RAA (Robot Assisted Activities) allow people to feel joy and pleasure and recover through contact with a robot. Considerable research has reported that RAA is useful for elderly persons, children hospitalized in pediatric wards and so on. We have developed a reaction time measurement system with hardware and software as evaluation method of RAA. In this paper, we reported the results of comparisons featuring such measurement systems and the basic results of the RAA evaluation experiment using reaction time measurement.

View full abstract

Recently various robots spread in the home, and, in the future, familiar partner robots seems to appear in human life. Therefore, it is necessary that robots learn desirable behavior in order to establish a life of the cooperation between robots and human. In this paper, we propose the methods using interactive GAs so that robots acquire the behavior which the human likes. Also, we apply this method to the pet robot AIBO.

View full abstract

Sensor fusion aims at obtaining information, which cannot be obtained by single sensor, by combining signals from multiple sensors. The main problem of sensor fusion is huge computational cost increasing exponentially with the number of sensors since the combination number of associations between signals and sensors is large. A general method for this problem is proposed in a form of nonlinear state space model to deal with the unknown associations. The target states and the associations are simultaneously estimated through the state estimation. In the estimation of the associations, we apply particle filters with clever proposal. The associations are estimated in probabilistic way to avoid being trapped by a bad solution. We adapt the method to a specific situation of omni-directional camera and two microphones, to a sound target.

View full abstract

Green caterpillar type robots have been studied widely to clarify their motion patterns. These robots were also applied to inspection medical treatment robots, maintenane robots or exploration robots. However, the method of using a single movement pattern is not the best strategy from a viewpoint of the movement efficiency when the environment with a different diameter in a duct exists variously. In this study, we consider a semi-looper type robot whose motion pattern can adapt to the change of situation.

View full abstract

This paper presents a new path generation method for a mobile robot equipped with an omni-directional camera and a laser rangefinder. The omni-directional camera can be used for environmental recognition, and a can capture 360-degree image without a dead angle. The laser rangefinder is capable of detecting range profiles that can accurately identify the distance of obstacles. This information can be used for smooth path generation. Here, in order to achieve smooth path generation and/or tracking, we applied omni-directional image and laser range profile data to generate an appropriate path by using a clothoidal curve. The curvature and its derivative are estimated by segmented Hough transform, and path points are generated based on the Delaunay triangulation method. The path generated by Clothoid minimizes the effect of lateral force to the passenger. The validity of the proposed method is verified by simulations.

View full abstract

This paper describes how to realize a real-time 3D positioning for outdoor mobile robot based on multi-rate sensor fusion technique. In order to achieve accurate 3D positioning to compensate vertical resolution as well as to identify the direction of the mobile robot, The cascade type Kalman filer can enable to reduce significant computational task and can enable to estimate a real-time 3D positioning and its direction.

View full abstract

Recently, the number of accidents for an electric four-wheeled vehicle increases. User's driving skill and knowledge shortage are enumerated as the cause. Especially, it is difficult to judge other traffic participants' movement and to decide the content of the evasive action. Up to the present, we had developed an automatic driving system that can correspond also to a dynamic obstacle by building the driving algorithm based on the predictive fuzzy control into the controller, and adding the driving knowledge to human safety. In this paper, the pavement is assumed as an actual life environment, and the behavior of the automatic driving system was confirmed by the simulation.

View full abstract

This paper describes the sensor configuration and dynamic robust tracking algorithm for a mobile robot capable of following a walking human in 2D space. Since the back shape of a walking human is not constant as that of a vehicle would be, it is difficult to identify the human solely by vision sensor and image signal processing. Here, we employ a sensor fusion technique and combine an omni-directional camera and a laser rangefinder to solve this problem and achieve stable tracking and following ability.An omni-directional camera is used for robust image acquisition and processing, and a 2D laser rangefinder is used for environmental recognition by detecting obstacles. We present the advanced sensor fusion technique of the omni-directional camera and laser rangefinder, which yields dynamic range-adjusting tracking, avoidance of collision and stable human tracking. The Kalman filter based on multi-rate sensor fusion theory is employed to make the system functional. Human tracking necessarily involves the problem of short-cut passing of human following trajectory. In order to avoid the short-cut problem, we adopted a path-tracking algorithm to generate a trajectory as similar as possible to that of a walking human. The path-tracking algorithm calculates the trajectory to be passed from the target position and the direction, and a control signal is generated based on the calculated trajectory.

