We propose a new kind of architecture for physically-grounding agents that is an infrastructures for information services in ubiquitous environment. In the agent architecture, several kinds of agents work cooperatively, e.g., user-agent which serves as proxy of user, service agent, and spatio-temporal resource managing agent, by which software agents can trace users’ moving history, understand their intention and preference, and negotiate each other, with protecting users’ privacy using temporal identifiers. We also propose the concept of mass-user support and its example called mass-user navigation in theme park.

View full abstract

The target of ubiquitous computing environment is to support users to get necessary information and services in a situation-dependent form. In order to support users, we propose a location-based information support system by using Compact Battery-less Information Terminal (CoBIT). A CoBIT can communicate with the environmental system and with the user by only the energy supply from the environmental system and the user. The environmental system has functions to detect the terminal position and direction, to read and write information proper to the terminal, to communicate with terminals, to recognize the objects in the environment communicating with other computers in order to realize situated support.

View full abstract

An implementation of an indoor laser radar with higher system performance is described for the location-based information service environment. The system emits multiple eye-safe laser beams to My Buttons to provide simultaneous parallel information services.

View full abstract

In this paper, we propose a selflocalization method utilizing soccer robots. Generally selflocalization is one of most important functions for autonomous motion generation and cooperation in such robot soccer environment. Proposed based on Kalman filter technique utilizes to integrate position information using landmark and odmetry information. Dead-reckoning information and omni-directional visual information is utilized to confirm effect of proposed method using mobile robot.

View full abstract

In this paper, we propose a memory-based localization algorithm. This method estimates the position and the orientation of a robot by the matching process using azimuth information of observed landmarks and all possible sets of azimuth information calculated beforehand. All landmarks can be identical, and this method is effective even when it is difficult to identify which landmarks are detected.

View full abstract

In RoboCup environment, cooperative task realization among the robots is required. For realizing such task, it is necessary that each robot should acquire self-information and teammate information. However, it is very difficult that one robot gets accurate information of the other robots. Therefore, we propose a cooperative task execution system utilizing recognition ability of each robot and mutual communication among teammate robots. We had some experiments using real robots and confirm functional validity of our system.

View full abstract

We participated in the RoboCup Small-Size League with robots equipped with only onboard camera (local vision) taking up the position that decentralized autonomous robot system is more significant. However, there were many difficulties to use onboard camera. Therefore, we have developed a pseudo-local vision system which creates onboard camera information approximately from ceiling camera image, and replaces the image processing part in the local vision system. In this paper, we evaluate resolution, processing period, and time delay of this system compared with our local vision system.

View full abstract

When making autonomous robots, modeling of a robot’s action poses a probrem. To overcome this problem, we are trying to extract human behavioral intelligence and to transplant it to the robot’s program. As preparation for that, we developed an immersive remote control system by which a operator can get in the robot virtually at low cost. Omnidirectional camera image on the robot is transformed into perspective image and project onto the three screens around the operator. The operator can send commands to the robot by using a joystick.

View full abstract

Gifu Prefecture, Japan, has been endeavoring to produce a new robot industry in order to promote the development of industry. A humanoid robot “NAGARA” was developed under academic, business, and governmental circles in order to accumulate robot technology and know-how. In this paper, we describe NAGARA’s architecture (hardware and its walking algorithm).

View full abstract

UTTORI United is a joint team for the robot soccer competition RoboCup since 1997, which is composed of Utsunomiya University, Toyo University and RIKEN. This paper reports the system design philosophy of the robot system in UTTORI United, current implementation of the idea, and some considerations of some issues to be resolved.

View full abstract

The WinKIT participated in the middle size robot league of RoboCup 2002 Fukuoka/Busan, and we got the 2nd place. This paper describes the actual state and the problem of the WinKIT team in Kanazawa Institute of Technology.

View full abstract

This paper presents a new system design of RoboDragons for RoboCup small size league. Main issues are a system configuration, an image processing, robot agents and a strategy. Our team is placed in the RoboCup competition at Fukuoka Dome in this June.

