LATENT SEMANTIC ANALYSIS FOR CAREER FIELD ANALYSIS AND INFORMATION OPERATIONS

Darrell Laham and Marcia Derr

Knowledge Analysis Technologies

Boulder, Colorado

Winston Bennett

Air Force Research Laboratory

Mesa, Arizona

This paper reviews two current Air Force Research Laboratory Human Effectiveness Directorate (AFRL/HEA) efforts that are maturing Latent Semantic Analysis (LSA) tools for the Air Force. The first effort is developing new LSA-based agent software that helps decision makers to identify required job knowledge, determine which members of the workforce have the knowledge, pinpoint needed retraining content, and maximize training and retraining efficiency.  Modern organizations are increasingly faced with rapid changes in technology and missions and need constantly changing mixes of competencies and skills.  Assembling personnel with the right knowledge and experience for a task is especially difficult when there are few experts, unfamiliar devices, redefined goals, and short lead-times for training and deployment.  LSA is being used to analyze course content and materials from current training pipelines and to identify appropriate places in alternative structures where that content can be reused.  This saves time for training developers since the preexisting content has already been validated as a part of its earlier application. 

AFRL/HEA’s second research effort involves a demonstration of a combined speech-to-text and LSA-based software agent for embedding automatic, continuous, and cumulative analysis of verbal interactions in individual and team operational environments.  The agent will systematically parse and evaluate verbal communication to identify critical information and content required of many of today’s AF operators.  LSA is promising new technology that has significant potential for assisting operators in the performance of their tasks because it can “listen” and in almost real-time evaluate free-form verbal communication from a variety of sources and match content to stored language dictionaries.  One application of this technology being explored is tracking and scoring the tactical communications that occur between the members of a four-ship air combat flight and their weapons director to identify areas of training need and as an additional tool for assessing the efficacy of DMT scenarios and missions.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

GRAPHICAL EXPERT SYSTEM FOR ANALYZING NUCLEAR FACILITY VULNERABILITY

C. David Sulfredge, Robert L. Sanders, Douglas E. Peplow,

and Robert H. Morris

Oak Ridge National Laboratory

Oak Ridge, Tennessee

and

LTC Todd A. Hann

Defense Threat Reduction Agency

Alexandria, Virginia

Nuclear facilities are a vital part of the nation's infrastructure providing approximately 20% of the nation's electricity and representing a vast investment of national resources.  They also have the potential for serious radiological releases following an incident, which could adversely affect public health and contaminate large areas.  Following the September 11^th attacks, it has become clear that nuclear facilities rank very high among possible targets that must be safeguarded from terrorists.  Conversely, mission planners need a targeting tool to estimate what sort of damage would be required to defeat operations at a nuclear facility and how serious the collateral consequences of an attack might be.  The Defense Threat Reduction Agency (DTRA) and the Oak Ridge National Laboratory (ORNL) are currently developing a novel graphical expert system, the Visual Interactive Site Analysis Code (VISAC), to address both of these needs.

For a single software package to meet these design goals, VISAC must integrate the three concepts of target geometric modeling, damage assessment capabilities, and event/fault tree methodology similar to that used in the nuclear industry for evaluating accident consequences.  As currently written, VISAC is a Java-based graphical user interface (GUI) that can analyze a variety of accidents/incidents at nuclear or industrial facilities ranging from simple component sabotage to an attack with military or terrorist weapons.  The list of damaged components from a scenario is then propagated through a set of event/fault trees to determine the overall facility kill probability, the probability of an accompanying radiological release, and the expected facility downtime.  An initial version of VISAC is now being used for test calculations by the DTRA.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

AN ACOUSTIC TARGET SCORING AND EVALUATION TECHNOLOGY

Dr. Niall McNelis and Chris Yaniger

AAI Corporation

Timonium, Maryland

This paper addresses research into the application of acoustic-shockwave measurement technology to facilitate the instantaneous scoring of targets (including virtual targets). The scoring data is statistically analyzed and provided in real-time to an operator (or directly to the gunner/gunnery team) for concurrent gunner/weapon system performance evaluation at US Army firing ranges. In particular, this paper discusses the theory of operation, the system components, the usage options, and the value of instant feedback and performance evaluation. When a supersonic projectile travels through the atmosphere, a shock wave is propagated away from the nose of the projectile at the speed of sound. This shock wave propagates normal to the shock front. These shock waves can be detected by an array of sensors that are carefully designed to detect the fast risetimes. The shock wave's sharp risetime allows the accurate calculation of the projectile trajectory geometry. This in turn allows the projectile missed distance, trajectory azimuth and elevation, and projectile velocity to be derived. In addition to solving for the geometrical solution, the first two hundred microseconds of the acoustic pressure profile are examined to extract projectile signature information. This signature information is sufficient to classify the round type.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

MISSION COMPLEXITY SCORING IN DISTRIBUTED MISSION TRAINING

Todd Denning

L3 Communications, Link Simulation and Training Division

Mesa, Arizona

Winston Bennett, Jr. and Peter Crane

Air Force Research Laboratory

Warfighter Training Research Division

Mesa, Arizona

At last year’s conference (Crane, Robbins, Bennett & Bell, 2001) we presented a first look at an empirically-based methodology for quantifying the complexity of Distributed Mission Training (DMT) scenarios for simulation-based, team training.  The outcome of that effort was the development of a scenario complexity index derived from subject matter expert judgments which has been used to assess the effects of training in a building-block syllabus.  Over the past few months, training specialists and subject matter experts at Air Force Research Laboratory’s Mesa Research site have been characterizing DMT scenarios in terms of specific learning objectives linked to mission-essential competencies and to the underlying knowledge, skills, and experiences that are required for successful combat performance.  As part of scenario characterization, we have identified the mission characteristics and levels of those characteristics that are important for complexity indexing.  Techniques for determining overall scenario complexity and for relating scenario characteristics to mission essential competencies have been developed.  This paper presents results from a validation study comparing the new, analytically-based complexity methodology with the empirically-based approach presented last year.  In addition, applications of the new assessment to both F-15 and F-16 weapon system capability and scenario characteristics to learning objectives while controlling overall complexity will be discussed together with plans for developing DMT instructor support systems.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

