Thomas J. Hebert, Ph.D., Associate Professor, UH; Bernard Henneberger, Graduate Student, UH

The problem of simultaneously tracking an object and estimating its position and orientation (or pose) using a CCD video camera is important to the NASA mission. The solution to this problem has application to automated docking of space vehicles, to the grasping of geometrically complex objects with the space shuttle's robotic arm, and to mobile robotics. Real-time object tracking requires a machine vision system to extract 3-D information from a video image sequence at video frame rates. This requirement places stringent constraints on the algorithmic approaches that can be proposed for solving the tracking and pose estimation problem.

This project addresses the theory, design of a hybrid optical/digital system for automated tracking of known objects. This system, shown in Fig. 1, incorporates a modified 2-F optical correlator that is based upon a liquid crystal (LC) spatial light modulator (SLM). A computer controls the zoom, rotation, pan and tilt of the camera that drives the input SLM. A set of filters for the complex transmittances of the second SLM are stored in memory on the system's computer to allow correlations with many different stored filters.

We have proposed a new general class of optimal solutions to the problem of object tracking and pose estimation with the system above. The solution framework, shown in Fig. 2, has a pre-processing stage, an optical correlation stage, a post-processing stage, and a non-linear least squares stage. The following denote:

This class of solutions offers three advantages, (a) clearly defined optimality, (b) suitability for real-time implementatoin on the hybrid optical/digital system, and (c) a component structure, whereby the robustness of the solution can be increased by inclusion of additional optical correlators. Preliminary experimentation using Silicon Graphics 3-D rendering language OpenGL with a CAD model of the Spartan-201 spacecraft, has demonstrated feasibility of the approach.

David M. Hoffman, Associate Professor, UH; Jung-Sook Kim, Research Assistant, UH

The quaternary aluminum-gallium-indium nitride (AlGaInN) system is a direct bandgap semiconductor tunable from 2 to 6.2eV. It is the most promising system for semiconductor device applications at blue wavelengths, a region of the spectrum not readily accessible using current technology. Extending the range of available semiconductor optical devices into the blue region would greatly advance display and graphics technologies like those used in the air and space industries. The 240-280 nm (4.75 eV) band, which is absorbed by the earth's ozone layer, also has important space applications.[1] Operating in this range, satellite-to-satellite communications would be shielded from earth monitoring, and imaging array detectors located on the dark side of the earth could provide for a very sensitive surveillance of objects coming up out of the earth's atmosphere.[1]

Figure 1. Plot of In(N(SiH(CH3)2)
(C(CH3)3))3(pyridine-p-N(CH3)2) from an X-ray crystallographic study.

Considerable effort in recent years has been devoted to growing and characterizing aluminum, gallium, and indium nitride thin films and films of their alloys because of their potential for applications in semiconductor devices.[1,2,3,4] High quality InN films remain the most difficult of the III-V nitride films to grow. The problems encountered with InN are related to its poor thermal stability at temperatures >500&degree;C, which is well below typical III-V nitride deposition temperatures. To realize the potential of the III-V nitrides in applications, new routes must be developed that allow the full range of AlGaInN alloys to be prepared at low substrate temperatures.

We previously established that tris(dimethylamido)aluminum and gallium, M(N(CH3)2)3 (M = Al, Ga), and ammonia are excellent precursors for the low temperature (<450&degree;C) chemical vapor deposition (CVD) of MN films.[5,6] With this successful low temperature route to aluminum and gallium nitride established, we sought to develop the analogous approach of using tris(amido)indium compounds as precursors to indium nitride. In this grant period, we have successfully synthesized several new indium amido complexes including the dimer [InCl(µ-(CH3)NSi(CH3)2N(CH3)Si(CH3)2N(CH3))]2 and In(N(SiH(CH3)2)(C(CH3)3))3. The latter compound is a liquid that appears to be stable and quite volatile, important attributes for a practical CVD precursor. To be certain that the compound was formulated correctly, we synthesized a solid derivative and carried out a single-crystal X-ray diffraction study (Fig. 1).

In summary, we have synthesized new indium amido complexes that are potential precursors to InN thin films. One of the complexes, In(N(SiH(CH3)2)(C(CH3)3))3, is the most promising precursor candidate because of its desirable physical properties. Indium nitride deposition studies using In(N(SiH(CH3)2)(C(CH3)3))3 and ammonia precursors are in progress.

