Search

This report describes the results of Honeywell's effort under the Optoelectronic Technology Consortium project. The program involved a consortium of Honeywell, AT&T, IBM and Martin Marietta. The overall goal of the program was to develop parallel fiber optic data link component and packaging technology and to demonstrate a link possessing 32 channels, each operating a 500 Mbps for an aggregate data rate of 16 Gbps. Baseline and alternate technologies were investigated. Honeywell's effort included: (1) the development of AlGaAs waveguide based four element electro-optic modulator arrays as an alternate transmitter technology, (2) packaging technology required to couple single mode fiber to, and multimode fiber from, the modulator array, and (3) a polymer waveguide based board level interconnect demonstration included an expanded beam connector feasibility study. Partway through the program we added a task to develop a second source capability for the Vertical Cavity Surface Emitting laser (VCSEL) arrays. During the program we demonstrated multi-gigahertz bandwidth modulator arrays with a contrast ratio >10dB, and a voltage-length product of approximately 5.5V- cm. The packaging technique developed allowed one to maintain a 1 dB additional coupling loss over the 0 to +123 deg C temperature range. The polymer waveguide demonstration showed that 60 optical waveguide channels could be coupled from board to-board through a 3rnm connector with a fraction of a millimeter board to board alignment tolerance. We also demonstrated the fabrication of VCSEL arrays using MOCVD as the epitaxial layer growth technology.

During this reporting period the effort was concentrated on carrying an experiment on measuring the coupling efficiency and tolerances involved in transferring data from board to board via polymer waveguides on the boards and an expanded beam connector between the boards. Waveguide propagation loss was measured as 0.3dB/cm, and the cOnnector loss was 3.4 dB. Translational tolerance was measured as +O.37/-O.25mm, while rotational tolerance was measured to be +0.19/-O.15 deg. These results all appear to be within a practical level and consistent with tolerances achievable between a motherboard and daughterboards.

During the current reporting period effort was concentrated on verifying the high speed characteristics of the waveguide modulators, and preparing to produce VCSEL arrays. The waveguide modulators were found to exceed the speed requirements for the link. The speed of the transmitter will therefore be limited by the speed of the driver circuit and the image matching between the driver and the modulator. Progress in the area of VCSELs included the elimination of the special contact layer, and an improved understanding of the factors limiting the ohmic contact resistance. During the remainder of the program we will package the waveguide modulators into a transmitter module with the driver chips and measure DC performance over temperature. We will also evaluate the board level interconnect demonstration. in particular we will examine the connector tolerance to misalignment

The Optoelectronic Technology Consortium has been established to position U.S. industry as the world leader in optical interconnect technology by developing, fabricating, integrating and demonstrating the producibility of optoelectronic components for high-density/high-data-rate processors and accelerating the insertion of this technology into military and commercial applications. This objective will be accomplished by a program focused in three areas. Demonstrated performance: OETC will demonstrate an aggregate data transfer rate of 16 Gb/s between single transmitter and receiver packages, as well as the expandability of this technology by combining four links in parallel to achieve a 64 Gb/s link. Accelerated development: By collaborating during the precompetitive technology development stage, OETC will advance the development of optical components and produce links for a multiboard processor testbed demonstration. Producibility: OETC's technology will achieve this performance by using components that are affordable, and reliable, with a line BER<10 to the - 15th and MTTF>10 to the 6th hours.

This report describes the Environment for Software Prototyping (ESP). Within ESP, prototyping is viewed as an experimental process whose results drive the evolution of a domain theory in the context of a megaprogramming environment. ESP is used to evaluate the nature and efficiency of technologies that address requirements unique to prototyping. ESP consists of a component shelf containing reusable components (i.e., a component consists of a specification and a set of implementations), a repository for reusable software architectures (i.e., an architecture defines the structural characteristics of a system), and a workbench (i.e., a workspace for conducting prototyping experiments). ESP provides an environment for bootstrapping a megaprogramming capability.