A COMPLEX SYNTHETIC ENVIRONMENT FOR

REAL-TIME, DISTRIBUTED AIRCREW TRAINING RESEACH

Dave Greschke

Air Force Research Laboratory

Mesa, Arizona

Edward Mayo

Qinetiq, Bedford, United Kingdom

Stuart C. Grant

Defense R & D Canada, Toronto, Ontario, Canada

During November 2001, AFRL (US), Defense R & D Canada and Qinetiq (on behalf of the UK MoD), conducted the first in a series of simulation trials to investigate the potential of an international collective environment for aircrew training. The trial involved simulating a mixed air-to-air and air-to-ground package using manned simulators in the US for the friendly force and manned simulators, computer generated forces, and human role players in the UK for both the friendly and enemy force. The Canadian site was a passive stealth node on the network.

The simulated operational environment was designed to be as realistic as possible. Front-line crews manned the simulators, while other military personnel took the roles of the command chain for both friendly and hostile forces. This allowed the simulated missions to be run as they would be in a real operational environment, with full pre-sortie briefings, crew planning, sortie execution and debriefing all conducted via a secure long haul link.

The implementation of the trial infrastructure involved significant development and integration effort, covering aspects such as long-haul secure data and voice communications; scenario development and management; data recording and analysis tools; planning, briefing and debriefing systems; and computer generated forces. This paper describes the trial infrastructure, explains its development, and reviews the lessons learnt during development and use.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

EXPLOITING HIGH FIDELITY SIMULATION FOR AIRCREW COALITION MISSION TRAINING

Heather M. McIntyre & Ebb Smith

Defense Science & Technology Laboratory

Bedford, United Kingdom

Winston Bennett, Jr., Ph.D.

U.S. Air Force Research Laboratory

Human Effectiveness Directorate

Warfighter Training Research Division, Mesa Arizona, USA

In November 2001, DSTL, QinetiQ and the U.S Air Force Research Laboratory (AFRL) completed the first synthetic Composite Air Operation (COMAO) experiment to assess the potential of networked simulation for providing Coalition Mission Training (CMT) for the front-line. The exercise, named trial VirtEgo (the Virt stands for Virtual and Ego stands for a conscious, thinking subject) was designed to satisfy both training and research objectives in a tactically relevant and instructionally valid environment. The operational goal was to prepare RAF pilots for the Combined Qualified Weapons Instructors  (CQWI) Operational Phase two weeks later in Scotland. The CQWI is the RAF’s fighter weapons school. The training research goal was to leverage current US and UK simulation-based training initiatives in a CMT exercise.  This effort examined the coalition training potential of linking the simulation facilities at Bedford (where the UK aircrew and expert White Force were co-located) with the USAF Research Laboratory, Warfighter Training Research Division (AFRL/HEA), Mesa, US where an experienced F-16 4-ship team participated as part of overall combat package. The UK portion was sponsored by Strike Command to provide synthetic COMAO training for students on the CQWI course, (namely 2 RAF Jaguar pilots and 2 Tornado F3 RAF crews). Key instructional features included specifying training objectives based on Mission Essential Competencies (MECs); applying and evaluating distributed planning, briefing and debriefing processes and technologies; and using common training and field evaluation measures with all participants. This paper provides an overview of the design, rationale and data from the experiment. It will also establish the “way ahead” for future coalition training research.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

DEVELOPING EXECUTABLE COGNITIVE ENGINEERING MODELS OF TEAMS

David W. Roberts, Kyle A. Crawford

Georgia Tech Research Institute

and

Center for Human Machine System Research

and School of Industrial & Systems Engineering

Christine M. Mitchell

Center for Human Machine System Research, School of Industrial & Systems Engineering

Georgia Institute of Technology

Atlanta, Georgia

The development of aids to design and evaluate team training are subjects of widespread and on-going research. Some approaches have proposed the use of models of the team to support the evaluation of team training. This paper describes an approach to modeling teams and to developing instructor support tools for team training based on a model of operator-task interactions, the operator function model (OFM). The OFM, and it computational implementation, OFMspert, have been successfully used to model and specify model-based decision-support and intelligent training tools for individual decision makers and operators in a range of applications. 

This research explores extensions of the OFM to teams: the team operation model (TOM) and a parallel extension of OFMspert (i.e. computational) for teams, the computational team model (CTM) explicitly represents the mission and individual team members. With these representations, patterns of behavior are identified and reused.  This paper describes the approach and illustrates it with an Army attack scenario.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

Behaviors that Emerge from Emotion and Cognition:  A First Evaluation

Amy E. Henninger, Ph.D.

Soar Technology, Inc.

Ann Arbor, Michigan

Randolph M. Jones, Ph.D.

Colby College

Waterville, Maine

Eric Chown, Ph.D.

Bowdoin College

Brunswick, Maine

This paper presents an initial evaluation of an emotions model developed for a sophisticated synthetic forces model. Sponsored by ARI, the objective of this research is to make the decision-making process of complex agents less predictable and more realistic, by incorporating emotional factors that affect humans. To this end, we have adopted an approach that promotes the emergence of behavior as a result of complex interactions between factors affecting emotions, integrated in a connectioniststyle model, and factors affecting decision making, represented in a symbolic model.

