In recent years, the opportunity that users receive information from information systems has been increasing. The timing, frequency and presentment of interruption have not been controlled in most of the systems. In this paper, we proposed the interaction start method which reduces the negative effect of work efficiency of the interrupted user by mediating an interruption request from outside, users and information systems, for developing the secretary agent that supports acceptable interaction start. The agent estimates the user uninterruptibility from PC operation records, then controls the timing of interruption based on it. Moreover, the agent uses the ambient appeal using the joint attention and mutual gaze animations of CG character for presenting the interaction request. We confirmed the usefulness of the both methods and a reduction effect of negative influence for the work efficiency by interruptions.

View full abstract

Some researches in the field of human symbiotic system and human-agent interaction have been actively performed. It becomes a background that the flexible understanding ability for human intent and the expressing ability of robot intension are required for the robot to live together with human. However the technology for realizing bidirectional communication between a human and a robot has not been established yet, therefore there are very few robots which are able to communicate to human smoothly. Therefore, the interactive emotion communication (IEC) -bidirectional interaction based on emotional behavior between a human and a robot has been proposed. We aim to give high interpersonal affinity to the robot for the amplification or relief of human emotion though the interaction by communicating emotional behavior mutually. In this research, we proposed the generation method of various robot emotional motions by using self-organizing maps because a robot emotional behavior has been defined uniformly until experiments. As a result, we were able to realize more complex interaction experiment which a robot expressed the mixed emotion by plural emotions. Next, we confirmed how human emotion changes by the interaction experiments with generating mixed emotional behavior. In consequence, more complex interaction as compared with the ordinary method was realized and the robot generating mixed emotional behavior became to grow the interpersonal affinity.

View full abstract

An internet service of video sharing using pseudo synchronization attracts the attention. As a typical example, there is Nico Nico Douga (Nico Video) that comments, which were submitted to a scene in a video by the Nico Video users, are superimposed to the same scene in the video played back by a viewer. The viewer watching the video with the superimposed comments feels as if the viewer shares emotion in the scene with the users even though the viewer watches the video alone. This paper describes an influence of the user's comments to the viewer's emotion and shows that the video sharing using pseudo synchronization causes the upsurge of sentiment of the viewer in a certain kind of emotion. First, five videos are selected that cause emotions “Joy”, “Anger”, “Sadness” and “Relaxation” of Russel's model. Next, the videos are submitted to Nico Video and comments superimposed by Nico Video users are collected and analyzed, then typical comments expressing the emotions are selected. Furthermore, new videos superimposing the typical comments are produced and an experiment is conducted to measure emotion strength while the viewer is watching the new videos and the original (no comments) videos. Finally, relation between the existence of comments and feeling upsurge of viewer are discussed and the experimental results are verified based on the psychology on emotions.

View full abstract

In this paper, we propose the music generation system equipped with a model of human cognitive biases. A musical piece is a construct formed by satisfaction of and deviance to the listener' s expectation. In order to realize the dynamics of expectation, we adopt the loosely symmetric (LS) model that is a kind of distorted conditional probability model based on illogical cognitive biases including the symmetry and the mutual exclusivity bias, proper to thinking and reasoning of human beings. Our system regards features of melodies as the simplest sequence of transitional events from preceding notes to succeeding notes, modifies and generalizes a given melody like human through LS model, and creates new melodies. The average information of the probability distribution to generate new melodies showed that the LS model creates a proper amount of complexity. We conducted an evaluation experiment with more than one hundred participants and found that the melodies generated by LS have a good balance between fulfillment (musical unity) and deviance (unpredictable quality) to expectation. The result indicates the possibility to generate expectation and more human-like melodies by application of human cognitive biases.