View full abstract

We propose a new method which achieves fast learning of the SOM having a competitive layer with graphical structure. The calculation of distance in the competitive layer with graphical structure is a ultimate factor which prevents fast learning. In the proposed method, a Bacth Learning SOM and Dijkstra method are introduced.In addition, input vectors are applied in the heuristical way which contributes the fast learning. The effectiveness of the proposed method is verified by applying it to module placement of a gate array.

View full abstract

Recently, a lot of social systems have been computerized using Information Technology (IT). It is no exception in hospitals. Hospital support systems are also rapidly being computerized using such as Diagnosis Reservation System, Ordering System for Drug Prescriptions, Electronic Medical Record (EMR) and so on. In the immediate future, all of medical data will be stored to the database server in the hospital, and it is considered that complete paper-less and film-less system in hospitals become reality. These stored data, however, is not used effectively for analysis because of the great deal of the data and its variety --- image data, numerical data, text data and so on. Previously, knowledge discovery methods from numerical medical database were studied and discussed actively. Text mining method for medical data, however, has not been enough discussed. This study discusses knowledge discovery method from Electronic Medical Records using Self Organizing Map (SOM) and its possibility. As the first step of this research, this paper describes the outline of brief sentence mapping method using SOM from Incident Reports. This paper also discusses the keyword extraction method from EMR and the coding method of SOM.

View full abstract

The system that displays the health condition by using the blood test result of obtaining when examining physical condition has been developed by using the Self Organization map. It is developing as an application that moves to the native by using this algorithm by the Windows system made by the Microsoft company. As for the system under development, the function etc. to forecast the health condition from information on the plural of each user year were added and the user added the improvement to an easy-to-use way. As a result, the end user became easy-to-use.

View full abstract

SOM achieves a probability density distribution approximation of input vectors using weight vectors. Binary SOM in which each element of the input vectors and the weight vectors is represented by ``binary'' integer is same to the SOM with respect to the approximation. In this paper, we propose a new updating equation of the binary weight vectors based on significance of binary input vectors. In the proposed updating equation, the binary weight vectors with low significance factors are adjusted to the binary input vectors with high significance factors. This updating equation achieves a probability density distribution approximation of the significant binary input vectors. The modified Binary SOM employing the proposed updating equation is applied to GA as a reproduction operator in order to maintain a genetic diversity of population.

View full abstract

This paper describes a new algorithm of the Self-Organizing Relationship (SOR) network for an efficient digital hardware implementation. In SOR network, the weighted average takes heavy cost for calculation. To overcome this problem, we propose a fast calculation algorithm for the weighted average by using "active units" only. Beside this, we also propose a new generating technique of the membership functions by presenting its width on power-of-two to a given parameter, to suit well with digital hardware bit-shift process. The experimental results show that the proposed algorithm can reduce 93% of the calculation cost without declining the calculation accuracy.

View full abstract

A financial statement is a financial report that companies submit to the Ministry of Finance. The bankruptcy analysis that is one of the analysis for managements is very important. It is, however, very difficult for a user without knowledge to adopt indexes for a good analysis because financial statement includes many indexes. Furthermore, such user may not understand an analysis result. In this paper, we develop a financial-analysis tool in which indexes are automatically adopted and analysis result is represented visually using self-organizing map (SOM) and genetic algorithm (GA). The proposed method is applied to industrial classification, and its effectiveness is discussed.

View full abstract

A bankruptcy prediction is more important than a changing of the stock price for investors because they can avoid a falling risk of the stock price. There were past methods using discriminant analysis such as Altman's Z-score. However their methods had a problem that a precision is low when there are a difference in the number of the bankruptcy group and the existence group samples. This paper describes a construction of bankruptcy prediction system available even in such a case. First, the system select cash flow indices by factor analysis to treat the money changes. Second, the system makes SOM map and predict bankruptcies. Third, we use t-test, and verify prediction accuracy. Finally, we compare a result of our method with those of past methods and show usefulness of our method.

View full abstract

Self-Organizing Maps preserves neighborhood from input space to output space. In this paper, quantitative analysis of the preservation was done on the conventional rectangular SOM and the spherical SOM. Three types of measure were used for the evaluation and compared.

View full abstract

The development of Kansei evaluation model is presented for necktie design using Self-Organizing Maps (SOM). The feature of neckties was retrieved from the design and colors. The multidimensional feature vectors, including Kansei evaluation values, were input to SOM and the maps were formed. This model enables us to predict our feeling and emotion numerically when we look at the design of neckties. The comparison of the SOM method to the conventional one is discussed here and the usefulness of the SOM method is shown.

