzy subject, m,m',m":modifiers, F:fuzzy predicate, tau:truth qualifier). We show that Q' is resolved step by step for non-increasing type(few,...) and a non-decreasing type (most,...).

View full abstract

Fuzzy logic, which began tafter he paper "Fuzzy sets"(Zadeh) was published, was gone, by crticisms by Morgan et al., and Haack and so on, to somewhere. But to where?

View full abstract

We have many kinds of data of time series such as stock prices. We have proposed a method to express a global trend of them in a natural language via representative values on the fuzzy intervals in the temporal axis and local features via the position of large difference between the original data and the data generated from the global trends. In the previous research we used the fixed fuzzy intervals for various data of time series. However, humans usually tune them with local features for every data of time series. In this paper, we propose a method to tune the fuzzy intervals in the temporal axis by reducing difference between the original data and the data generated from the global trend. As a result, the fuzzy intervals have been tuned by local features, with the global trend being expressed in more appropriate word.

View full abstract

In this paper, we propose a method for classification and retrieval of news contents based on automatically extracted tags. A criteria of tagging is set based on the result of a subjective experiment. A tag crowd is used for visualization of the tags, in which an important tag is displayed with large size. The level of importance of the tag is calcurated by the frequency and time of its appearance. As a result of evaluation of the method, the contents of whole news site are easily comprehended by the tag crowds, however, the tag crowds depends on the news source. The accuracy of retrieval is evaluated as precise.

View full abstract

A metaphor typically takes a form "A is B". This study researches the metaphors formed as above and propose a model which understand a metaphoric expression based on relevance theory. In relevance theory a metaphor is paraphrased by utterances, assumptions and implicatures.And an implicature is made from utterances, assumptions of a listener and their deductions. In this study, first we extracted knowledge about the utterance from a news corpus and generated assumptions. Next we obtained implicatures operating the utterances and the assumptions. Third we chose the best paraphrase of the metaphors calculating the relevancies of implicatures and the utterances. We used a co-occurrence degree as a relevance degree. Forth we evaluated the proposed model simulating some metaphors.

View full abstract

A conceptual fuzzy set (CFS) expresses the meaning of a word in the distribution of activations of other words. It was proposed attempting to express word ambiguity influenced by varying contexts. In English sentences a generation method has been proposed paying attention to word sequences just before the target word. First, we predict a next word of word sequences just before the target word by using the modified confabulation model and obtain a synonymy. Second, we generate a CFS taking weighted summation of the synonymies reflecting contextual information. But in Japanese sentences the meaning of a word is determined by modification structures rather than a word order and it is difficult to presume the meaning of a target word using a word sequence. In this study we propose a method to generate a CFS using a modification structure in Japanese sentences. As a result of experiments, we show that the proposed method is effective to generate a CFS reflecting contextual changes.

View full abstract

We propose a method to compare and align concepts in different hierarchically structured conceptual systems that are provided as parts of electronic dictionaries. We define the similarity of two concepts in terms of word sets associated with them, and present an effective algorithm to obtain the most similar concept set to a given concept. Using this method, we attempt to align the Goi-Taikei semantic system with the EDR conceptual system, and demonstrate its validity. We also discuss applications of such a conceptual alignment to semantic analysis of sentences.

View full abstract

In this paper, we discuss a method to classify blog documents into some categories by extracting common topics from various blog documents and calculating the similarity of the topic. From blog documents about a specific topic, we extract specific topic words, which are common topic words among them. We classify blog documents by extracting topic words of them and measuring the similarity with the specific topic words. In addition, by comparison with the classification by human, we evaluate the effectiveness of this method.

View full abstract

Recently as digital cameras and web cameras have been commonly used in our everyday lives, we can easily obtain quite a few movies. However, it is difficult to fine a particular part of the obtained movies. From this, in this paper we apply image processing for the obtained movies and then propose a method to explain human's behavior in a particular space by observing how a human beings interacts the objects in the space. By this, we aim to develop a system that enables us to retrieve a particular human behavior by words.

View full abstract

In conventional studies, it is considered that differences of actuators are not important for realizing intelligent behavior, so, conventional robots employ motors as actuators.However, animals have muscles, and recently, it is reported that physical properties of the muscles like viscoelasticity play an important role in realizing intelligent behavior.In this paper, we consider that learning time (time required for adapting themselves to the environment) works as selection pressure, and actuators like muscles are acquired in evolution.To discuss this hypothesis, we employ a manipulator, and evolve the manipulator in simulation. Fitness of the manipulator is calculated from learning time. Simulations have been conducted, and as a result, manipulators that have actuators with viscoelasticity have been obtained. By analyzing the result, we found that the body image of the manipulator has consisted of the viscoelasticity, and the body image has reduced learning time.

