Recently, the demand for 3-dimensional humanoid animation on the Internet has increased. We have developed an animation system to generate human actions based on a hierarchic human body description and action library. Using this system, 3D human body structure and behavior can be managed in a unified way and can be used with VRML. We were able to describe a human structure model and action information as well as develop methods to generate action animation utilizing libraries in the internet. This paper explains the concept and the structure of the system and its application on the Internet.

抄録全体を表示

Walking velocity is perceived by various senses. We investigated the accuracy of velocities perceived by the sense of sight and leg movement. In the first experiment the subjects adjust their walking velocity to the velocity displayed in the HMD. On the other hand, in the second experiment the subjects adjust the velocity displayed in the HMD to their walking velocity. The findings of both experiments indicate that the velocity perceived by the sense of sight is higher than actual walking velocity.

抄録全体を表示

We describe a newly developed locomotion interface system, called the GSS (Ground Surface Simulator). The GSS system is a novel locomotion device that provides a ground surface simulation with an active treadmill. This ground surface simulation function achieves a more realistic virtual walking environment. In this paper, we describe the objective and background of the GSS development program, the basic concept of the GSS system, its technical basis of implementation, and an overview of the first prototype system development. The first prototype system has a 60cm X 150cm walking surface on its treadmill belt, and has a ground surface simulation function.

抄録全体を表示

This paper describes a software tool for construction of a virtual environment with sense of move. A locomotion interface is implemented in various mechanical configurations. In most case, the software of the virtual environment is tightly connected to the control program of the locomotion interface. This problem is a hazard for development of further applications. We developed a modular software tool called LHX-L, which supports various type of locomotion interface. LHX-L is composed six modules. Application of the locomotion interface is easily reconfigured by exchanging these modules. The effectiveness of LHX-L is exemplified through various applications.

抄録全体を表示

This paper presents a new locomotion interface device for walking about virtual space. Traveling on foot is the most intuitive way for locomotion. An infinite surface driven by actuators is an ideal device for creation of sense of walking. We selected a torus-shaped surface to realize the locomotion interface. The locomotion device employs twelve sets of treadmills. These treadmills are connected sideby-side and driven in perpendicular direction. The effect of infinite surface is generated by the motion of the treadmills. The walker can go in any direction while his/her position remains localized in the real world. The device is applied to study of human performance on spatial recognition

抄録全体を表示

This paper describes locomtion interface that presents uneven virtual surface. The term "haptics" is currently being generalized beyond just hand-object interaction. One of the new frontiers of haptic interface is full-body haptics that includes foot haptics. Force applied to a foot is essential in walking on uneven virtual surface, such as staircases. We developed a 4DOF motion platform for each foot. Two motion platforms move according with positon of the feet. These devices are mounted on a turn table, which rotates the devices according with rotation of the body. This mechanism simulates omni-directional uneven surface.

抄録全体を表示

The principle of the scalp skull and brain dipole tracing (SSB/DT) method and its application were described. The SSB/DT method enabled to estimate the location of current dipoles from EEG (10-20 international system). The main principle of the arithmetic algorithm consisted in the calculation of the local minimum of the evaluated function; that is, the difference of the recorded and the estimated potentials on the scalp. Before calculation, the three-layered head model of individual subjects was constructed based on X ray-CT photographs taken 5 mm interval of the horizontal plane. The conductance of the scalp, skull and brain, which is very important for precise calculation of the current dipole, was estimated by passing a small amount of rectangular current (0. 1 mA x 5 msec) at the scalp. Several physiological studies conducted in our and other laboratories demonstrated the usefullness of the SSB/DT method.

抄録全体を表示

We developed a rendering system that can simulate the focal blur of the human lens in real time. The system enables us to provide focal blur information in 3D computer graphic images while the observer's eyes are moving around naturally. Using this system, we found that focal blur synchronized with the observer's eye movement can be effective in judging the depth arrangement of three spheres in a virtual space. We extended the graphics system to a stereo presentation by using twin projectors with polarizing filters, and found that focal blur information emphasizes depth perception and reality in virtual space even though binocular disparity is involved.

抄録全体を表示

R-Cubed (R^3: Real-time Remote Robotics) is the concept and technology for people to telexist anywhere in the world by controlling remote robots as his avatars in real time base through the network. An R-Cubed system is supposed to allow people to control remote robots from terminals installed on homes, offices, or any public booths connected to Internet or other dedicated networks. In this paper some efforts devoted to the design of R-Cubed systems are shown, including the specification of RCML (R-Cubed Manipulation Language) that describes remote environments and robot characteristics, and RCTP (R-Cubed Transfer Protocol) that is in charge of network communication between users and robots in real time base. Finally an experimental implementation example and evaluation to the behavior of this system is reported.

抄録全体を表示

A haptic device provides force feedback by means of a simulator in which the force, position, velocity etc. of a probe are used as input data. If there exist dimensional errors in the device mechanism and installation errors in the sensors, the simulator cannot calculate correctly actuator torques/forces and the displayed force becomes distorted, which means the force is not proportional to the operator's force. In the present paper, the magnitude of the distortion which depends largely on the type and dimension of the mechanism is evaluated kinematically and statically by proposing Nonlinearity Index. A numerical example is shown to illustrate the characteristics of the index. Moreover, an optimal kinematic design system for haptic devices by means of Genetic Algorithm with consideration of the nonlinearity of displayed force in addition to the operability, spatial output-isotropy and size of the mechanism is proposed and applied to a device as a master-arm of a master-slave system assisting precision task such as assembling small packaging substrates.

抄録全体を表示

This paper presents a robot teaching for multi-fingered robots based on the intention of human motion in virtual reality. The human motion measured in virtual reality environment is segmented into plural primitive motions. The segmentation is made using not only 3D motion of the human and virtual object but also virtual reaction generated at the hand of human attached a force-feedback-glove. Each primitive motion is represented as a human motion command by a smooth time function to avoid the vibration mode of robot. The robot teaching command with respect to the object coordinate frame is generated from the human motion command. An experimental result of pick & place task is presented.

抄録全体を表示