This paper explains how we used the concept of emerging behaviors to test our framework. This includes a description of the behaviors we used in our prototype, the design of our experiments, a representative set of behavior patterns that emerged as a result of exercising our model over the design space, and our project’s lessons learned.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

JUST-TALK:   AN APPLICATION OF RESPONSIVE VIRTUAL HUMAN TECHNOLOGY

Geoffrey Frank, Curry Guinn, and Robert Hubal

RTI International

Research Triangle Park, North Carolina

Pamela Pope and Martha Stanford

North Carolina Justice Academy

Salemburg, North Carolina

Deborah Lamm-Weisel

North Carolina State University

Raleigh, North Carolina

In this paper, we describe an application of responsive virtual humans to train law enforcement personnel in dealing with subjects that present symptoms of serious mental illness. JUST-TALK provides a computerized virtual person to interact with the student in a role-playing environment.  Students were able to converse with the virtual person using spoken natural language and see and hear the virtual human’s responses–a combination of facial gesture, body movements, and spoken language.  The JUST-TALK project, funded by the National Institute of Justice Office of Science and Technology and developed by RTI International, involved integrating virtual reality training software within a 3-day class at the North Carolina Justice Academy. The course was structured to include classroom-based lecture, videos, discussion, live human role-playing, and virtual human role-playing.  

A scientific evaluation of the class and the software system was carried out by North Carolina State University.   This assessment investigated the contribution of natural language interfaces and virtual reality technology to learning in this applied setting.   Results of the evaluation are extremely encouraging.   The vast majority of students (88 percent) found the simulation easy to use.   A majority of the students said the virtual trainer enhanced their learning in the course.  As a training tool, students rated the computer simulation on par with other training methods including lecture, role-play and discussion. A total of 59 percent of students felt the simulation was better for learning or comparable to role-play; 77 percent felt simulation was better than or comparable to lecture; and 59 percent felt the simulation was better than or comparable to discussion.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

USING MULTIPLE CLIP CENTERS ON SGI HIGH END MACHINES

Michael Rapoport, B.Sc.

Visualization Team Leader IDF GFC BL

Tel-Aviv, Israel

Maj. Amit Sirkis, M.Sc.

R&D Manager IDF GFC BL

Tel-Aviv, Israel

The IDF Ground Force Command Battle-Lab uses a virtual battlefield as a test-bed for new operational and tactical concepts. The virtual battlefield includes numerous simulators, each having several viewpoints of the virtual world.

A single-pipe multi-channel Vega-based application is used as a visualization basis for all Battle-Lab simulators. With introduction of the ClipMap technology that allows circumventing texture memory limitations, running simulations on large terrain databases with geo-specific textures of high resolution became possible.

The major problem that arose was creation and run-time management of several ClipMap regions from existing Vega-based application.  In other words in order to be able to use the ClipMapped texture technology on one hand and to keep the capability of rendering several channels from one SGI pipe on the other hand, we needed to extend our application to support the creation of several ClipMap regions.

This article describes the process and techniques that were used in order to implement multiple ClipMap regions and its limitations, and by such solve the above problem. As a result of our research, we succeeded in obtaining a multi channel application, each channel having its own virtual texture pyramid.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

Image Fusion for Natural Color Visual Simulation AND Materials Classification

Kathleen Minear and Morris Akbari

Imagelinks Inc.

Melbourne, Florida

This paper presents an image fusion algorithm applicable to both visual simulations and materials classification.  Natural colors are preserved enabling color matching between multi-resolution (example: landsat 15m, irs 5m, and quickbird .6m fusions) and multi-image fusions (example: entire countries).  The radiometry is preserved enabling the fused image to be used in materials classification. This approach is demonstrated on data gathered from various sensor platforms (satellite/aerial) and from various sensor types (electro-optical/radar).  With the emergence of new high resolution data sources such as quickbird and ikonos, multi-resolution image fusions for visual simulations have grown in complexity.  Visual simulators require fused data to seamlessly pan from 150 meters to under 1 meter.  With the proposed algorithm, realistic color and texture map generation is not needed. They are inherent in the coarse resolution, multi-spectral, and high resolution, panchromatic, input sources providing natural geospecific terrain rendering.  

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

MEMS-BASED LIGHT VALVES FOR

ULTRA-HIGH RESOLUTION

PROJECTION DISPLAYS

Francis Picard*, Céline Campillo*, Keith K. Niall**, Carl Larouche* and Hubert Jerominek*

*INO      **DRDC Toronto

*Quebec, Quebec, Canada       **Toronto, Ontario, Canada

Ultra-high resolution projectors will improve the visual systems of military flight simulators dramatically.  There are changes in aspect angle and aspect rate which fixed-wing fighter pilots can discriminate at long standoff distances, but which cannot be displayed with adequate resolution by the visual systems of contemporary flight simulators.  At present the limit of display resolution is fixed by the capacity of the display’s projector. This issue is being addressed by INO and its partners working toward the development of a new light-modulating micromirror MEMS. This unique device incorporates 25 µm x 25 µm microbridges acting as flexible micromirrors.  Each micromirror corresponds to one pixel of an image and is capable of modulating light intensity in analog fashion, with switching speeds in the range of 5 µs. A linear array of micromirrors is combined with a scanning system, a microlaser light source and a Schlieren-type optical system to produce a 256 grey-level image. The result is a MOEMS-based system that can write thousands of image lines at a frame rate of 60 Hz. 