The Engineering Information System (EIS) was created in response to increasing concerns about the exchange of engineering information between contractors and the government, contractors on a single team, and contractors on different teams who share technical data and diverse tools. The EIS specifications provide a means for managing heterogeneity among hardware and software platforms, data formats, tools, site-specific policies and methodologies, and interfaces to create multiple-domain, multiple-life-cycle environments. An Engineering Information Model (EIM) and the EIS framework comprise the EIS. The EIM is a semantic model of information found in an EIS. An EIM realization, called the reference schema, Is a set of object types that form a specific logical organization of data and operations. The EIS framework contains automated services embodied in software to support the use of the EIM to manage and control the data and activities of the engineering process. Current computer networking gateways require significant enhancement to meet the needs of cost-effective and concurrent engineering information access. The use of integration information gateways is the key to achieving the levels of technical interchange needed between organizations and various engineering environments. The EIS technologies can provide the Integration necessary to allow transparent information access across the information exchange networks.

The Optoelectronics Technology Consortium has been established to position U.S. industry as the world leader in optical interconnect technology by developing, fabricating, intergrating and demonstrating the producibility of optoelectronic components for high-density/high-data-rate processors and accelerating the insertion of this technology into military and commercial applications. This objective will be accomplished by a program focused in three areas. (1) Demonstrated performance: OETC will demonstrate an aggregate data transfer rate of 16 Gbit/s between single transmitter and receiver packages, as well as the expandability of this technology by combing four links in parallel to achieve a 64 Gbit/s link. (2) Accelerated development: By collaborating during precompetitive technology development stage, OTEC will advance the development of optical components and produce links for a multiboard processor testbed demonstration; and (3) Producibility: OETC's technology will achieve this performance by using components that are affordable, and reliable, with a line BER<10-15 and MTTF>10 6 hours.

This report summarizes the assessment of the Rome Laboratory Software Life Cycle Support Environment (SLCSE) for potential application of knowledge- based technology. The work performed under this contract included three major tasks: Task 1: Analyze SLCSE; Task 2: Specify Approaches for inserting knowledge-based techniques; Task 3: Develop a Software Requirements Specification and Software Development Plan for incorporating near-term knowledge-based enhancements into the SLCSE. This report provides an overview of these three tasks, and concludes with areas for further evaluation.... Knowledge-bases, Software engineering environments.

A computationally efficient real-time trajectory optimization and guidance approach for hypersonic aircraft is described. The optimization approach is based on Euler-Lagrange energy state approximations. A three-dimensional, spherical-earth, aircraft-motion model with constraints on temperature, dynamic pressure, acceleration, and angle of attack is employed. Climb-to-orbit, return-from-orbit, flight-to-designated-landing-site, unpowered-abort, and powered-abort flight conditions are considered. Different performance criterion are used for different problems. Solution methods of varying computational complexity and performance capability are developed. An exact solution to the optimization problem using iteration on adjoint variables is developed.

This report describes advances in robust control in three areas: Optimal H infinity control singular values, and dynamic inversion. The H infinity results are a thorough treatment of the theory as it has been developed over the last three years. The structured singular value section describes an application of the technique to represent inertia parametric variations in the Space Station. The dynamic inversion section addresses global stability of aircraft pitch axis dynamics using a dynamic inversion control approach. This document is the first draft of the final report for the program New Methods in Robust Control. The emphasis of this program was to develop mathematical theory to help control system designers faced with challenging control problems associated with advanced aerospace vehicles. Relevant applications include flight control systems for new Air force fighter/bomber aircraft, the F-18 HARV research vehicle, the NASP vehicle, the next generation launch system (ALS or NLS), and the Space Station.

