Stephen R. Ellis

Discriminability of Prediction Artifacts in a Time-Delayed Virtual Environment:

Overall latency remains an impediment to perceived image stability and consequently to human performance in virtual environment (VE) systems. Predictive compensators have been proposed as a means to mitigate these shortcomings, but they introduce rendering errors because of induced motion overshoot and heightened noise. Discriminability of these compensator artifacts was investigated by a protocol in which head tracked image stability for 35 ms baseline VE system latency was compared against artificially added (16.7 to 100 ms) latency compensated by a previously studied Kalman Filter (KF) predictor. A control study in which uncompensated 16.7 to 100 ms latencies were compared against the baseline was also performed. Results from 10 subjects in the main study and 8 in the control group indicate that predictive compensation artifacts are less discernible than the disruptions of uncompensated time delay for the shorter but not the longer added latencies. We propose that noise magnification and overshoot are contributory cues to the presence of predictive compensation.

Discrimination of Changes of Latency during Voluntary Hand Movement of Virtual Objects

Eight subjects abilities to detect changes in system latency during voluntary lateral hand movement of virtual objects were studied in an immersing virtual environment. A two-alternative forced choice procedure was used in which discrimination of latency was studied with respect to three reference latencies: 27, 94, and 194 msec. Results show that subjects are able to reliably detect changes definitely less than 33 msec and probably less than 16.7 msec. Strikingly, for the short latencies we examined, subjects ability to detect latency changes does not depend upon the base latency we used as a reference. Thus, the discrimination we studied does not appear to follow Webers law and may provide evidence for quick adaptation to the reference latencies used.

Improved temporal response in virtual environments through system hardware and software reorganization

Excessive end-to-end latency and insufficient update rate continue to be major limitations of virtual environment (VE) system performance. Beginning from a typical baseline VE in which a spatial tracker is polled to deliver data via an RS-232 interface at each update of a single application program, we examined a series of hardware and software reconfigurations with the aim of reducing end-to-end latency and increasing update rate. These reconfigurations included: (1) multiple asynchronous UNIX processes communicating via shared memory; (2) continuous streaming rather than polled tracker operation; (3) multiple rather than single tracker instruments; and (4) higher bandwidth IEEE-488 parallel communication between tracker and computer. Starting from an average latency of 65 msec and an update rate of 20 Hz for a standard 1000 polygon test VE, our most successful implementation to date runs at 60 Hz (the maximum achievable with our graphics display hardware) with approximately 30 msec average latency. Because our equipment and architecture is based on widely available hardware (i.e., SGI computer, Polhemus Fastrak) and software (i.e., Sense8 WorldToolKit), our techniques and results are broadly applicable and easily transferable to other VE systems.

Human Control in Rotated Frames: Anisotropies in the Misalignment Disturbance Function of Pitch, Roll, and Yaw

Comparative misalignment disturbance functions (MDF) have been measured for rotations between display and control axes for pure pitch, roll, and yaw misalignments in a high fidelity virtual environment. Twenty participants manually moved a virtual cursor using position control to touch 3-dimensionally, randomly presented nearby targets having a constant Fitts Index of Difficulty. Results show a peak disturbance near 120° of rotation for all axes with Roll being distinguishably more disturbed. Some reasons for observed anisotropies, nonlinearities and an equiaxial spiral feature are briefly discussed and modeled

Three Dimensional Tracking in Augmented Environments: User Performance Trade-Offs between System Latency and Update Rate:

Three-dimensional tracking performance was measured as a function of system latency (35–335 msec) and update rate (10–30 Hz). Twelve subjects used a custom, see-through head mounted stereo display to control the position of a virtual response cursor with hand and body movements. User performance trade-offs between latency and update rate were measured with objective and subjective measures and a possible performance model was evaluated. The results indicate that earlier findings suggesting that latency influenced tracking performance more than did update rate, could be due to previous studies having tested latency over a larger dynamic range. Iso-performance contours are used to compare objective performance with subjective perception and performance judgments.

Influence of Head Motion on the Judged Distance of Monocularly Presented Virtual Objects

Human subjects localized a monocularly viewed, space-stabilized virtual object presented on a head-mounted, see-through display. They either kept their head stationary or rocked it laterally to produce motion parallax. Their distance estimates had less variability in a head moving condition than in a head stationary condition, but in general were much less precise and much less accurate than comparable stereo-based localizations.

Effect of Head-Slaved Visual Image Roll on Spatial Situation Awareness:

We examined whether the inclusion of a third head-slaved “roll” degree of freedom (dof)–in addition to pitch and yaw dofs–to control the orientation of a remotely-viewed or computer-synthesized scene can enhance spatial situation awareness. Six subjects were required to match the position and orientation of stationary target markers on a remote taskboard by manually placing response markers on an identical local taskboard. Subjects could only view the remote taskboard through images transmitted to a head mounted display (HMD) from a motorized pitch-yaw-roll camera platform; they could see neither the local taskboard nor their own limbs. Results show that, while systematic overshoot errors in azimuth judgment occurred regardless of the roll condition, the addition of the roll dof to the platform had no statistically discernible effect on the subjects ability to match the position (i.e., azimuth and elevation) of the remote targets. Absence of the roll dof, however, did affect the subjects judgment of target orientation when their heads were at maximum elevation (pitch) and azimuth (yaw) combinations.

Spatial sensor lag in virtual environment systems

This paper describes a testbed and method for characterizing the dynamic response of the type of spatial displacement transducers commonly used in VE applications. The testbed consists of a motorized rotary swing arm that imparts known displacement inputs to the VE sensor. The experimental method involves a series of tests in which the sensor is displaced back and forth at a number of controlled frequencies that span the bandwidth of volitional human movement. During the tests, actual swing arm angle and reported VE sensor displacements are collected and time stamped. Because of the time stamping technique, the response time of the sensor can be measured directly, independent of latencies in data transmission from the sensor unit and any processing by the interface application running on the host computer. Analysis of these experimental results allows sensor time delay and gain characteristics to be determined as a function of input frequency. Results from tests of several different VE spatial sensors are presented here to demonstrate use of the testbed and method.

Localization of a Monocularly Presented Virtual Object with Delayed Visual Feedback

Observers adjusted a pointer to match the depicted distance of a monocular virtual object viewed in a see-through head-mounted display. Distance information was available through motion parallax produced as the observers rocked back and forth. The apparent stability of the virtual object was impaired by the time delay between the observers head motions and the corresponding change in the object position on the display. Localizations were made for four time delays (31 ms, 64 ms, 131 ms, and 197 ms) and three depicted distances (75 cm, 95 cm, and 113 cm). The errors in localizations increased systematically with time delay and depicted distance.

Misalignment Effect Function Measurement for Oblique Rotation Axes Counterintuitive Predictions and Theoretical Extensions

The Misalignment Effect Function (MEF) describes the decrement in manual performance associated with a rotation between operators’ visual display frame of reference and that of their manual control. It now has been empirically determined for rotation axes oblique to canonical body axes and is compared with the MEF previously measured for rotations about canonical axes. A targeting rule, called the Secant Rule, based on these earlier measurements is derived from a hypothetical process and shown to describe some of the data from three previous experiments. It explains the motion trajectories determined for rotations less than 65° in purely kinematic terms without the need to appeal to a mental rotation process. Further analysis of this rule in three dimensions applied to oblique rotation axes leads to a somewhat surprising expectation that the difficulty posed by rotational misalignment should get harder as the required movement is shorter. This prediction is confirmed. Geometry underlying this rule also suggests analytic extensions for predicting more generally the difficulty of making movements in arbitrary directions subject to arbitrary misalignments.