Finite-element analyses have been performed to describe mechanical properties of the micromirrors. Several examples will be given from both static and dynamic electromechanical simulation. The micromirror fabrication process will be summarized. The physical characteristics of the micromirrors will be reported, including their response time and damage threshold. Finally, future plans including the development of 2000 x 1 linear pixel arrays with the associated control electronics, will be described.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

LEVELING THE FIELD:

REDUCING DIFFERENCES IN LEARNING

FROM VIRTUAL ENVIRONMENTS

Barbara Barnett and Bruce Perrin

The Boeing Company

St. Louis, Missouri

In recent years, Boeing has investigated the effectiveness of virtual technologies for aircraft maintainer training.  In these studies, we collected data comparing transfer performance following various types of virtual maintenance training to that achieved with more traditional hardware-based training.  These data revealed that performance following virtual environment (VE)-based training was marked by significant inter-subject variability.  In addition, our results indicated that this transfer performance was mediated by the spatial visualization aptitude.  Individuals with higher spatial visualization aptitudes learned effectively from VE-based training, while those with lower visualization aptitude did not.  Results for mockup training showed no similar pattern -- mockup trainees performed equally well on the training transfer task regardless of their spatial visualization aptitude.  The current study further investigates the role of spatial visualization with regard to the effectiveness of an immersive VE-based trainer.  Does the same spatial visualization relationship exist within an immersive virtual training environment?  And, if so, are there additional factors that could help to minimize the impact of spatial visualization, therefore making VE-based training more effective for individuals with a broader range of aptitudes?  Results indicate that training transfer for the immersive VE-based training was similar to that of other VE-based training conditions.  Performance in the training transfer task was highly variable and was mediated by spatial visualization aptitude.  Additional experience factors of general maintenance/tool familiarity and 3-D video game exposure did not significantly reduce the effect of spatial visualization aptitude on training transfer performance.  Although additional practice within the immersive VE prior to maintenance training suggests a trend towards minimizing the association between visualization and post-training performance, the relationship between visualization and post-training performance was still statistically significant.  Implications for future research and training are discussed.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

EMERGING WEB-BASED 3D GRAPHICS FOR EDUCATION AND EXPERIMENTATION

Curtis Blais, Don Brutzman, Jeffrey Weekley,

and LT James Harney, USN

Naval Postgraduate School, Monterey, California

This paper describes current work in the evolution of open standards for 3D graphics on the World Wide Web (Web3D) and provides examples of application of emerging Web3D authoring, visualization, and simulation tools for military education and experimentation.  The paper presents an overview of the current state of Web3D standardization activities, including establishment of conformance tests and a reference implementation.  The paper describes examples of the application of web-based 3D graphics for exploring complex military battlespaces supporting Limited Objective Experiments in Anti-Terrorism / Force Protection (AT/FP) and emerging joint command and control concepts for Web-based information management.

In February 2002, the Web3D Consortium (www.web3d.org) announced completion of the draft of the X3D (Extensible 3D) specification, the proposed next-generation standard for describing 3D content on the World Wide Web (http://www.web3d.org/x3d.html).  X3D is a scene graph architecture and encoding that improves on the Virtual Reality Modeling Language international standard (VRML 97, ISO/IEC 14772-1:1997).  X3D uses the Extensible Markup Language (XML) to express the geometry and behavior capabilities of VRML.  The paper provides a brief overview of X3D and the current status of the standardization process and supporting efforts (e.g., development of an open source scene authoring and visualization tool and conformance test suite development), updating information presented in the I/ITSEC 2001 paper and conference briefing.

To demonstrate the capabilities of the emerging X3D standard, the Naval Postgraduate School is performing research toward development of scenario authoring and Web-based visualization capabilities. The paper describes the current status of research activities, including application of Web-based 3D graphics to a Navy Force Protection Limited Objective Experiment (LOE) and a Joint Futures Laboratory LOE investigating peer-to-peer Joint Interactive Planning concepts. The paper discusses technical challenges in representing complex military operations in Web environments and describes work in progress to demonstrate application of Web-based technologies to create and explore complex, multi-dimensional operational scenarios. The paper concludes with discussion of future research directions for application of Web-based 3D graphics in military education, training, and experimentation.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

APPLYING ACTIVE NETWORKS TO THE JOINT SYNTHETIC BATTLESPACE

Victor Skowronski, Matthew Dorsch, Thomas Kostas

Northrop Grumman Information Technology

Reading, Massachusetts

The Joint Synthetic Battlespace Architecture presents new challenges to Active Networks. Previous simulations had employed federates that performed the same basic function and were relatively self-contained. This allowed the simulation designer to partition the simulation geographically in order to take advantage of Active Networks. The Joint Synthetic Battlespace Architecture uses a number of specialized simulations working in concert. The increased need for communication within such a heterogeneous federation requires new strategies for making maximum use of Active Networks.

This paper describes how a simulation using the Joint Synthetic Battlespace Federation Object Model was adapted to a network using Active Networks. Strategies for assigning federates to hosts are discussed, and how well such assignments succeeded in reducing network traffic.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

THE DEPLOYABLE VIRTUAL TRAINING ENVIRONMENT

Michael P. Bailey and LtCol Robert Armstrong

Tecom USMC, Quantico, Virginia

The Deployable Virtual Training Environment (DVTE) project, sponsored by the Director, Expeditionary Warfare (N75) in the Office of the CNO, is a collaborative effort between the Program Executive Officer, Expeditionary Warfare (PEO EXW) and Marine Corps Training and Education Command (TECOM).  The intended product of this effort is to provide enhanced shipboard operational training simulators on amphibious ships for embarked Marines and to field a flexible, deployable, training system that addresses requirements for combined arms MAGTF and Naval Integration training. DVTE was established to provide a shipboard and shore training system that maintains and enhances embarked Marine war fighting proficiency.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