Systems containing both compact real parametric uncertainty and frequency-weighted bounded operator uncertainty are addressed. It is shown that any parameter adaptive control system is robustly stable provided only that: (1) the unknown parameters lie in a known compact convex set, (2) the control design rule is lipschitzian, (3) the control design rule would produce a robust controller if given perfect parameter information, and (4) a specified robust parameter estimation algorithm is applied in lieu of perfect parameter information. It is also shown that the asymptotic robust performance level may be arbitrarily close to that of the non-adaptive design which would result from perfect parameter information. Keywords: Mathematical matrices; Robust adaptor control; Asymptotic robust performance.

The topic is the theory of structured singular values in robust control analysis. Classical methods of invariant theory are applied to the problem of computing the structured singular value for the nonrepeated diagonal complex block structure. An algorithm for computing the structured singular value for the four-block problem is described. (CP), See also Volume 1, AD-A225 164.

This report describes research performed while developing specifications for a support environment (framework) suitable for serving as the common basis for integration and continued development of the many lifecycle facets of the Knowledge-Based Software Assistant (KBSA). The KBSA will be a system which provides significant automation in the development and lifetime support of large software systems. A history of the effort including a background for many of the design decisions and the lessons learned throughout the project are included.

A computationally efficient, real-time trajectory optimization and guidance approach for hypersonic aircraft is described. The optimization algorithms compute, in flight, minimum fuel trajectories within constraints from the current aircraft position to its final destination. In flight trajectory computation can provide a high degree of vehicle autonomy which could greatly reduce aircraft ground support costs. The optimization approach is based on Euler-Lagrange theory and energy-state approximations. A three-dimensional, spherical earth, aircraft motion model, with constraints on temperature, dynamic pressure and stall, is employed. An exact optimal iterative solution method and an approximate closed-form feedback solution method are developed. In the exact approach, solutions are computed By iteration on adjoint constants. Each iteration requires a complete forward trajectory integration. In the approximate closed-form feedback approach, the adjoint constants are expressed in terms of the states. No forward trajectory calculations are required, thus, the computations are minimal. Minimum-fuel climb to orbit, powered abort, and unpowered abort trajectories are computed with both methods. The approximate closed-form feedback solution closely matches the optimal iterative solution., This article is from 'Hypersonic Combined Cycle Propulsion Panel Symposium (75th) Held in Madrid, Spain on 28 May - 1 June 1990 (La Propulsion Hypersonique a Cycles Combines),' AD-A253 867, p3-1 thru p3-22.

The objective of the contract 'Research to provide automated training and testing capability for Air Defense weapon systems' is to demonstrate the training and testing potential of a portable, microcomputer-based instructional delivery system utilizing videodisc laser image access, videographic overlay of computer generated imagery, task action responses via equipment console mock-up, and courseware authoring/control with enhanced Programming Language for Interactive Teaching (PLANIT). A 20-minute sample Skill Qualification Test (SQT) Hands-On Component will be developed for Military Occupational Specialty (MOS) 25L--the AN/TSQ-73 (Missile Minder) air defense artillery command and control system operator/repairer. This report provides PLANIT programming for a representative 3 of 12 tasks identified as critical AN/TSQ-73 console operator's tasks. The programming strategy allows errors to be tested in a realistic manner, by an algorithm which evaluates console mock-up switch sequences in terms of optional order and previously activated switches. The programming for these tasks can serve as a model for programming additional tasks., Prepared in cooperation with Perceptronics, Inc., Arlington, VA.

This volume contains appendixes related to 'Training Effectiveness as a Function of Training Device Fidelity.' The appendixes consist of computer program listings and user documentation, raw rating and performance data, illustrations of training devices, and instructions to subjects.

This program explored the feasibility of a generalized approach to acoustic sensor operator training and resulted in recommendations concerning implementation. The program involved the analysis of the task, skill, and knowledge requirements for acoustic sensor operators (ASO) across a representative sample of sensor systems.