References
1S. Strite, H. MorkoV. "GaN, AlN, and InN: A Review," J. Vac. Sci. Technol. B10 (1992): 1237.
2S. Strite, M. E. Lin, H. MorkoV. "Progress and Prospects for GaN and III-V Nitride Semiconductors," Thin Solid Films 231 (1993): 197.
3J. H. Edgar. "Prospects for Device Implementation of Wide Band Gap Semiconductors," J. Mater. Res. 7 (1992): 235.
4T. Matsuoka. "Current Status of GaN and Related Compounds as Wide-Gap Semiconductors," J. Crystal Growth 124 (1992): 433.
5R. G. Gordon, D. M. Hoffman, U. Riaz. "Low Temperature Preparation of Gallium Nitride Thin Films," Proc. Mat. Res. Soc. Symposium 242 (1992): 445.
6D. M. Hoffman, S. P. Rangarajan, S. D. Athavale, D. J. Economou, J. -R. Liu, Z. Zheng, W. -K. Chu. "Chemical Vapor Deposition of Aluminum and Gallium Nitride Thin Films from Metalorganic Precursors," J. Vac. Sci. Technol. A. 14 (1996): 306.

Ben H. Jansen, Ph.D., Professor, UH; Demetri C. Mavrofrides; Graduate Student, UH; Arunachalam B. Kavaipatti, Graduate Student, UH; Vonda W. Woodie, Doctoral Student, UH

Following space flight, the ability to locomote is often severely impaired, a debility that would jeopardize a crew member's safety in an emergency. Understanding the physiological mechanisms underlying the neurosensory adaptation processes associated with spaceflight is essential to the development and evaluation of countermeasures to optimize crew health, safety, and performance. Being able to objectively quantify the physiological changes induced by spaceflight and/or the effects of countermeasures is a key requirement on the way to understanding. Therefore, we are developing software tools to aid in understanding the nature of the inflight neurosensory adaptive processes responsible for post-flight changes in (free) locomotion patterns and the post-flight adaptive processes leading to recovery of normal locomotion.

We work with electromyographic (EMG) collected by NASA from astronauts before and shortly after Space Shuttle and long duration flights (Mir Space Station) and made available to us by the Motor Performance Laboratory of NASA/JSC. Astronauts were asked to walk on a treadmill at a constant speed, and EMG data were obtained from four muscles from each of the legs (Rectus Femoris, Biceps Femoris, Tibialis Anterior, and Gastrocnemius), plus toe-off and heel strike information using foot switches.

Typically, data from about twenty contiguous steps were collected from an astronaut per session, with at least two sessions before, and two to three sessions after spaceflight (i.e., at day of landing, two days, and four days later). An example of the EMG obtained during one gait cycle is shown in the left panel of Fig. 1. The EMG is rectified and low-pass filtered (10 Hz cut-off) to obtain estimates of the amplitude envelope. Typically, we normalize the linear envelope by subtracting the mean and dividing by the standard deviation. The middle panel of Fig. 1 shows the result.

The method we are in the process of developing proceeds in three stages. First, multi-channel EMG recordings are transformed to trajectories, which provide a view of the phasing and activation levels of all the muscles simultaneously. Trajectories are constructed in a four-dimensional space, where each axis represents one muscle. At instance t = nT, where T = 1/fs (I>fs is the sampling frequency) the four muscles of the right leg define the point R(l), B(l), T(l), G(l) in 4-D space, where R stands for rectus femoris, B for biceps femoris, etc. Incrementing l and "connecting-the-dots" result in the desired trajectory.

Next, representative gait cycles ("templates") that account for most of the variability in the data are determined using a clustering algorithm. In the third and last stage of our method, all the gait cycles from one (pre- or post-flight) session of a particular astronaut are compared to the templates, and a count is taken to determine how often each template is characterized as being most similar to gait cycles in question. In this way, we obtain a histogram for each session, expressing how often each template occurred in that session. Changes in gait following space-flight will show up as changes in these histograms, which can be compared using statistical tests.

An example is presented in Fig. 2, where data from two astronauts (A2402 and A3453) are shown. Ten templates for each astronaut were required to adequately represent the data collected from two pre-flight sessions (A and C) and two post-flight sessions (D and E). The trajectories for the three most frequently occurring templates are shown as well. Two views of each trajectory are provided, showing the Biceps, Rectus and Tibialis muscle in one case, and the Biceps, Rectus and Gastrocnemius in the second view. The five templates labeled with an "A" come from pre-flight sessions, and the six labeled with a "D" from post-flight data. As one can see, the two pre-flight sessions of each astronaut have roughly the same distribution, with the majority of the gait cycles assigned to the pre-flight templates. Histograms from the post-flight sessions differ from the pre-flight ones, with the most profound differences observed in the D-session, which took place on the day of the landing. In case of astronaut A2042, eight of the 20 gait cycles were similar to those seen pre-flight, and this number increased to 15 in session E, which took place two days after landing. This information suggests a gradual return to pre-flight gait patterns. Similar observations can be made for astronaut A3453, but the changes are even more dramatic, with session D having zero cycles resembling pre-flight gait patterns, whereas session E has eleven such cycles.