View full abstract

In human learning, we memorize the given data as it is in the begining and we reach to find general knowledge from the data. This means that knowledge representations are switched. We proposed and simulated a learning method that switches knowledge representation. However, human never always switches knowledge representation in the same fashion. Our emotions influence our learning. In this paper, we introduce an emotion model to change parameters for the learning. Russell's circumplex model classifies emotions into four elementary emotions, joy, anger, sadness and relaxation. The degrees of elementary emotions are calculated by using fuzzy rules from the frequencies of reconstructing knowledge-set and switching knowledge representation, the incorrect answer ratio and the number of rules. They increase and decrease for the threshold of incorrect answer ratio of parameters for switching knowledge representation, reconstructing knowledge-set and switching reasoning method.

View full abstract

In order to realize intelligent agent such as autonomous mobile robots, Reinforcement Learning is one of necessary techniques in behavior control system. However, applying the reinforcement learning to actual sized problem, the "curse of dimensionality" problem in partition of sensory states should be avoided maintaining computational efficiency. In multi-agent reinforcement learning, the problem is emerged owing to the high dimensionality of each agent states. In this study, we evaluate the learning performance of agent that represents the input states as relative expressions through numerical experiments of pursuit problem.

View full abstract

In order to realize intelligent agent such as autonomous mobile robots, Reinforcement Learning is one of the necessary techniques in control system. Fuzzy Q-learning is one of the promising approaches for implementation of reinforcement learning function from high ability of model representation. However, in applying fuzzy Q-learning to actual application, the number of iterations for learning also becomes huge as well as almost all Q-learning application. Furthermore convergence performance is often deteriorated owing to its complicated model structure. In this study, implementation method of fuzzy Q-learning is discussed in order to improve the learning performance of fuzzy Q-learning. The modular fuzzy model construction method based on fuzzy Q-learning is proposed in this paper. Through numerical experiments of mountain car task and acrobot task, we found that the proposed construction of modular fuzzy model improved the performance of fuzzy Q-learning.

View full abstract

This paper describes the concept of "Emotional Fitness" which is an exercise for mental and mind health care and proposes the optimum exercise intensity for Emotional Fitness. First, the heart beat stability (HBS) after the exercise is defined by the coherence of fluctuation of heart beat rhythm. Secondly, the optimum exercise intensity which makes HBS maximum is shown through treadmill exercise experiments. Finally, the correlativity between HBS and the hedonic tone measured by the two dimensional mood scale is discussed and the validity of the optimum exercise intensity for Emotional Fitness is shown.

View full abstract

Estimation of a caregiver's view is one of the most important capabilities for a child to understand the behavior demonstrated by the caregiver, that is, to infer the intention of behavior and/or to learn the observed behavior efficiently. We hypothesize that the child develops this ability in the same way as behavior learning motivated by an intrinsic reward, that is, he/she updates the model of the estimated view of his/her own during the behavior imitated from the observation of the behavior demonstrated by the caregiver based on minimizing the estimation error of the reward during the behavior. From this view, this paper shows a method for acquiring such a capability based on a value system from which values can be obtained by reinforcement learning. Experiments with simple humanoid robots show the validity of the method, and the developmental process parallel to young children's estimation of its own view during the imitation of the observed behavior of the caregiver is discussed.

View full abstract

Both self-learning architecture(embedded structure) and explicit/implicit teaching from other agents(environmental design issue) are necessary not only for one-shot behavior learning but more seriously for life-time behavior learning. This paper presents a method for a robot to understand unfamiliar behaviors shown by others through the collaboration between behavior acquisition(reinforcement learning) and recognition of observed behaviors, where the state value has an important role not simply for behavior acquisition but also for behavior recognition(observation). That is, the state value updates can be accelerated by observation without real trials and errors while the learned values enrich the recognition system since it is based on estimation of the state value of the observed behavior.The validity of the proposed method is shown by applying it to a dynamic environment where a human and two robots play soccer.

View full abstract

The purpose of this study is to extract kansei rules that connect human kansei with design elements from the questionnaire data of bouquet flowers. In this paper, three data having class with the grade of certainty are generated by the questionnaire data in each sample. These data are analyzed using the fuzzy C4.5. We analyzed data of 20 Adjective Pairs, extrated kansei rules. Also, we conducted the verification experiment, and confirmed the effectiveness of kansei rules.

View full abstract

This paper proposes a method that generates utterances to estimate the degree of coincidence between the shared beliefs assumed by a user and a robot in an object manipulation task. In the proposed method, a belief system of the robot consists of two parts: a shared belief function and a global confidence function. The shared belief function expresses the shared belief assumed by the robot for motion, object images, speech, and other task-specific features. The global confidence function represents the degree of coincidence between the shared beliefs assumed by the user and the robot, outputs the predictive probability that each other's utterances are understood correctly. This function is learned incrementally online through human-robot interaction with objects. By using it, the robot can generate utterances which the user can understand correctly with target probabilities. The experimental results revealed that the robot and the user could adapt their behaviors through the mutual inference of the shared beliefs assumed by them.