INTELLIGENT OPPONENTS FOR VIRTUAL REALITY TRAINERS

Randolph M. Jones and Robert E. Wray

Soar Technology, Inc

Ann Arbor, Michigan

and

John E. Laird  and Andrew Nuxoll

University of Michigan

Ann Arbor, Michigan

We are exploring the requirements for and prototypes of intelligent opponents for a virtual reality environment in which U.S. Marines can train to clear buildings of enemy soldiers.  This virtual reality trainer is one of the components of the Virtual Training and Environments (VIRTE) Program of the Office of Naval Research.  Autonomous, intelligent computer generated forces, such as TacAir-Soar, have been shown to be effective in vehicle-level entity domains.  Creating models of individual humans engaged in combat with direct fire weapons presents a host of new challenges.  One challenge for the current project is that intelligent agent development is concurrent with development of the simulation technology by other researchers, meaning the final virtual environment is not yet available for the agents.  Our solution is to use a commercially available, first-person perspective shooter game (Unreal Tournament) as a prototyping environment.  We describe requirements for intelligent opponents, our approach and the current system architecture.  To date, we have developed agents that navigate through buildings and defend themselves.  Creating much more realistic behaviors is a larger challenge and we introduce a number of potential research questions that could, if pursued, increase the realism of the VIRTE agents.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

Implementation of Synthetic

and Instructional Agents in Virtual Technologies and Environments

Ronnie Soles and Stephanie Lackey

Naval Air Warfare Center Training Systems Division

Orlando, Florida

The complex and dynamic environments encountered by our warfighters call for creative solutions to training challenges.  The forward-deployed nature of amphibious forces, practical operational concerns, and perishable nature of complex cognitive skills and team process behaviors create the need to train our Marines and Sailors while at sea.  Virtual Technologies and Environments (VIRTE), an advanced research initiative sponsored by the Office of Naval Research (ONR), aims to develop prototype simulation training systems for deployed amphibious forces.  These training systems will provide the opportunity to refresh team skills and will include various environmental conditions that may impact the training exercises.

VIRTE will incorporate many technologies into the development of these amphibious forces training systems.  This paper will be concerned with development and integration of Computer Generated Forces as teammates (Synthetic Agents) and an intelligent tutoring system (Instructional Agent).  Synthetic Agents (SAs) provide the capability to replace individual crewmembers in training scenarios.  Synther technology creates an atmosphere where the entire crew need not be present in order for one or more crewmembers to participate in meaningful training.  Automated instruction, a key feature of the VIRTE program, is accomplished via intelligent tutoring.  The Instructional Agent (IA) monitors trainee behavior and performance, assesses trainee actions, and provides after action review of trainee performance. VIRTE advances the current state of human behavior representation by integrating these powerful tools to emulate human operators and instructors with technology that uses artificial intelligence to simulate human reasoning.

Synthetic and Instructional Agents will be applied to virtual training platforms of the Navy’s Landing Craft, Air Cushion (LCAC) vehicle and the Marine Corps’ Advanced Amphibious Assault Vehicle (AAAV).  The model development process of the SA and IA will be discussed.  This paper will discuss agent development, progress achieved, anticipated challenges, and plans for performance evaluations.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

AN AUGMENTED REALITY SYSTEM FOR MILITARY OPERATIONS IN URBAN TERRAIN

Mark A. Livingston, Lawrence J. Rosenblum, J. Edward Swan II

Advanced Information Technology, Naval Research Laboratory, Washington, DC

Simon J. Julier, Yohan Baillot, Dennis Brown

ITT Advanced Engineering and Sciences, Alexandria, Virginia

Joseph L. Gabbard and Deborah Hix

Systems Research Center, Virginia Polytechnic Institute

and State University, Blacksburg, Virginia

Many future military operations are expected to occur in urban environments.  These complex, 3D battlefields introduce many challenges to the dismounted warfighter.  Better situational awareness is required for effective operation in urban environments.  However, delivering this information to the dismounted warfighter is extremely difficult.  For example, maps draw a user's attention away from the environment and cannot directly represent the three-dimensional nature of the terrain.

To overcome these difficulties, we are developing the Battlefield Augmented Reality System (BARS).  The system consists of a wearable computer, a wireless network system, and a tracked see-through head-mounted display (HMD).  The computer generates graphics that, from the user's perspective, appear to be aligned with the actual environment.  For example, a building could be augmented to show its name, a plan of its interior, icons to represent reported sniper locations, and the names of adjacent streets.

This paper surveys the current state of development of BARS and describes ongoing research efforts.  We describe four major research areas.  The first is the development of an effective, efficient user interface for displaying data and processing user inputs.  The second is the capability for collaboration between multiple BARS users and other systems.  Third, we describe the current hardware for both a mobile and indoor prototype system.  Finally, we describe initial efforts to formally evaluate the capabilities of the system from a user’s perspective through scenario analysis.  We also will discuss the use of the BARS system in STRICOM's Embedded Training initiative. 