View full abstract

The KU-Boxes team started up in 1997 to develop a standard platform for the Small Robot League of RoboCup. This paper presents recent improvements in KU-Boxes2002 and future research problems that we would like to study.

View full abstract

This paper describes about the Local vision system for the DNTC robot that is organized DENSO Technical College. This system realizes small size and high-speed image processing by using the CMOS image sensor camera. Moreover, image processing program adopted the most suitable algorithm for the local vision system.

View full abstract

[in Japanese]

View full abstract

We developed a soccer robot system based on RoboCup F-180 League in order to research cooperative behavior of mobile robots in the multi-agent robot system. The system consists of ten mobile robots, their remote-brain PCs, a communication server, and a vision server. Agent robots are made of LEGO MINDSTORMS^TM Robotics Invention System. Each of them has a programmable microcomputer unit called RCX with infrared communication.

View full abstract

The control to move in all the directions in the machine is becoming indispensable in the present RoboCup Small-size robot league. Our group manufactured this kind of machine at a low cost which is apt to be expensive. In this paper, we introduce this type of method. We also report the evaluation of the mobility of the all-direction-move-machine equipped with gear wheels.

View full abstract

We think creating a soccer robot team over several semesters provides an excellent teaching material for students in engineering courses. It not only covers themes from hardware to software, but also topics in system engineering materials. This paper reports an experiment whether RoboCup small robots can be used as undergraduate educations materials.

View full abstract

Department of Adaptive Machine Systems, Graduate School of Engineering, Osaka University, has been conducting education of creativity for master course students. The class provides several projects, each of which consists of two or three students and a researcher from a company. As one of these projects, we are working on the project named “Development of RoboCup Small Size League Robot”. This paper gives some introductions on the engineering education effects through this project.

View full abstract

As a tool to examine the role of gaze direction in human motor control, the authors built a measurement system which could estimate 3D gaze positions in a real-time manner, without restricting subjects’ head movements. The present report describes its structure and measurement/calibration algorithms, and shows some results of numerical simulation for estimating the expected error of the proposed system.

View full abstract

A new type of pressure-conductive rubber was developed. Its hysteresis characteristics was lower than former materials. As it was the paste type rubber, sensing elements can be made by dipping or painting method. By using this method, we can make the tactile sensor on the free form 3D object that is one of the important configurations of the sensor. In this report, the tactile sensor on the semi-sphere surface was made by dipping method was described and experimental result was shown.

View full abstract

In this paper a real-time and mega-pixel motion capture system based on intelligent scanning is developed, and the real-time tracking results in kHz order are shown. As an application system, a gesture robot system with high speed gesture recognition which can win JANKEN games at any time is introduced.

View full abstract

This paper discusses dynamic preshaping for a capturing robot driven by a single wire. To control the finger posture, there are three essential factors, wiring, spring distribution among joints, and mass distribution in each finger link. Giving the final link posture of finger, mass distribution, and tension with respect to time, we explore how to determine both the wiring and the spring distribution. We first show how to find them under a quasi-static motion, where all dynamic effects are neglected. We further explore an approach to find them under a dynamic motion where the tension increases fast enough.

View full abstract

We have developed the 100G capturing system whose maximum acceleration is two times larger than that of conventional robots. The robot can complete the capturing motion within 50ms. This paper discusses the experiment what the robot capture a ball dropping in the air with the assistance of 1ms vision system proposed by Ishikawa and his group. The vision system first detects the object and computes the trajectory of the ball. When the trajectory has an intersection of the capturing region of the robot, the computer sends the starting command to the robot with an appropriate time interval.

View full abstract

In this paper, a visual feedback control algorithm for robotic catching is proposed. The proposed algorithm is based on a nonlinear mapping of visual information to the desired trajectory. This method is applied to a catching task, and a reactive and flexible motion is obtained owing to real-time high-speed visual information. Experimental results on catching a moving object using a high-speed vision chip system are presented.

View full abstract

Nirvana Technology Inc. (NTI) was established in September, 2002 as a first ATR spin-off company. Its aim is to start business based on the technologies developed at ATR, especially at ATR Media Integration & Communications Research Laboratory where the author was the director/president. This paper briefly introduces the several basic features of NTI; Why NTI was established?, what kinds of technologies does it have?, and so on.