The use of operational equipment as a primary means for providing training and hands-on practice of electronics maintenance tasks in electronics- oriented class A schools can be costly, inefficient, dangerous, and inadequate. Therefore, the Navy has initiated an engineering development effort to develop, test, and evaluate a general-purpose electronic equipment maintenance training (EEMT) system that can significantly reduce reliance on operational equipment at the class A school level. Planning for system test and evaluation (T and E) is required in support of this effort. This report reports on the development of EEMT DTEMP (Device Test and Evaluation Master Plan). This plan will be used to (1) direct and control all T and E phases, and (2) identify contractor and government resources, long-range planning programming, and budgeting required to support T and E effort., See also AD-A102 200, AD-A087 715 and AD-A098 304.

Imagery simulating the output of a moving target indicator (MTI) radar was used in two experiments to determine the effect on the ability of radar operators to track targets moving in one particular area, when the radar was being switched intermittently to provide coverage of a second area as well. In both experiments, we investigated the effects of switching away from the area of interest after each complete scan of that area, varying the length of each interruption that occurred before returning for another scan of the primary area. The frames of simulated MTI radar imagery used for each condition were shown in time compression to twelve operators in both experiments. The results of the two experiments seem to present a coherent picture. Increasing the length of the interruptions in coverage from 15 to 30 seconds had little effect on tracking performance. With interruptions of 45 to 90 seconds, performance was worse with smaller target units, containing ten vehicles, than it was for larger units. However, with interruptions of 120 seconds, there was a decrement for all target units, large and small, compared to the level of tracking performance achieved with shorter interruptions., This article is from 'Proceedings of the Symposium: Psychology in the Department of Defense (9th) Held at Colorado Springs, Colorado on 18-20 April 1984', AD-A141 043, p445-449. See also AD-P003 323.

Simulated moving target indicator (MTI) radar imagery was used in the two experiments performed in this, one of an on-going series of studies in which we are investigating conditions of intermittent radar coverage. In the first of these experiments there were two, and in the second experiment there were two, three, or four consecutively-collected frames of imagery before each interruption in coverage. In both experiments there were twelve operators and the imagery was presented in time compression. The results of the two experiments showed that the use of more than one frame of imagery before the occurrence of the interruptions in radar coverage resulted in improved performance. For the larger, 25- and 50-vehicle units, two consecutively- collected frames, and for the smaller 10-vehicle units, four consecutively collected frames helped to prevent the decrement in tracking performance that otherwise occurred with 120-second interruptions., This article is from 'Proceedings of the Symposium: Psychology in the Department of Defense (9th) Held at Colorado Springs, Colorado on 18-20 April 1984', AD-A141 043, p450-454. See also AD-P003 322.

Ada is a programming language designed in accordance with requirements defined by the United States Department of Defense: the so-called Steelman requirements. Overall, these requirements call for a language with considerable expressive power covering a wide application domain. As a result, the language includes facilities offered by classical languages such as Pascal as well as facilities often found only in specialized languages. Thus the language is a modern algorithmic language with the usual control structures, and with the ability to define types and subprograms. It also serves the need for modularity, whereby data, types, and subprograms can be packaged. It treats modularity in the physical sense as well, with a facility to support separate compilation. In addition to these aspects, the language covers real-time programming, with facilities to model parallel tasks and to handle exceptions. It also covers systems programming; this requires precise control over the representation of data and access to system-dependent properties. Finally, both application-level and machine-level input-ootput are defined., See also AD-A090 709.

This report documents research on the potential use of speech technology in a manned penetration bomber aircraft (B-52/G and H). The objectives of the project were to analyze the pilot/copilot crewstation tasks over a three-hour-and forty-minute mission and determine the tasks that would benefit the most from conversion to speech recognition/generation, determine the technological feasibility of each of the identified tasks, and prioritize these tasks based on these criteria. Secondary objectives of the program were to enunciate research strategies in the application of speech technologies in airborne environments, and develop guidelines for briefing user commands on the potential of using speech technologies in the cockpit. The results of this study indicated that for the B-52 crewmember, speech recognition would be most beneficial for retrieving chart and procedural data that is contained in the flight manuals. Technological feasibility of these tasks indicated that the checklist and procedural retrieval tasks would be highly feasible for a speech recognition system.