Findings presented in Fig. 2 demonstrate the sensitivity of the proposed method and are even more remarkable considering the fact that our NASA collaborators (Drs. Layne and Bloomberg) have not been able to assess that pre- and post-flight EMG responses following short-duration space flight are significantly different using more traditional (single-channel) EMG measures.

Stephen E. Morse, OD, Ph.D. Associate Professor, UH; Bai-Chuan Jiang, Ph.D., Associate Research Professor, UH; Brian G. Morse, Undergraduate Student, UT Austin; Tara Aye, Graduate Student, UH

Recent reports have shown that approximately 50 percent of the individuals who use virtual reality (VR) head-mounted displays (HMD) have adverse visual symptoms. These reports have documented changes in users' oculomotor functions. However, they did not establish if the observed oculomotor changes were similar for symptomatic and asymptomatic individuals.

Purpose
We measured oculomotor function of both symptomatic and asymptomatic individuals to determine if there were fundamental oculomotor performance differences that might account for the adverse visual symptoms reported during or after VR HMD use.

Method
Before and after wearing a commercially available bi-ocular VR HMD for 20 minutes, the 13 volunteer subjects had binocular vision assessments that included the following clinical measures: associated and dissociated horizontal and vertical phorias; fixation disparity, gradient accommodative convergence to accommodation ratio (AC/A), stereopsis, and near point of convergence. Measures were selected for speed of administration to allow assessment of any oculomotor aftereffect of the VR test session. Before and after each VR test session, subjects completed a symptom questionnaire.

Results
Five of 13 subjects were symptomatic and eight were asymptomatic. We found that the gradient AC/A ratio of symptomatic subjects was consistently reduced after a period of VR HMD use. In contrast, asymptomatic subjects displayed no consistent change in gradient AC/A. None of the other clinical tests showed significant differences between the symptomatic and asymptomatic groups.

Conclusion
While no specific vision test identified individuals susceptible to vision symptoms before VR HMD use, we did find that symptomatic individuals have consistent changes in oculomotor functioning after VR HMD use. The reduction in gradient AC/A displayed by the symptomatic subjects could be attributed to accommodation, vergence, or their interaction. We speculate that sympathetic nervous system arousal displayed by symptomatic subjects could lessen accommodative response and reduce the gradient AC/A, a mechanism we are currently investigating.

Lawrence S. Pinsky, Ph.D., Professor, UH

For the last several years, there has been an active effort to convince NASA that existing computer simulation tools developed for use in high energy physics experiments are applicable to the problem of predicting the radiation environment in and around spacecraft, or in a habitat on the lunar surface. These Monte Carlo simulations have been used to model the propagation of high energy particles through structures as complex and detailed as the Space Shuttle. Such a simulation permits detailed accounting of the effects of the radiation on constituent parts of the spacecraft and its crew. The technique offers a considerable improvement over the present methods used by NASA to estimate the radiation environment.

Objectives
NASA has provided a one year grant to pursue the concept to the extent necessary to submit a formal proposal. The NASA, grant which did not provide funds for summer salary, was supplemented by ISSO funding in the amount of one month's support to enable the demonstration.

Because NASA has a funding cycle that begins with proposal deadlines by April 15 each year, the objective is to develop a demonstration simulation that can be used within a proposal for submission by April 15, 1997. That proposal will request funds to support a nominal 3-year initial program to develop a formal integrated simulation.

Methodology
We have decided to try to adapt the evolving code known as FLUKA for use in a space simulation. FLUKA is the present incarnation of a Monte Carlo transport code that has evolved over the last decade to become the most physics-correct transport code in existence. FLUKA presently suffers from having a more primitive interface and lesser graphics capabilities in comparison with better known codes such as the famous GENT code developed at CERN in Geneva Switzerland. It is our hope to be able to combine the superior physics treatments found in FLUKA with a better interface package. Such a package is being developed at CERN under the project name ROOT. The ultimate goal therefore is to obtain a code capable of simulating the space radiation environment and the effects of that environment as it propagates through a geometrically realistic spacecraft. Spacecraft details will be input from industry standard CAD file descriptions of the spacecraft.

Status
The effort to develop a demonstration code is presently underway at the University of Houston, where the latest version of FLUKA has been installed and is in the process of being checked out. The initial simulation will focus on the simulation of protons only in the primary flux, and calculate how they propagate through a simple slab. This geometry was chosen to allow a reasonable comparison with the results of prior calculations. Dr. Iota Foka of Geneva Switzerland is assisting with the project, although she receives no direct ISSO support.

ISSO * 1995-1996 * Annual Report