View full abstract

In this research, an index of the relaxation degree for each user is obtained by the electroencephalogram information, and a system with the automatic tuning function of the sound is developed. We construct beforehand the index for measuring the relaxation degree using the electroencephalogram information. At first, the change of electroencephalogram power content ratio for θ, α, and β wave's frequency band are displayed in the three-dimensional state diagram, and we analyze the electroencephalogram feature. Based on the hypothesis about the relationship between the electroencephalogram and the relaxation, the relaxation tendency is grasped in the three-dimensional state diagram. Then, we construct an index which can be measured quantitatively that a subject is in the relaxation state. As a result of experiment for six subjects, we confirm an efficiency of the relaxation degree for the change of electroencephalogram of the subject. We have already developed the Interactive Chaotic Amusement System (ICAS) that generates the sound deciding the musical note, duration and volume by using network of chaotic elements that is able to control a synchronous and an asynchronous behavior. We aim to construct the system to tune the parameters of ICAS automatically from the generated sounds to the relaxation sound. We make six subjects use the proposed sound generation system and compare the result of relaxation effect estimated by this system with the questionnaire result after subjects listen to the generated sounds. As a result, it is confirmed that the results of subject's questionnaire and the index of the proposed method are correlated.

View full abstract

For practical use of Brain-Computer Interface (BCI), it is safe and suitable that electroencephalography (EEG), which is noninvasively without inserting an electrode in an encephalon. In previous study, a specific change in the feature of EEG signal accompanying cognitive phenomenon has been selected as the target and the BCI system has been designed to detect the specific change. However, EEG is far from stable and hardly reproducible even among recorded wave from an individual. In addition, same cognitive task induces not always same effect on EEG signal in everybody. In this study, we abolished that premise assumptionabout the target feature of EEG. Instead of that, we have developed a BCI system with EEG feature search unit and designed the system to detect the most obvious change accompanying the cognitive phenomenon. This method enables us to select the most suitable EEG signal to control devices. This BCI system succeeded in automatic exploring for every EEG recording sites and every frequency bands of the EEG signal and it finds for a specific site and a frequency band with remarkable changes of EEG signal, accompanying cognitive task.

View full abstract

Behavioral biometrics is one of the most promising authentication technology in terms of various application scenes such as human-computer interaction and context-aware computing. Especially, human gait is one of the most popular trait among behavioral biometrics. Recently front-viewed gait verification has been becoming important technology because it has some advantages; human-compatible ability to identify his/her acquaintance and less installation condition of camera. It is significant to develop the technology with effective use of the advantage of front view gait so that we can develop an interactive robot with person authentication ability and a personal adaptive system with high security. In this study we develop a novel person verification method by soft biometric features observed from gait, face and body in frontal view. The facial and the body measurements, head movement time series derived from gait are independently evaluated by DP matching and Euclidean distance, and they are integrated by AND/OR calculation or fuzzy inference. As the result of verification experiment with seven subjects data, 0% false acceptance error in average can be achieved when false rejection error is 0%.

View full abstract

Reinforcement learning is a technique developed for a single agent. If it's used for the cooperative behavior in multi-agent environment, one of the main problems is how to benefit from mutual interaction during the learning process. In this research, we propose an interactive learning system with cooperative ability through the interaction of reinforcement value among agents. In this method, when each agent repeats trial and error, the confidence degree between other agents could be generated and updated based on the degree of goal achievement and cooperation. The adoption strategy of reinforcement value is determined through the confidence degree. Each agent is able to adopt reinforcement value of others, and an interactive learning system can be built among agents. Therefore, each agent could learn the available experience from others. The cooperative behavior and group strategy of multi-agent is also learned through the interaction with environment and other agents.

View full abstract

Impression evaluation of robot behaviors is important to enhance the effectiveness of robot-assisted therapy. Here, we discuss impression evaluation for mixed walking motion of a quadrupedal animal-type robot, AIBO. To verify the effects of different behaviors on human sensibility evaluation, we performed a comparison between the normal pre-installed gait and an animal-like gait generated by mimicking canine walking, with optimization of the path of the legs using a genetic algorithm and application of a zoological basis. Moreover, we analyzed the relationship between behaviors and impression by evaluating three mixed walking motions.