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

A DATA-BASED PARADIGM FOR DYNAMIC, INTERACTIVE GRAPHICAL HUMAN MACHINE INTERFACES

Mark Snyder

CG²

Glendale, Arizona

Graphical human machine interfaces (HMIs), which for simulation systems can include instrumentation, cockpit displays, virtual controllers, and interactive operator stations can be among the most time and resource consuming parts of a training system development.  Typical problems encountered in developing and integrating graphical HMIs include low quality graphics that do a poor job of representing ‘the real thing’, inflexible software environments, changing requirements, and tool limitations that force developers to ‘just code it by hand’.  Code generation tools are sometimes employed to try to meet these requirements, but reuse of the generated code in multiple simulation and training applications can prove difficult.  A data format based approach, such as is used for 3D models in visual systems, would seem to be a good solution for creating graphical HMIs that could be easily reused and even treated as commodities, like 3D databases are today.   Today’s graphical HMIs, however, have several unique characteristics that make existing modeling and scene graph systems a poor fit for this application, such as complex interaction requirements, advanced special-purpose graphics, and extensive data interfaces.  This paper describes a system of tools, data formats, and runtime libraries that make it possible to treat high quality graphical HMIs as databases and reuse them across a wide range of applications.  Furthermore, the database system supports generating graphics as C code for embedded system usage if required.  Several example applications using the system are presented and the results are quantified in terms of application development speed and ease of reuse.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

VIRTUAL COCKPITS UTILIZING RETINAL SCANNING DISPLAYS

Jack D. Clevenger

Microvision

Bothell, Washington

and

Thomas B. Adams

Quality Research

Huntsville, Alabama

The Virtual Cockpit Optimization Program (VCOP) focuses on optimizing the workload of the pilot in today’s and future advanced military aircraft.  The concept of the virtual cockpit program is to provide the pilot with information such as sensor imagery, flight data, and battlefield information in a clear, non-confusing and intuitive manner to increase situational awareness, thus making the aircraft easier and safer to fly while also improving mission performance.   The Virtual Cockpit Optimization Program recently completed a simulation demonstration and Human Factors evaluation of the integrated advanced technologies in a rotorcraft simulator at the Army’s Advanced Prototyping, Engineering and eXperimentation (APEX) Laboratory at Redstone Arsenal, AL.  VCOP will continue to utilize simulation to evaluate the effectiveness of the technologies as it progresses towards a flight evaluation.

The majority of the VCOP activity involves the integration of advanced, independently developed technologies into a single system that represents a significant leap ahead in cockpit design philosophies.    Rather than concentrating on the aircraft and how it can be retrofitted to meet the needs of the next generation warfighter, VCOP furnishes pilots with the necessary enhanced capabilities to perform their jobs more efficiently.  VCOP comprises the following six independently developed technologies:

1.  Full color, high resolution, high brightness helmet-mounted display (HMD) that incorporates Retinal Scanning Display (RSD) technology

2.  Three dimensional (3D) audio

3.  Speech recognition

4.  Tactile situational awareness

5.  Intelligent information management

6.  Crew-aided cognitive decision aides

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

CIGI, A COMMON IMAGE GENERATOR INTERFACE

Robert Lechner and Willard Phelps

Boeing Training Systems and Services

St. Louis, Missouri

For years, image generator suppliers have been developing real-time interfaces for their respective products. While these interfaces may be specific in nature for a given system, they all share common attributes and controlling methods that allow a host simulation to communicate to the image generator. Unfortunately, these interfaces are deemed proprietary to the respective companies. Thus each time a new image generator comes along, the integrator must perform a non-recurring effort to integrate the new image generator with their system. The integration effort includes interface software design and testing as well as debugging particular issues that evolve during integration. Of certain benefit to the simulation community would be a standard interface that can be assimilated among the industry that provides a generic communication methodology and controlling mechanism for any image generator.

This paper will present the development, design, and distribution of a Common Image Generator Interface (CIGI) that encapsulates a generic interface and provides standard objects for control within an image generator. The purpose of CIGI is to overcome the proprietary nature of existing interfaces by providing it open source, and allow for a plug-and-play solution for any image generator application. CIGI is based on a building block process. Each building block is generic in nature and represents a related group of data. These generic building blocks are used to specify such things as high-level image generator commands, out-the-window view portals, entities and special effects, atmospheric effects, mission functions, and sensor simulations. By mixing and matching the available building blocks the user can create full-featured simulations.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

An Antialiasing Quality Metric for Visual Simulation and Training

Jeffrey D. Potter
Quantum3D
Orlando, Florida

Now that even the lowest cost image generators provide some form of hardware-accelerated antialiasing, there is a need for a quantitative ranking of their quality. Previous efforts to measure rendered image quality tend to be broad, subjective, and based only on static scenes. Their standard of reference is photo-realism. However, in the visual simulation and training domain, what’s most important is not photo-realism per se, but the “suspension of disbelief.” The goal is for the maximum amount of training to occur while minimizing the rendering anomalies that cause negative training. This paper proposes an antialiasing quality metric targeted specifically for visual simulation and training. The proposed metric pulls together the properties of visual perception and the classic measures of Kell factor, MTF, and Johnson’s Criteria to suggest an improved measure of a rendering subsystem’s ability to provide the necessary level of quality to train the human subject.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

Dynamic Terrain Modeling and Simulation for Simulators of Soil Handling Construction Equipment

Norbert Härle

STN Atlas Elektronik GmbH

Bremen, Germany

Armies use special equipment to enable access ways to rivers, to remove debris and barricades, and to build bulwarks. Training on this equipment in Germany is difficult because of the many environmental regulations. Especially, building an access path to rivers is forbidden because of the severe impact onto the environment. Bulldozing and excavating is restricted to small areas where the soil has been moved for many years and lost the consistency, which is normally found in nature. Handling simulators are a way to enable realistic training and training on working near rivers.

Such a simulator requires many features, which are additional to traditional simulators. It needs dynamic terrain such that the image generator is able to display a change in the environment as well as a change in the database such that the change in the terrain is stored and retrieved when the position is accessed later again.

A big challenge is the modelling of the soil mechanics such that the deformation of the ground is displayed correctly and the force feedback to the tool and vehicle is calculated realistically.

A project was funded by the German Government and STN Atlas Elektronik to investigate the feasibility for building such a simulator. During this project, a model for soil mechanic simulation, an image generator for dynamic databases and a storage mechanism for the changes in the database was developed and successfully implemented as prototype.

