Toru Yamaguchi

A cloud based chat robot using dialogue histories for elderly people

The authors developed cloud based chat robot for exhilarating elderly people. Our robot is designed to keep users from being bored. Conventional chat robot systems use huge phrase database in advance. However, fixed phrase involves an immense amount of time and effort to make. The proposed system uses dialogue histories which are collected from users. This paper reports the effect of updating the phrase database by using histories.

Influence of approaching patterns of Telepresence Robot for personal space

The robot called “Telepresence Robot” is on business in recent years. However, the robot does not consider users personal space when the robot approaches the user. To better develop the system of the robot, here we study about how the robot approaches a user. The three factors was considered to influence the scope of the users personal space. The first factor is the position where the robot starts to decelerate when the robot approaches a user. The second factor is the method the robot decelerate. The final factor is pre-action by sound when the robot approaches a user. We verified those 3 factors, as a result, it was revealed that pre-action influenced personal space. By the appropriate approaching, the conversation between Telepresence Robot and human becomes natural and interactive.

Exercise Support Robotic System by Using Motion Detection

In this research, we propose an exercise support robotic system. The system collects human daily life activity data by motion detection as a part of big data, and analyzes it for providing better service. The Normal exercise tools such as pedometer need to be carried while exercising. However, sometimes people forget to take such exercise sensor, when it happened many times, people will feel hard to continue exercise regularly. Thus, the proposed system focuses on user’s behavior when exercise by Kinect, and based on it to recommends user a suitable exercise pattern through an interaction robot. Firstly, this paper discusses the user’s exercise habits and their weariness in exercise, secondly, describes the proposed system approach, the method of detecting human motion, and maintaining exercise motivation. Finally, the experiment and the results are shown.

Effectiveness of dialog contents for obtaining personal attribute

It looks easy to make human relationship, but it is difficult because of various factors. Moreover, it is difficult to continue a dialog with those who have never met before. Dialogs arent continued without pleasant topics. However, if pleasant topics are understood, dialogs are continued and he/she wants to talk about himself/herself. It is supposed to be applied to dialog with robot as well. If robots understand users favorite topics and make a fit dialog for user, user may talk with robot aggressively. However, existing dialog systems have not encouraged users to talk. We develop a chat system in order to resolve the difficulty. The experiment was carried out to evaluate the system. It is important that dialog systems need to change dialogs in order to let users to talk. This paper describes chat system with robot to obtain users information.

Acquisition of the human characteristics through dialogue with a robot at home

Users life-logs include various and useful information. Life-logs can be used for health care and improve daily living recasting his/her daily life. Our aim is monitoring a users life for acquiring users life-logs. In previous research, life-logs is acquired using users GPS data or estimating daily activities. These methods are useful for estimation of what he/she did or where he/she went, however it is difficult to estimate users likes and tastes. Therefore, this paper focuses on the dialogue with the robot. In this research, the authors experiment to what feature can be acquired through dialogue with the robot for the acquisition of the users likes and tastes.

Attendant robot for elderly service using sensor network system

The authors developed an Attendant Robot system for the elderly. The robot system is used to prevent user from being at a loss. The system uses ZigBee-3D acceleration sensor and Kinect sensor. Attendant Robot mounts the ZigBee access point, and the ZigBee accelerometer tag is attached to users. The ZigBee access point knows that user comes near to Attendant Robot by finding the tag. This tag has a unique ID, so Attendant Robot reacts to correlative ID only. Kinect sensor detects people who stand in front of the Attendant Robot. This system recognizes user by using function of Kinect and ZigBee sensor. Attendant Robot navigates user using this system. In this paper, the authors set up the proposed system and performed experiment.

Wearable finger motion input interface system with GMM foreground segmentation

A wearable fingertip based input interface software has been developed. High speed detection of fingertip position for maintaining real-time software capability, stable finger motion analysis as well as reliable direction pointing have been added to the real-time JAVA based foreground segmentation software. The developed interface system can be run on a single microprocessor with a glasses-type single camera.

Software implementation approach for fingertip detection based on color multi-layer GMM

Prototype fingertip detection software has been developed for see-thru type head mount displays. The employed algorithm is newly introduced multi-resolution GMM (Gaussian Mixture model) foreground segmentation for fingertip extraction. The fingertip detection is realized on successive foreground/background binary images. The algorithm is extended to remove reflection areas as well as those of shadows by a nearby room light for indoor scenery. The employed algorithm can understand smart phone like control motions such as clicks and scrolls.

An Effectiveness of Model-Based Development with User Model in Consideration of Human

In this paper, we proved an effectiveness of a model-based development method with user model which can operate a product as an agency of human in the model-based simulation. In this method, the user model is generated on the basis of the user profile which is defined to clarify the features of the product’s user. This method is effective to evaluate the products which is used by various users with different features, and is also effective to evaluate the products with the risks of damaging the human. And this method does not require any hardware sample production for evaluation. In order to investigate the effectiveness of this method by comparing the results between the simulations and the experiments, we applied this method to the development of an approach function and an emergency stop function of a service robot.

A possibility of the power supply leveling by using electric vehicles as storage batteries

After the big earthquake on March 11, 2011 at Fukushima, there are not enough power supply sources in Japan since almost all nuclear power plants stopped for checking each safety level, and no alternative power supply source is available. Then there is a risk that the amount of electric power demand would exceed the maximum amount of electric power supply. Once the power demand would exceed the power supply, a widespread blackout would occur. To avoid this kind of blackout, Japanese citizens are required to reduce the electric power demand during the peak power demand hours. But this means that if the peak of the power demand can be cut down, there is no reason to reduce the power demand. To cut down the peak of the power demand, we need to change the point of view from the increase of power supply and/or the decrease of power demand to the power supply leveling. Then we decided to study a possibility of the power supply leveling with the storage batteries. But since the storage battery is not popular as a home electric appliance, then we started to study it with electric vehicles (EVs) as storage batteries. This study must depend on behaviors of each family and each family has a different life style. Then in this study, we chose the approach by the simulation with user model.