To stabilize our position in space we use visual information as well as non-visual physical motion cues. However, visual cues can be ambiguous: visually perceived motion may be caused by self-movement, movement of the environment, or both. The nervous system must combine the ambiguous visual cues with noisy physical motion cues to resolve this ambiguity and control our body posture. Here we have developed a Bayesian model that formalizes how the nervous system could solve this problem. In this model, the nervous system combines the sensory cues to estimate the movement of the body. We analytically demonstrate that, as long as visual stimulation is fast in comparison to the uncertainty in our perception of body movement, the optimal strategy is to weight visually perceived movement velocities proportional to a power law. We find that this model accounts for the nonlinear influence of experimentally induced visual motion on human postural behavior both in our data and in previously published results.

We present a coordinated ensemble of scalable computing techniques to accelerate a number of key tasks needed for vision-based gesture interaction, by using the cluster driving a large display system. A hybrid strategy that partitions the scanning task of a frame image by both region and scale is proposed. Based on this hybrid strategy, a novel data structure called a scanning tree is designed to organize the computing nodes. The level of effectiveness of the proposed solution was tested by incorporating it into a gesture interface controlling a ultra-high-resolution tiled display wall.

In this paper, the effects of three visual factors: scene complexity, stereovision and motion parallax on correct perception of a virtual object's size were analyzed in an immersive virtual environment. We designed a controlled experiments set to incorporate visual conditions that reflected all twelve different configuration combinations of the three visual factors. Under each visual condition, subject performed the task of making judgments of the sizes of a virtual object displayed at five different distances from him/her. A total number of eighteen subjects participated in our study. The subjects' judgments and the corresponding actual sizes of the virtual object were recorded. Based on the collected data, two quantitative measures of subjects' performance were derived and analyzed. The results of our experiments were consistent across the majority of the subject population and suggested that scene complexity and stereovision could have significant impact on the performance of a user of virtual environments to make correct judgments on a virtual object's size. On the contrary, motion parallax, either produced by the virtual environment or by the observer, might not be a significant factor in determining that performance

With technological progress, wide field-of-view (FOV) displays will become increasingly common. Wide FOVs provide a more immersive environment and produce stronger self-motion perception. The objective of this study was to investigate the relationships between FOV and scene content on postural stability in an immersive environment. 10 subjects were tested using two different scenes (a simple radial pattern and a "meaningful" city scene) at six FOVs (30/spl deg/, 60/spl deg/, 90/spl deg/, 120/spl deg/, 150/spl deg/ 180/spl deg/) using a within-subjects design. Subjects exhibited more postural disturbance with increasing FOV. A surprisingly large increase in disturbance was found for the interval between 150/spl deg/ and 180/spl deg/ using the city scene. No statistically significant difference was found for effects of scene content. Two groups (postural stable group and postural unstable group) were identified during experiment. These groups performed differently in the two scene conditions. Future research plans are described.

We assessed the effects of network latency and jitter on a cooperative teleoperation task in a collaborative virtual environment. Two remote partners worked together to manipulate shared virtual objects over a network. The task was to minimize the time to transfer a ring through one of four paths with the least number of collisions. The performance of human subjects was measured and analyzed quantitatively as a function of network latency: 10 and 200 msec delays with and without jitter. Jitter had the greatest impact on coordination performance when the latency was high and the task was difficult. These results are discussed in light of current and future CVE tasks.

The design, implementation, and testing of virtual environments is complicated by the concurrency and real-time features of these systems. Therefore, the development of formal methods for modeling and analysis of virtual environments is highly desirable. In the past, Petri net models have led to good empirical results in the automatic verification of concurrent and real-time systems. We applied a timed extension of Petri nets to modeling and analysis of the CAVE/sup TM/ virtual environment at the University of Illinois at Chicago. We report on our time Petri net model and on empirical studies that we conducted with the Cabernet toolset from Politecnico di Milano. Our experiments uncovered a flaw in the way a shared buffer is used by CAVE processes. Due to an erroneous synchronization on the buffer, different CAVE walls can simultaneously display images based on different input information. We conclude from our empirical studies that Petri net-based tools can effectively support the development of reliable virtual environments.

Adaptability is one of man's advantages over machines. Perhaps one of the reasons for our limited understanding about human adaptation during manual tracking tasks is that we have only limited tools to identify the model coefficients (especially delay time) of an adapting human operator. In this paper, we introduce a discrete time recursive delay identifier (RDI) capable of simultaneously estimating a human operator's nonstationary delay time and linear model coefficients. At its core lies the extended Kalman filter (EKF). Our goal to obtain fractional delay time estimates was realized by using the bicubic interpolation scheme as part of the EKF to provide subsample magnitude and derivative estimates of the observed input/output time series. While this theoretically limits the RDI applicability to band-limited or differentiable signals, this is seldom a concern in practice. Based on data from simulated and experimental time varying tracking tasks, we show the RDI's potential to substantially increase our understanding about human adaptations thus perhaps offering new avenues for machine adaptation.

In this paper we present a method for compression of stereo images. The proposed scheme is a frequency domain approach based on the suppression theory of binocular vision. By using the information in the frequency domain, complex disparity estimation techniques can be avoided. The wavelet transform is used to obtain a multiresolution analysis of the stereo pair by which the subbands convey the necessary frequency domain information.

Human operator characteristics were measured during a fixed-base visual tracking task where the field of view (FOV) varied from 10 degrees to 120 degrees . Using the critical tracking (CT) task, five subjects were tested at 10 degrees , 20 degrees , 40 degrees , 80 degrees , and 120 degrees FOV. The measured effective time delay declined exponentially as the FOV increased. The corresponding root-mean-squared (RMS) error followed a U-shaped curve with the majority of the RMS reduction at 40 degrees FOV. A second experiment, in which two subjects were tested at 10 degrees , 40 degrees , and 120 degrees FOV, used a time-invariant plant to allow the measurement of human describing-function parameters. The crossover frequency, increased at least 5% and the RMS error dropped by at least 20% at a FOV of 40 degrees or 120 degrees compared to 10 degrees . The results from these two experiments show that a FOV as small as 40 degrees can produce performance improvements of the same magnitude as a FOV as large as 120 degrees . In the final experiment, where only the central 10 degrees of the scene rotated, performance was unexpectedly best at 10 degrees and poorest at 40 degrees and 80 degrees .>