The decision-aid being developed is identified as SUMMET (survivability Measures and Methods Evaluation Technique). SUMMET will aid TACOM managers in deciding which of many countermeasure alternatives to improve combat vehicle survival should be funded for development. SUMMET development has progressed to a point where the decision-aid concept is available to try, to test, and to evaluate. Computerized aids are available to support trying, testing, and evaluating the SUMMET concept.

New methods are presented for analysis and synthesis of control systems subject to structured uncertainties. The technical approach involves the structured singular value, as an analysis tool and H infinity as a synthesis tool. Alternative formulations are compared with the H infinity approach, extensions of mu to handle real parameter problems are presented, and the issue of the convergence of mu synthesis to a global optimum is studied. A comprehensive solution for the synthesis of general optimal controllers is given for linear lumped time-invariant systems. The existence of an optimal solution for H inifinity optimization is proved, and some properties of this solution are discovered. The method called Euler's constant iteration is presented and its convergence properties are established. A new algorithm for solving a class of algebraic riccati equations is obtained Explicit error bounds for model reduction in the synthesis process are derived. The constructions use standard matrix operations and linear algebra applied to state-space representations of linear systems.

A total systems approach methodology and computer-aided design tool for integrated flight control design and evaluation has been formulated. This methodology is a composition of classical system design steps and system performance, reliability/survivability, packaging, and cost evaluation techniques assembled for interactive computer-based execution. The interactive sensor integration program developed under this methodology addresses the sensor requirements of the flight control, guidance and navigation, and weapon delivery functions as directed interactively by the designer and the mission, the system configuration, cost, power, weight, volume constraints, and total system performance. The demonstration program contains performance reliability, size, power, weight, cost data for gyros, accelerometers, radars, seekers, and air data systems in addition to analysis tools for evaluating navigation system accuracy, probability of target acquisition, and probability of kill for selected targets and weapons.

Honeywell has conducted a computer-assisted analysis in order to predict the range of ambient luminance conditions under which Helmet Mounted Display (HMD) imagery can be usefully resolved by an observer. This report examines those variables relating to ambient luminance, display luminance, and the attenuation of each in terms of optimizing display contrast minimizing differences in luminance to the two eyes. Visual performance predictions were made based upon comparing the analytically derived data with criteria values obtained from a limited number of technical reports. These criteria values were used even though it was recognized that the monocular see-through HMD is a unique system, thus limiting the applicability of data obtained from studies of related but still dissimilar systems such as head-up displays (HUDs) and occluded monocular HMDs. Among the factors affecting the quality of the displayed imagery, those variables most crucial to display luminance and contrast were systematically varied on a computer.

This report summarizes research performed to develop a methodology for deciding requirements for display dynamics that ensure good flying qualities. To this end, a model-based approach was pursued that uses the optimal control model (OCM) of human operator behavior. One aspect of the investigation was to further establish the utility of the OCM in this context by correlating model-based quantities with flying qualities data from existing data bases. Another aspect was to consider how various factors at the pilot/display interface map into usual OCM perceptual parameters. A third element of the investigation was to examine the specific case of an STOL vehicle in an approach and landing configuration, with an explicity modeled processing delay in the HUD. This case study showed the negative impact of the delay and demonstrated how the introduction of additional display dynamics can compensate for its effect. Finally, a methodology for deciding display dynamics requirements has been suggested that is based on multivariable control theory, specifically the structured singular value. The methodology provides a means for assessing robustness of system performance and pilot workload to variations in pilot behavior and with respect to variable aircraft dynamics. It also incorporates a synthesis procedure that selects display dynamics so that a balance is achieved between optimal performance and adequate robustness. From such synthesis results, display dynamics requirements can be abstracted. A detailed example that illustrates this approach is provided. Keywords: Man machine systems, Flight control systems, Head-up displays.