The paper will discuss the results of the methods, which can be used for this kind of simulator as well as present the architecture and simulation results of the prototype.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

AUTOMATIC CREATION OF INTELLIGENT SIMULATED ENTITIES THROUGH CONTEXT BASED REASONING AND GENETIC PROGRAMMING

Hans K. Fernlund and Avelino J. Gonzalez

Intelligent Systems Laboratory

School of Electrical Engineering and Computer Science

University of Central Florida, Orlando, Florida

Context-Based Reasoning (CxBR) has recently made significant progress in the area of modeling Computer Generated Forces in training simulations. Context-Based Reasoning is based on the concept that humans think and act in terms of contexts. This approach has been proven to work well in simulated environments. One problem has been that the time and effort required for gathering and analyzing knowledge from subject matter experts for the development of intelligent entities (i.e., models of tactical human decision-making) is high. To be able to reduce these costs, new methodologies and tools must be developed that automate the creation of these simulated entities. We describe in this paper one way to accomplish this through the use of Genetic Programming in conjunction with CxBR. Genetic Programming (GP) addresses computer programs that evolve new, better programs by themselves, i.e. automatic programming. GP will genetically breed a population of computer programs using operations and selection mechanisms inspired by Darwin's evolution theory. The result of the genetic process is a computer program that will solve most types of predefined problems in almost any area such as classification, planning, mathematic, optimizing, control. This paper presents a new approach for automatically creating intelligent simulated entities. To automate the creation process, learning from observation is used. This strategy learns the behavior of a subject matter expert by merely observing his behavior. The need for a methodology like this has already been identified for some applications, e.g. after-action-review programs, modeling human behavior in battlefield simulations and in street traffic flow simulators. By developing a tool to automatically create models of simulated entities, a new area of rapid creation of such entities could emerge.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

Developing Joint Operations on Urban Synthetic Terrain (JOUST) Technology

LtCol Chris Hadinger

Defense Modeling & Simulation Office

Alexandria, Virginia

Kevin Roney

Science Application International Corporation

Arlington, Virginia

The Services currently have very limited opportunity to conduct live joint urban operations training and testing due to facility limitations and range safety and environmental concerns. Since the US military is increasingly likely to conduct operations on urban terrain, this deficiency must be addressed. Developing a distributed range capability for urban operations appears to provide the best solution to this problem. The Services have separate initiatives totaling over $200M beginning in FY04 to build separate Service-unique urban operations facilities. At this time, the plans for these facilities are not well integrated and no concept currently exists to integrate live, virtual, and constructive training assets in a way that provides meaningful training.

The Defense Modeling and Simulation Office (DMSO) is conducting research into integration of simulation technology and distributed range capability for urban operations. The overall objective of JOUST is to allow the joint warfighter to conduct training and test operations for urban warfare using distributed live test and training capabilities as well as distributed simulation technology. JOUST will integrate Naval, Air Force, and Army assets in a mixture of live-fire ranges, military operations urban terrain facilities as well as, virtual and constructive simulations.

This paper focuses on the challenges of creating a cost affordable solution for distributed joint urban operations testing and training. The keystone issue here is to define a distributed system component architecture that will leverage DoD investments in both live-fire ranges and simulation technology. Our approach, to develop a prototype Joint Urban Operations Distributed Synthetic Range (DSR), is based upon working through various architecture alternatives and defining fundamental assumptions. We will also present our near term JOUST objectives along with a set of defined use cases for FY03 experimentation. The results of our efforts will be valuable in defining a reusable joint urban operations test and training environment.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

DISPLAYING THE BATTLESPACE ENVIRONMENT

EVOLUTION OF A HEAD-MOUNTED DISPLAY

Jim Melzer

Product Manager, Head-Mounted and Miniature Displays

Kaiser Electro-Optics, Carlsbad, California

Rita Simons

Visual Systems Engineer, AVCATT-A

United States Army STRICOM, Orlando, Florida

In today’s increasingly sophisticated simulation world, a realistic, high intensity, task-loaded display of the battlespace environment has become the expectation of the military user community. To effectively train aviation aircrews, the visual system must support realistic collective/combined arms training with the required fidelity to fly nap-of-the-earth (NOE) or conduct multi-ship operations. Meeting this high level of expectations requires that the visual system be capable of performing all necessary collective tasks and supporting individual tasks with no negative training transfer or physical impacts on crewmembers. An evolving technology that can meet these needs is the Head-Mounted Display (HMD). This paper will address the use of HMDs in aviation simulators and will follow the evolution of the SIM EYE XL 100A from its early Wide Eye™ stage, to its current use in the Army’s Aviation Combined Arms Tactical Trainer – Aviation (AVCATT-A) Reconfigurable Manned Simulator program. Finally, it will address possible future improvements that can be incorporated into the HMD to further satisfy Army aviation users.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

IMPROVING PERFORMANCE MEASUREMENT WITH VIRTUAL ENVIRONMENT TECHNOLOGY:   VIRTUAL METRICS

Elizabeth Sheldon, Ph.D.

John Burns, Ph.D., Scott Casey, and Janice Giebenrath,

Sonalysts, Inc., Orlando, Florida

LT Joseph Cohn, Ph.D.

Naval Research Lab, Washington, DC

Robert Breaux, Ph.D.

Naval Air Systems Command Training Systems Division

Orlando, Florida

Virtual Environment (VE) training systems are fast becoming the tool of choice for supporting a wide array of military training needs.  As the technology advances, this trend will become even stronger, with these systems ultimately competing with more traditionally accepted methods of training for precedence in assorted curricula.  One area in which this faith in VE systems is warranted is performance measurement.  In particular, the data collection capabilities afforded by VEs readily lend themselves to the implementation of new, advanced measures of performance that were not possible to develop in the operational environment.  The critical challenge that developers now face is identifying and validating these Virtual Metrics and determining how best to include them in the ever-expanding array of evaluation tools available to the operational training community.  