View full abstract

InterDesign Technologies, Inc., which is funded jointly by Toshiba and Mitsubishi Corporation, was established in March last year as a spin-off venture company from Toshiba R&D Center. Our company provides software tools and solutions for system level design of embedded systems and system LSIs such as SoC(System on Chip). The product has been developed by Toshiba’s long term(more than 10 years) R&D efforts related to system level design and methodology. This paper describes the vision and business activities of our company and also refers to a Japanese model of a venture company to activate Japanese industries.

View full abstract

Japanese SBIR(Small Business Innovation Research), which was introduced Japan in 1999 and has been implemented by the government. The Program for Promotion of Technology Innovation on Specified Issues as the main program of the Japanese SBIR conducted by Japan Small and Medium Enterprise Corporation showed highly success rate in the point of the commercialization of the product.

View full abstract

In this paper we try to extract various uneasy factors in our life. Then, we try to construct structural models among these factors using DEMATEL (Decision Making Trial and Evaluation Laboratory). For the purpose of analyzing priority among these factors we revised the DEMATEL and found effective factors to be resolved in order to realize future safe, secure and reliable (SSR) society.

View full abstract

We consider a system controlled by well organized and highly practised actions which are served by subroutines of a human operator. In this paper, the system is modeled by a piecewise affine system containing a hysteresis switch which idealizes a human’s discrete action. By using the model, we model an error classified as a slip which occurs when the intention fits the goals and the environmental situation but the action is not carried out as planned.

View full abstract

In this paper, we consider simultaneous stabilization problem in a behavioral framework. In the behavioral approach, control system synthesis includes not only a design of feedback controllers but also a design of optimal plant for desired specification. Thus, the solution for the above problem yields an interconnected system with high reliability. We derive necessary and sufficient condition for the existence of an simultaneous stabilizing system and provide one of algorithms for synthesis of reliable control systems.

View full abstract

The paper describes a mobile manipulator for monitoring and support of human activity in home and office to keep security. The mobile manipulator can be equipped with cameras at the tip of the arm. The camera pose can be changed flexibly by controlling both mobile platform and manipulator poses. It is useful to complement function of fixed-monitoring cameras in monitoring-support systems.

View full abstract

This paper proposes a method for person tracking and activity recognition. Our system can perform person tracking and activity recognition simultaneously. The system estimates the status of tracked person by using shape parameters and state transition probabilities. An action of tracked person is recognized by using MAP calculated in the tracking process.

View full abstract

3-D Object motion tracking method without feature extraction is necessary for monitoring human action in an usual civil-life scene. To realize a fast and robust object tracking method, we propose a model-based method using intensity images. At first, the 3-D shape and reflectance of the object are measured by using a rangefinder constructed with a camera and projector. Then, several CG images with varied object pose and position are generated and then compared to the input image. The pose and position of the object are estimated by minimizing the residual of the CG and input images. We made a special CG generator which is a precise simulator of the real camera to generate a CG image identical to the input image. We confirmed the ability of our method and achieved a 3DOFs real-time object tracking system in 4.5frame/sec speed.

View full abstract

We report our recent work about basic studies on crisis management systems for reliable communication/transportation network design and evacuation guidance of residents against large-scale disasters. First, we survey recent developments in modeling of evacuation problems. Next, we describe our efficient algorithm for evacuation and transportation network design to optimally locate facilities of safety.

View full abstract

In recent years, our life becomes more convenient rapidly. There are, however, various sources of uneasiness and anxiety. Our end goal is to give people safety and relief through information presentation on computer systems. Few researches on safety and relief have been done from a standpoint of system users. We focus on the role of color and voice, and propose basic guideline of information presentation for safety and relief.

View full abstract

This paper describes the framework and methodology of co-creation engineering. First, it shows some different systems and problems that concern with the research framework, and also introduces some examples of our research work. Then, it surveys the co-creative features between “self” and environment in human being at 3 levels of “self” referring to human and social sciences.