The objective of this study was to determine the effects of reduced training devices fidelity on learning and performance of a perceptual-motor maintenance task. Bicycle wheel truing was chosen for study. Five devices including the actual equipment were procured, or designed and built. The device fidelity was systematically varied in physical and functional similarity to the actual equipment. One hundred naive high school and vocational technical school students served as paid subjects; 20 were trained in each device condition. All subjects were then tested on the actual equipment. The results indicated that significant skill was acquired under all training conditions. The amount of skill acquired did not differ as a function of overall fidelity (i.e., with physical and functional similarity at the same level). However, further analysis in which these two dimensions were separated showed a significant effect of physical similarity. High physical similarity resulted in higher performance on the transfer of training task than low physical similarity. Further research was proposed in the context of specific experiments. Finally, recommendations for the organization and communication of research results via a computerized database were presented., See also Appendixes, ADA133104.

This report presents a conceptual framework for identifying factors that may impact training simulator effectiveness. A research strategy is proposed for the empirical determination of necessary levels of training simulator fidelity. Preliminary studies consistent with this strategy are described to explore the effects of device fidelity on the transfer of training of perceptual-motor and cognitive maintenance tasks.

When completely independent images are presented to each eye, fusion normally cannot occur. Instead, either an involuntary alternation occurs between the two images (binocular rivalry) or one of the images is 'suppressed,' and visual performance is degraded. If the images are only partially independent (see-through display), the observer can control this alternation and presumably reduce degradation. During the development of one version of the Honeywell Helmet Mounted. Sight and Display (HMS/D) system, it was suggested that presenting a gun-sight reticle one eye and target imagery to the other eye either sequentially, or with an inter-ocular delay interval (IOD), might minimize binocular rivalry. It was hypothesized that if binocular rivalry did exist, and if it occurred even when only temporal summation linked the images to the two eyes, visual performance would be best when display imagery was presented to only one eye, and worst when presented to both simultaneously. Further, performance would fall off in a regular fashion between these two extremes as the IOD was decrease. Statistical analysis of the data failed to confirm these hypotheses, and it was concluded that the influence of binocular rivalry on target recognition tasks was negligible with a see-through display.

This study was carried out in order to validate experimentally the predictions of previous analytic work (Cohen, 1973). The objective was to determine the best combination of filter coatings and display luminances required for viewing the Helmet Mounted Display (HMD) against a wide, but operationally realistic variety of background luminances. The predictions relating beamsplitter transmittance, visor transmittance and display luminance to performance were largely supported by the data.

The objective of this effort is to develop, design and analyze techniques that can be used to meet future line of sight and jitter control requirements of large space structures that have onboard disturbances. This effort studied parameter identification on a representative large space structure and the impact of on-orbit identification on the design of a disturbance controller for the spacecraft.

A by-product of future automated tactical systems is a proliferation of data to be organized and analyzed. To help cope with this problem and bring about an efficient man-computer interface, an evaluation of the decision aiding literature was undertaken. Special emphasis was placed on those aids which might be useful in tactical systems and could be implemented in the G3 versions of SIMTOS (i.e., simulated tactical operations system). It was found that more than one decision aid was necessary to impact significantly on an ultimate decision. Therefore, a decision support complex was designed for incorporation into the SIMTOS. The support complex utilized a normative approach and an adaptive approach in regard to estimating the pre-combat tactical situation and in regard to allocating fire resources during combat., See also Volume 3, ADA040563.