View full abstract

In this research, we propose a method to acquire operation characteristics in the control strategy by an expert operator. In this paper, we use CMAC learning algorithm which is one method of the neural network. The ordinary CMAC learning algorithm is able to learn multiple inputs and one output control mapping, but the control output to learn generally becomes plural in case of human operator's control characteristics. Therefore, in this method, we propose the modified CMAC learning algorithm to learn three operation parameters when a human operator controls an omni-direction mobile robot.

View full abstract

In this laboratory, we have already proposed the fuzzy state division type reinforcement learning to acquire high-level behavior selection rules of the hierarchical fuzzy behavior control by reinforcement learning automatically. In this research, we propose a method which we make learn to the robot efficiently by giving appropriate rewards by human. Furthermore, we define the weighted value of attack and defense for the robot by using the fuzzy rule and incorporate the method to let it learn the behavior strategy according to the situation. We performed the comparison experiment with the conventional reinforcement learning method by the soccer robot simulation.

View full abstract

In the field of the pet robot and robot therapy, the creatural motion is important for the robots imitated the form of various animals. This paper presents the generation method of animal gait for quadrupedal robot. Here, we have employed AIBO as experimental quadrupedal robot and created the gait of AIBO in imitation of animal gait. At first, we have optimized the orbit of mono-leg, which can output propulsive force efficiently, by imitating dog gait and Genetic Algorithm. Moreover, we have generated the quadrupedal gait of AIBO using both the optimum orbit of mono-leg and "Trot" gait, which is classified the gait of walking dog based on the zoology.

View full abstract

Skulltronics is a novel system design paradigm that require skills upon the users. Puzzles which need skills and body performances to solve them can be called as a skulltronics game. This paper shows how to make skulltronics puzzles and how to solve them.

View full abstract

In recent years, many researches trying to make relation between multimedia and information of user's impression using the soft computing methods has been developed. But most of these researches aimed at developing multimedia search system, classify things into groups, and so on. In this research, we developed automatic pattern generation system based on user impression and suggest to the new application of the soft computing methods.

View full abstract

This paper presents a librarian robot that has sleep and wake functions. This robot is equipped with a laser range finder for predicting library users' behaviors, a microphone for conversation with them and a stereo vision. A lot of processes run in parallel in this system, movements of these processes are controlled by our proposed mathematical Activation-Input-Modulation (AIM) model, that can express consciousness states, such as wake or sleep based on stimuli detected by the external sensors. In a waking state, sensory information with stimuli is mainly processed. In a sleep state, most processing is paused, or information stored in memories during waking is mainly processed. Moreover, this system has two kinds of memories. One is a memory stored when external stimuli are detected, the other is a memory stored when no stimulus is detected. Both dynamic and gradual changes of sensory information can be stored by these functions.

View full abstract

Most conventional image retrieval systems use the title and keywords added to image. In an image retrieval system Feelimage developed by us, we use color distributions and exif information such as date and time, where we retrieve them with fuzzy sets. All fuzzy sets in our system are fixed, but fuzzy sets, for example, "evening" in our life are not fixed. This is the reason why the sunset time varies every day. In other words the "evening" is defined by the attribute "date". In this paper, the fuzzy set "evening" is defined by the sunset time calculated from a date and a place. In addition, we define various time zones such as "morning" and "midnight" as fuzzy sets by the sunrise and sunset times. A retrieval experiment in 102 landscape images shows the improvement of the quality of the retrieval.

View full abstract

We investigated preferences for 20 landscape images. Because the paired comparison for many targets is too effortful for subjects, we adopted the four-fold choice. We asked subjects to choose the most preferable image from the four images displayed on the Web site, and they repeated this selection process 20 times. Then, assuming the Thurstone model, we estimated the average preference scores of all images with the MCMC estimation. Then, we assumed subjects can be divided into different groups according to their preference patterns. We assigned the subject to groups based on the likelihood and estimated the average preference scores for the 20 images for each group simultaneously. Some of the subjects were assigned on two groups. In addition, we simulated our experiment. Supposing that the preference scores for each image and the assignment of subjects to the groups were given, we provided artificial answers and analyzed them by the same procedure described above. As a result, we found that the experiment can estimate each group's preference well.

View full abstract

Fuzzy modeling is an effective technique as the modeling technique of the nonlinearity system. The part where the nonlinearity is strong automatically increases the number of rules in this text according to the degree of the nonlinearity of the object system, and a weak part proposes the fuzzy modeling technique of decreasing the number of rules. The proposal technique repeats that fuzzy modeling is done by dividing the input space of the object system and the model centering on the maximum error margin point of the object system and the obtained model in each division space. The technique for proposing it than the number of fuzzy rules roughly generated one after another this time was obtained for the modeling object that had been experimented as a result this time and the modeling result of the accuracy of this level was obtained by a less number of rules.

View full abstract