View full abstract

The purpose of this study is to extract kansei rules from the kansei data of bouquet flowers, and to apply them to bouquet flower systems. These rules are easy to understand for humans, and connect human kansei with design elements of bouquet flowers. However, in this study, it is difficult to analysis for kansei data of bouquet flower using the quantification theory type I, because those data are included for attributes in a relation of linearly dependence. In this paper, kansei data of bouquet flowers having class with the grade of certainty was created, and the data has been analyzed using the C4.5 with certainty factor for class classification. Also, we tried to extract the kansei rules of bouquet flowers easy to understand for humans from generated decision trees, and to apply there rules to the bouquet flowers display system. As a result, obtained rules could be interpreted to fit human kansei, and showed some degree of effectiveness. In addition, from evaluation experiments of system using obtained kansei rules, we showed that subjects could select bouquet flower images to match his/her kansei.

View full abstract

“Shodo,” which is Japanese calligraphy, is a form of artistic writing, used for writing Japanese language. It is the oriental beauty art similar to water color and oil painting. This study aims to propose and develop a method to design Japanese calligraphic fonts which contain artistic representations similar to actual “shodo” art. We especially focus on “kasure (in Japanese),” which is one of the most characteristic visual effects of “shodo” and adopt an approach using example based texture-synthesis in the field of non-photorealistic rendering to achieve high quality artistic fonts. Our proposed method doesn't require any technical expertise and high level operation skill to design “kasure” of Japanese calligraphic fonts.

View full abstract

The authors examine the impressions people gain from reading newspaper articles and propose a method for quantifying the strength of these impressions. Our target impressions are limited to those represented by three bipolar scales, “Happy-Sad,” “Glad-Angry,” and “Peaceful-Strained.” In order to compute the strength of each impression as an “impression value,” that is, a real number between 1 and 7, it is generally required to quantify the power of features extracted from articles to influence on their impressions and record it in an impression lexicon. An impression lexicon is usually constructed for each kind of impression and is used only to compute an impression value for an impression. In this study, we focus on the fact that the impressions are not independent of one another and adopt a new approach that recalculates each value with the values that were computed using each impression lexicon. First, we define “word unigrams” and “word bigrams” as features extracted from articles and construct a total of six impression lexicons (i.e., one lexicon for each of the three kinds of impressions multiplied by two kinds of features) from a newspaper database. Next, we apply multiple regression analysis for each impression, where the values computed from articles using each lexicon are used as one of the explanatory variables, and an average of the values that respondents used to rate each article using the corresponding impression scale in questionnaire surveys is used as the objective variable. Consequently, a multiple regression equation is obtained for each impression, which represents a correspondence relationship between the variables. We also perform five-fold cross-validation using the data obtained in the surveys to verify the effectiveness of the proposed method. The results show that the average root-mean-square errors for unlearned data are 0.68, 0.57, and 0.62 for respective impressions.

View full abstract

A model is described that estimates object posture by using a number of ellipses. First, the ellipses are rotated and arrayed in the annular direction. Next, the ellipses that change long-and-short axis ratio are arrayed in the radial direction. Finally, a radial-ring network model is built by connecting each of these ellipses to form a concentric network. A variable template network (VTN) model is based on the same concept: an ellipsoid is rotated, enlarged and reduced, and then several a 2D appearance figure is projected onto each 2D plane surface. The VTN model is constructed by connecting these figures to form a grid. Although object postures can be estimated using this model, there are multiple figures with the same rotation angle and form in the VTN model. Similarly, since the discrete degree at the time of projection is constant, if the skewness of a 3D object is high, this can not estimate its posture accurately. The proposed model does not suffer these problems, enabling it to estimate object position and posture quickly and accurately.

View full abstract