A review of the decision aiding literature has shown that a multitude of decision aids is needed to impact on an ultimate decision. A decision support structure was designed for use in a simulated tactical operations system (SIMTOS). The present research provided a preliminary evaluation of the decision support complex. Ten participants were divided into two groups. One group used the decision aid throughout the simulated defensive scenario while the other group did not. Both groups participated in the same G3 planning and combat activities. The aided group (i.e., the group using the decision complex) used more of the available resources than the unaided group. Although there was no significant difference between the two groups in terms of distance surrendered, or friendly and enemy force attrition, the trend favored the aided group. The general results show that the concept of a decision support complex is a sound one. ARI is continuing to utilize the decision support complex in further studies of tactical decision making., See also Volume 1, ADA040606.

In an experiment designed to evaluate the effectiveness of an automated decision support system on tactical information processing and decisionmaking performance, 25 experienced Army officers played the role of Division G3 in the Simulated Tactical Operations System (SIMTOS). Results of the evaluation demonstrated that the concept of decision support is sound, although specific measures of tactical performance were insensitive to player actions. The decision support complex increased player's ability to interact effectively with SIMTOS and increased their satisfaction with the system. Further development of automated decision support for tactical decision making is warranted.

The automation of Molecular Beam Epitaxy system shutters was completed for the substrate and the Mercury, Cadmium Telluride and Telluride sources. As reported earlier, the automated shutter operations allow for much more precise thickness control of the superlattice layers during the deposition. Also, from signal-to-noise considerations, performing the nonlinear optical experiments requires the use of thick (2 micrometer) superlattice samples which involve several hundred layers of Mercury Telluride and Cadmium Telluride. This can best be accomplished by using an automated Molecular Beam Epitaxy system. The current superlattice design calls for alternate layers of HgTe and CdTe with thicknesses of 100-150 A. A p-type sample with this structure is expected to yield a third order susceptibility (x(3)) of 100 times larger than that of the bulk HgCdTe. A number of test runs were carried out with the automated MBE system. These runs resulted in five superlattice samples of various thicknesses. One sample is now being used in a preliminary experiment of optical phase conjugation to establish baseline data.

This report summarizes our progress and results during this reporting period. Our experiments show that the growth of thick (approx 5 micrometers) superlattice samples of HgTe/CdTe requires the use of an MBE system with in situ diagnostic capability to assure stability of growth parameters for periods of several hours. Our efforts with our in house MBE system have not succeeded in growing thick superlattice samples needed for the optical experiments.

This report describes the research results on Honeywell's Hierarchical Multisensor Image Understanding program. Honeywell is developing a unified framework for the different hierarchical levels of image processing such as segmentation, detection, classification, and identification of outdoor scenes and across different sensor modalities such as millimeter wave, infrared, and visible. Current activities on the project are reviewed under the following headings: (1) A Survey of Multisource Information Fusion Systems; (2) The Role of Structure in Human and Machine Perception; (3) A Knowledge Based Image Segmentation System; (4) The Use of Optical Flow as a Depth Cue in Scene Analysis and (5) Belief Maintenance for A Fuzzy Reasoning System. Keywords include: Image processing; Image understanding; Artificial intelligence; and Scene analysis.

This report, in the form of a set of notes, details Honeywell's research results of the past year in Robust Multivariable Control Theory. These notes are made up of four major parts. Part 0 gives a review of the required notation and mathematical background. Part 1 reviews recent results on the problem of analyzing the performance and robustness properties of systems. Part 2 presents the results on synthesis which are the highlight of this report, and Part 3 outlines how the methods of the previous parts apply to control of large space structures. (Author)

In this report we analyze the feasibility of subarc-second measurement of earth crust warp and/or earth spin axis deviation, in less than one day, using ring laser gyroscopes and accelerometers and/or tiltmeters. It is marginally feasibile, using laser gyros equivalent to the Honeywell GG1389, either unidirectionally carouseled or with enhancement equivalent to the Honeywell closed-loop lockin correction (CLIC). Experimental results on the GG1389 show that input axis stability is more than adequate, and that its carouseled or CLIC-enhanced random drift and output resolution suffice marginally. Design and build of a one-axis feasibility demonstrator is recommended, using one CLIC-enhanced GG1389 ring laser gyro, two state-of-the-art tiltmeters, an Ultradex indexer, and a Hewlett-Packard micro-computer. Keywords include: Crustal motion, Earth spin axis deviation, Northfinder, Subarc-second, Measurement, Polar axis deviation, Laser gyro, Ring laser gyro.