View full abstract

Co-creation of narratives has been used o form empathy among people, particulary in the area of narrative therapy. We will analyze the underlying structure of this mechanisms by referring to Schank-Abelson’s theory in artificial intelligence and Heider’s theory of interpersonal psychology.

View full abstract

Recently, university evaluation aiming at quality assurance and improvement attracts attention. Higher education and academic research, the mission of universities are difficult activities to evaluate. Evaluation based on self-study and peer review are methods to cope with these difficulties. This paper discusses such evaluation process from the viewpoint of co-creation systems.

View full abstract

How did money originate? This question has been answered mainly in two doubtful ways for long: commodity & cartal theory of money. But in this lecture, we propose another way: a model that operationally shows the exchange process which is simultaneously robust and yet precarious, and realizes an origin of money. At this point the individualization process of the external institutions is argued.

View full abstract

The essential feature of life is expressed as a weaken paradox. A pair of triplets consisting of two sets and a binary relation is given with an infomorphism [1] between triplets, and that is a pair of intent and extent of a system. The weaken paradox is defined by the mixture of an infomorphism and relations. It results in a dynamical system towards a collapse of the system in its own right.

View full abstract

Conventional ideas of “actions” has been discussed as byproducts of cognition. On the contrary, actions have move more significant meanings for collaboration between two and/or more actors. In this paper, an idea of “epistemic action” is introduced (in contrast to pragmatic action), and we implement this idea in the human-machine shared autonomy systems.

View full abstract

Usually, an identification of an open system in a world means a process of specifying its border, resolving it into components, and extracting a regularity of its time evolution with input from the environment. This process formally corresponds to the extraction of simulation relation. In the real world, however, the simulation relation suffers ceaseless collapse and resets. In this study, I clarify this situation, and propose a new framework for extracting systems, which positively exploit the collapse of (a modified) simulation relation. Some features of the proposed framework are reviewed.

View full abstract

Instead of a pair dynamics and the boundary condition, we propose system consisting of a pair of time-dynamics and space-dynamics. Both time-dynamics and space-dynamics defined and are transformed by their own dynamics, and then re-enter the other dynamics. We call it the time-space re-entrant system. The time-dynamics is due to logical self-reference, on the other hand space dynamics to frame problem. We propose the system as calculation of “within a finite universe”.

View full abstract

I present the overview of brain mechanisms underlying communication, stressing the importance of the meta-cognition of uncertainty. As a particular instance, I discuss the cognitive processes of mind reading.

View full abstract

Since Kolers and Grünau pointed the significant roles of color-perception in apparent motion is 1975, some theoretical models have been persued to describe the neuronal mechanisms of apparent motions. As for the brightness-perception, Grossberg et al proposed a theory based on the physiological analysis for neural networks in a ventral stream. However, it is difficult to explain the color-changing apparent motion from their theory. In this note, we extend the Grossberg’s idea and propose a new theoretical model to explain the apparent motions with color-perception.

View full abstract

In our daily lives, we feel stimuli to our different sensory modalities, such as vision and tactile sense, are synchronized. If our visual stimuli are delayed, visual and tactile perception can’t be synchronized and consistent. We think this asynchronization must influence now as so-called duration, and change the dual cognition of time-space. We have an experiment using delayed video, and evaluate the change.

View full abstract

Characters (letters) are divided into ideogram(e.g. Chinese character) and phonogram(e.g. hiragana). The difference between them can be quantitatively estimated, in term of the complexity of character’s structure. We have show that the reading depends both on the ratio of the chinese characters in a sentence, reading direction.

View full abstract

This paper describes a design framework, and an interface system for supporting human interaction with a sense of copresence between remote people. Our “inter-real virtual space” design presents a virtual reflection of a local physical space into a remote locale, and provides an interface space bridging remote locales in order to locate selves in common communication place. We have implemented a networked “lazy Susan” which includes a co-actuated table, the disk of which synchronizes with virtual disk and remote disk in real time as well as virtual avatar of remote person appears. Some results of trials indicate that our “lazy Susan” is available for supporting bodily interaction and collaborative work between remote locations.

View full abstract