Kazunori Shidoji

Detecting drowsiness while driving by measuring eye movement - a pilot study

A function to warn a drowsy driver is a prospective device to be included in intelligent transportation systems. We compared the eye closure rate (PERCLOS) with the vertical eye movement frequency between wakeful and drowsy states of drivers who were driving a car in a driving simulator. Three subjects participated in the experiment. In the drowsy state condition, both measures showed high values in frequency, which was significant enough to distinguish between the two states with a threshold. The threshold level, which can distinguish the two states, is defined as the rate at the cross point of the curves of two possibility density functions made by the frequency of the rates of the two measures at the two states. The distinction error was evaluated for three analysis durations. As a result of the analysis, it was found that the better measure was different for each subject.

The cause of traffic accidents when drivers use car phones and the functional requirements of car phones for safe driving

This paper reports how driving behaviour is affected by the use of mobile phones. We measured the reaction time (RT) in various conditions to detect what kind of effects mobile phone use has in the face of accidents. It is important to study RT in driving because the stopping distance of the car varies with the drivers RT. For this study, we prepared a system which measures the time from when a lamp mounted on the windshield of the car lights up to when the driver steps on the brake pedal. The RT was measured under three conditions: a) During a conversation using a handset b) During a manual task with a handset c) During a manual task with a handset, with eyes on the road ahead. The results of our experiments showed that the RT was prolonged unexpectedly when the driver took his eyes off the road to make or to take a phone call. The main effect of the car phone appears to be on the distraction of visual attention, rather than on physical or mental demand in operating the phone and engaging in a conversation.

New stereoscopic video camera and monitor system with central high resolution

A new stereoscopic video system (the Q stereoscopic video system), which has high resolution in the central area, has been developed using four video cameras and four video displays. The Q stereoscopic camera system is constructed using two cameras with wide-angle lenses, which are combined as the stereoscopic camera system, and two cameras with narrow-angle lenses, which are combined (using half mirrors) with each of the wide-angle cameras to have the same optical center axis. The Q stereoscopic display system is composed of two large video displays that receive images from the wide-angle stereoscopic cameras, and two smaller displays projecting images from the narrow-angle cameras. With this system, human operators are able to see the stereoscopic images of the smaller displays inserted in the images of the larger displays. Completion times for the pick-up task of a remote controlled robot were shorter when using the Q stereoscopic video system rather than a conventional stereoscopic video system.

Effect of overlap rate between stereoscopic images on performance in a teleoperation

According to the results of the simulated teleoperation experiment, the larger the ratio of the overlapping area of stereoscopic images, the smaller the completion times and the number of errors. For this paper we did an experiment using the actual stereoscopic video system. We examined the performance of the teleoperation of an insert task in two experiments. In experiment 1, we set three fixed overlap rate conditions for the stereoscopic image pairs. (High overlap rate condition): The convergence point of the two cameras was set at the goal point where a cylindrical object was inserted. When subjects fixated their eyes on the goal point, the overlap rate of the images from the cameras was 95%. (Middle overlap rate condition): The convergence point of the cameras was set at the center of the working area. When subjects fixated their eyes on the goal points, it was 76.7%. (Low overlap rate condition): Convergence point of the camera was set at the point in the situation where the ratio of the overlapped area was 49% when subjects fixated to the goal point. Completion times and the numbers of errors of the insert task were measured. As a result, these were smallest at the high overlap rate condition. In experiment 2, we compared the performance between a fixed and a variable overlap rate conditions in a pick-and-insert task. The experimental results suggested that the number of errors of variable overlap rate condition was less than that of the fixed condition although the completion time of the former condition was not shorter than that of the latter condition.

An ellipsoidal trajectory model for measuring the line of sight

The relationship between the amount of rotation of the eyeball and the amount of rotation of the line of sight was investigated. The planar model and ellipsoidal model were compared as the trajectory of the center of the pupil. An evaluation of precision was carried out by measuring the discrepancy between the direction of the target and the estimated direction of the line of sight. The experimental results showed that the planar model was more precise than the ellipsoidal model. The center of rotation of the eyeball seems to translate along the oblique rotation of the eyeball.

Stereoscopic video system with embedded high spatial resolution images using two channels for transmission

Teleoperation requires both wide vision to recognize a whole workspace and fine vision to recognize the precise structure of objects which an operator wants to see. In order to achieve high operational efficiency in teleoperation, we have developed the Q stereoscopic video system which is constructed of four sets of video cameras and monitors. It requires four video channels to transmit video signals. However, four channels are not always available for a video system because of the limitation of the number of radio channels when multiple systems are used at the same time. Therefore we have tried to reduce the number of channels on this system by sending images from the right and left cameras alternately by field. In experiment 1, we compared the acuity of depth perception under three kinds of stereoscopic video systems, the original Q stereoscopic video system, the Q stereoscopic video system with two channel transmission, and the conventional stereoscopic video system. As the result of the experiment, the original Q stereoscopic video system enabled us to perceive depth most precisely, the Q stereoscopic video system with two channel transmission less so, and the conventional stereoscopic video system even less. In experiment 2, we compared the Q stereoscopic video system with two channel transmission to the original Q stereoscopic video system. The result showed that the operators were able to work more efficiently under the original Q stereoscopic video system than under the Q stereoscopic video system with two channel transmissions. In experiment 3, we compared the Q stereoscopic video system with two channel transmission to the conventional stereoscopic video system. It was found out in this study that the new stereoscopic video system we developed enabled operators to work more efficiently and to perceive depth more precisely than the conventional stereoscopic video system, although the number of channels for image transmission of this system was equal to that of the conventional stereoscopic video system.