The main function of an evidence accrual system for image understanding is to sequentially update information on scene objects based on new sensor data or on non-sensory information such as intelligence. This paper presents a concept for sequentially updating information on scene objects. Scene objects and background (clutter) are represented by attributed relational graphs in which nodes represent objects of interest and arcs represent inter-object relations. Dynamic recognition/identification of nodes is accomplished by a belief/disbelief measure. Our experimental results with infrared images show improvements in natural scene object recognition over traditional image processing methods. (Author)

This report describes the research results on Honeywell's Hierarchical Multisensor Image Understanding program. Honeywell is developing a unified framework for the different hierarchical levels of image processing such as segmentation, detection, classification, and identification of outdoor scenes and across different sensor modalities such as millimeter wave, infrared, and visible. Current activities on the project are reviewed under the following headings: (1) artificial-intelligence-based generic image segmentation and object recognition; (2) evidence-confidence paradigms for image understanding; (3) hierarchical systems theory for control structures; and (4) invariant methods in image understanding. Also discussed are scene analysis and attributed graphs., Original contains color plates: All DTIC and NTIS reproductions will be in black and white.

A new CCD Matrix Array has been fabricated including a Microprocessor Host-CPU and Controller Exerciser Electronics Unit. Each unit cell of the array performs analog arithmetic and logic operations on input data matrix elements in the Single Instruction Multiple Data architecture. Ultra-high signal and image processing throughput is demonstrated. (Author)

The Image Processing and Digitization Facility (IPDF) is a sophisticated system of fully integrated hardware and software for engineering advanced image-processing algorithms. The IPDF combines highly specialized hardware elements in a generalized software environment that provides consistent and convenient access to all elements of the system. The IPDF provides the research scientist with command-level access to all image functions to facilitate quick evaluation and development of new algorithm ideas. The IPDF then supports the implementaiton of efficient simulations to test algorithms over large databases of images. The IPDF is a research tool which provides all the facilities for design and development of state-of-the-art image-processing algorithms. (Author)

The objective of this effort was to generate the data for and develop a preliminary training technology handbook. The final form of this handbook is intended to assist hardware acquisition managers, training program developers, and others in estimating the composition and cost of training required for new weapon system acquisitions. This report is being released at this time as a working draft to provide for distribution to the research community. Further development will be required before the handbook is ready for broad application within the hardware acquisition community., Prepared in cooperations with Calspan Advanced Technology Center, Buffalo, NY.

This report presents results from a Naval Research Laboratory-sponsored investigation of dual polarization gyros. The dual polarization concept is a new fiber optic gyro configuration designed to provide independence from principal gyro bias error sources. Experimental results indicate an order of magnitude improvement in bias error performance. Important results from a new modeling capability are also included. In addition to examination of dual polarization gyro performance, a number of fiber gyro configurations have been examined. Results of this effort include development of an improved understanding of fiber optic depolarizers and identification of a previously unreported error team for gyros using 3 dB distributed couplers. (Author)

This study investigated, developed, and evaluated various Computer Image Generation (CIG) techniques to overcome the qualitative limitations of current CIG imaginery produced by edge-based systems. The study concluded with an integration of techniques into a system concept. This report describes the techniques investigated, the system concept developed, and the general hardware implementations which are useful for cost/benefit tradeoffs. The system concept presented is based on the use of textured terrain for realistic simulation. In areas where the technique was for training effectiveness and an algorithmically unified system. The approach also involves the display of terrain as curved surfaces represented by bicubic splines. (Author)