View full abstract

Combustion images in incinerators usually involve various noise elements in each of its images, since diverse wastes cause turbulent burning. This paper proposes a fuzzy membership approach basedon the squared Mahalanobis distances and the noise clustering (NC) by Dave for robustizing PCA to intra-sample outliers. The clustering prototypes of the images by the robust approach clearly represent the combustion statuses of an incinerator.

View full abstract

In order to develop a data mining system for huge database mainly composed of numerical attributes, there exists necessary process to decide valid quantization of the numerical attributes. Though the clustering algorithm can provide useful information for the quantization problem, it is difficult to formulate appropriate clusters for rule extraction in terms of cluster size and shape. In this paper, we propose a new method of fuzzy association rule extraction that can quantize the attribute by applying fuzzy clustering algorithm and extract rules simultaneously. From the results of numerical experiments using benchmark data, the method is found to be promised for actual applications.

View full abstract

In this paper, we propose the technique of clustering for motion patterns extracted from continuous input. Traditional techniques obtain clusters using the distance of each motions. Obtained clustering result tended not to be familiar for human sensitivity. In order to extract the rule for the motion categorization according to human sensitivity, we measured the human sensitivity to the motion appearance changed by position, scale, speed, and so on. Sensitivity parameters are applied to proposed clustering method. Representative motion patterns for corresponding basic motion clusters are generated according to the appearance trend. Experimental result shows the adequacy of proposed clustering method based on human sensitivity for motion appearance.

View full abstract

Recently, a lot of researches that focus on humans' subjectivity have been reported. In these researches, the SD (Semantic Differential) method is one of the most useful methods to quantify the humans' subjectivity. Some methods to compare individuals' impression using SD data have been proposed. This paper proposes the clustering method based on Orthogonal Procrustes Analysis (OPA). The proposed method can cluster the sets of data points based on the similarity of their structures. Therefore, this method can extract subjects among whom the distributed structures of the SD data are similar. This method enables us to compare the similarity/difference of humans' impression and analyze the relationships among the clusters. This paper applies the proposed method to the data group added arbitrary orthogonalization transformation as a preliminary experiment. This paper shows that the proposed method can form appropriate clusters to the objects with unknown shapes without giving prior knowledge about their shapes.

View full abstract

This paper presents switching fuzzy model construction with arbitrary switching planes and derive controller design conditions based on the switching Lyapunov function. In our previous papers, we proposed the switching fuzzy model with switching planes corresponding to quadrants. However, a switching fuzzy model differs with respect to switching planes. It is obvious that the model construction affects the stability analysis results. In this paper, we propose switching fuzzy model construction with arbitrary switching planes and derive controller design conditions based on the switching Lyapunov function in terms of LMIs.

View full abstract

In this paper, we consider new generalized delay-dependent stability conditions of Takagi-Sugeno fuzzy time-delay systems. In the literature, both delay-independent stability conditions and delay-dependent stability conditions for fuzzy time-delay systems have already been obtained. However, those conditions are rather conservative and do not guarantee wide stability regions. This is also true in case of the robust stability for uncertain fuzzy time-delay systems. Here we choose a generalized Lyapunov functional and introduce more free matrices to it, in order to obtain generalized delay-dependent stability conditions. These techniques lead to generalized and less conservative stability conditions. In fact, delay-dependent stability conditions thus obtained are given in linear matrix inequalities(LMIs) and guarantee a wide stability region. We take a simple example to illustrate our result. Comparison with other stability conditions in the literature shows our conditions are the most powerful ones to guarantee the widest stability region. Moreover, we consider the robust stability of fuzzy time-delay systems with uncertain parameters. Applying the same techniques made on the stability conditions, we obtain delay-dependent sufficient conditions for the robust stability of uncertain fuzzy systems.

View full abstract

Fuzzy system is applied to the anti-rolling and pitching device of a high-speed catamaran boat in order to improve the passenger comfort. A PID control was initially installed in the system, however no significant effects were observed for passengers although instrumental data showed good results. The fuzzy system has shown not only good instrumental data but also provided good comfort ride for passengers. Also, present experiment results demonstrate the effectiveness of the fuzzy system for the anti-rolling system of a high-speed catamaran boat, superseding the conventional PID Control.

View full abstract

This paper presents a switching fuzzy control technique based on virtual equilibrium convergence. We provide an iterative algorithm to automatically generate virtual equilibrium points and to design local stable controllers around the corresponding virtual equilibrium points. A key idea of this algorithm is to achieve a control system design guaranteeing a wider stabilization range by switching the local stable controllers. Simulation results demonstrate the utility of this algorithm.

View full abstract