Perception of absolute and relative distances in stereoscopic image

In 3-D movies and virtual reality and augmented reality systems, stereoscopic images are used to improve perceptions of realism and depth. The distance (depth) that we perceive can be classified into absolute distance, which is the distance between the observer and the objects, and relative distance, which is the distance between the objects. It is known that in the real environment these two distances are independent. Previous studies have reported that we underestimate or overestimate the absolute distance in stereoscopic images under some circumstances. We examined perceptions of both absolute and relative distances in stereoscopic images to investigate the depth perception of virtual objects. The results of the experiments showed that (1) the perceived absolute distance from the observer to images in front of the screen was nearly accurate and that to images farther behind the screen was underestimated; (2) this underestimation tendency increased in short viewing distances; and (3) the relative distance from the screen to images in front of the screen was overestimated, whereas that to images farther behind the screen was underestimated.

An effect of a large overlapped area of stereo pairs at the working point on a spatial multi-resolution stereoscopic video system

Teleoperation is a very useful technique in the case of carrying out tasks in hazardous and/or remote sites. In order to carry out the tasks in high operational efficiency, a video system to display the workspace for an operator is important. In this study, we developed a spatial multi-resolution stereoscopic video system which has variable convergence point, and confirmed the effect of keeping an overlapped area large at the working point by varying the convergence point continuously with the spatial multi-resolution stereoscopic video system. As the results of this experiment, it was clarified that the operational efficiency with teleoperation through the spatial multi-resolution stereoscopic video system was improved by keeping an overlapped area large, which resulted from varying the convergence point. That is, in order to carry out tasks with teleoperation in high operational efficiency, the system is required to vary its convergence point to keep an overlapped area large.

Effect of the ratio difference of overlapped areas of stereoscopic images on each eye in a teleoperation

It is reported that the efficiency of a teleoperation in stereoscopic images of the working site is lower than that in the direct viewing of the site. It is assumed that one of the causes of lower efficiency of the teleoperation in the stereoscopic images would be the difficulty of the fusion of image, which would be caused by the imperfect overlapping of image on each eye. Through most of a teleoperation the convergence of the stereoscopic cameras is fixed at a certain point, usually in the middle of the working area. When the plane of the operators eye-fixation-point is apart from the plane of the convergence point of the stereoscopic cameras, the two images do not overlap perfectly. It requires a great deal of effort for the images to be fused when the difference of the depth of two places is over a certain value. We hypothesized that imperfect overlapping of the images on the left and right eyes would cause a decrease in efficiency for a teleoperation. We examined the efficiency of a teleoperation in tow kinds of camera convergence conditions in a virtual reality (VR) environment. (Condition 1): The convergence point of the cameras follows the point on the target object on which subjects fixate with their both eyes, the ratio of the overlapped area of two images is always nearly at maximum. (Condition 2): The convergence point of both cameras is not set on the target object, but at the center of the hole- base. The large the difference between the plane of the cameras convergence point and the plane of the subjects fixation point becomes, the more the rate of the overlapped are of the two images decreases. We prepared four cylinders and ahole-base with four holes in which the cylinders were inserted. The subject was asked to insert a cylinder in a hole using a 3D mouse which allows free movement in the VR space. We measured the completion times of the operation and the number of errors in each condition to evaluate the efficiency of the operation. As a result of the experiment, the completion times of the operation and the number of errors under nearly-perfectly overlapped conditions were significantly smaller than the ones under the conditions in which the overlapping is less than maximum value. The experiment revealed work performance decreased when the ratio of overlapping of two images, which were projected on each eye, is less than maximum value. These result led to the conclusion that in order to achieve good performance in a teleoperation, the convergence point of the cameras should follow the target object on which subjects fixate with their both eyes.

Digital stereoscopic video system with embedded high resolution images

We developed a stereoscopic video system, which has high-resolution images for central vision and it is called the Q system. The Q system uses a compound image that is a wide-angle image with an embedded high-resolution image. However, the Q system could not be used under situations where many robots work at the same time. This is because it needs four channels of video signals and the available channels could be limited under such situations. Thus, we have developed a digital Q system. The system can be used under such restricted situations, because the required data transfer rate is adjustable by changing the compression rates for a high-resolution image and a wide-angle image. In addition, an experiment confirmed that even though the systems used the same data transfer rate, digital Q system could make teleoperation more efficient and more precise than a conventional stereoscopic video system.