The current paper explores this process, using the research, development, and evaluation phases of the Conning Officer Virtual Environment (COVE) shiphandling training system as a case study.  The process to be described progressed in three stages, each of which resulted in the testing and evaluation of a novel piece of technology, which ultimately led to Virtual Metrics, Alongside Box and Command Efficiency, for evaluating shiphandling performance.  In the first stages, the data collection capabilities of the COVE testbed enabled the development of an objective quantification of basic UNREP performance, which is traditionally subjective in nature, using “track histories”.  In the final stages, the evaluation team used these basic measures to develop the more sophisticated Alongside Box and Command Efficiency measures, which serve collective variables for quantifying shiphandling performance.  Finally, these Virtual Metrics were implemented to quantify the transfer of skills learned in the COVE system to the real world UNREP task.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

LOGIC AND STATISTICAL MODELING FOR LANGUAGE UNDERSTANDING

Bradley Cope and Stephen Boemler

NAVAIR Training Systems Division

Orlando Florida

Team leader training applications often include critical decision making based on multiple streams of communications where an overload, or interference from secondary sources, can occur. Artificial intelligence based parsing of language could aid in the presentation of the information contained in communications. Semantic and pragmatic context of a group of messages can be restricted in terms of specific domain information relating to events. Further, rule based software written in programmable logic can model the rules of the domain phraseology.

Computational linguistics makes use of syntactic parsing whereby a stream of symbols is analyzed for its conformance to some specific set of rules. For instance, if you read the previous sentence, you "syntactically parsed" it according to some grammar of "English" and determined that it was grammatical. In the simplest terms, parsers analyze some target string of symbols according to some specified grammar.

In a probabilistic associative chain, the occurrence of each word is determined by the immediately preceding word or series of words, where each response serves as a stimulus for the next response. In bounded phraseologies such as those used in training devices, it is possible, using programmable logic, to model grammars. Similarly, the association of words in very large grammars can be modeled statistically. Two innovative computational linguistic approaches were investigated and presented in this paper, the cross entropy syntactic identifier and the Markov model syntactic parser. The cross entropy syntactic identifier identifies phonetic similarity and the Markov model syntactic parser adds variability to otherwise rigid parsing techniques.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

SEMI-AUTONOMOUS FORCES INVOLVING ROBOTIC ENTITIES RESEARCH RESULTS

Dr. Peter Drewes

David Rees

Science Application International Corporation

Orlando, Florida

Computer generated forces have modeled unmanned systems for several years. However, this modeling is usually in support of manned systems instead of the concentration on the robotic system as a cohesive team member.  Robotics research has been focused on navigation and sensor environments within limited simulation domains. There has been little emphasis on effective utilization of modeling and simulation capabilities to create robust environments to test future unmanned systems’ performance, parameters, and mission effectiveness.

Research at SAIC concentrated on the group behavior of robotic entities operating in a live environment, connected and inter-operating in a constructive Computer Generated Force (CGF) environment. The Semi-Autonomous Forces Involving Robotic Entities (SAFIRE) research project involves operation of robotics within a live (real world) environment. Information and behaviors are fed back to a “monitoring station” where they are analyzed and presented. This behavioral testbed provides the opportunity to leverage traditional CGF system behavior insertions as well as determine limitations and directions. During the second year of research, the focus shifted towards the effective use of commanding/controlling the unmanned systems in a mixed mode environment (both autonomous and human control) as well as providing a more robust simulation environment. This environment consists of “plug and play” physical characteristics for testing new unmanned system designs, in addition to providing live feedback into the constructive simulation. These features allow a more complete training and analysis concerning the unmanned system’s mission capabilities, detection, and survivability under various conditions.  This paper is a lessons learned from the research project.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

Software Reusability in the

Battle Lab:   Developing a Test-Bed for Evaluation of C4I Tactical Systems

Einav Kiperman, M.E., Amit Sirkis, M.Sc.

Israeli Defense Force Ground Forces Command Battle Lab

Tel Aviv, Israel

Parallel with the desire for developing Command, Control, Communications, Computers, and Intelligence  (C4I) systems at the low tactical level, there evolved a need for a tool that allows characterizing and evaluating these systems. Such a tool should also support the development process of the C4I systems and allow evaluating new operational concepts.

In the last few years, the Israeli Defense Force Ground Forces Command Battle Lab (IDF GFC BL) succeeded in synthesizing technical capabilities and a methodological concept that enables developing a virtual test-bed for the evaluation of C4I systems.  This concept has been successfully implemented by the BL to investigate and evaluate C4I systems on several occasions.  The test-bed is based on the creation of a virtual battlefield into which the tactical forces, using the C4I systems, are immersed. In this article, we describe the various infrastructures used for creating, in virtually no time, such a test-bed at the platoon and company levels. The main infrastructure modules include the Image Generator (IG), Computer Generated Forces (CGF), application infrastructure, radio infrastructure, and C4I infrastructure.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

USE OF HPC PARADIGMS IN ENTITY BASED SIMULATIONS

Dr. David R. Pratt

Science Applications International Corporation

Orlando, Florida

The High Performance Computing (HPC) community has developed several processing paradigms that warrant examination as future software architecture foundations for closed-form and limited interaction, entity-based simulation systems.  The four most important of these are the vector processor, the shared memory processor, the Beowulf Cluster, and the network of workstations (NoW).  Understanding the processing model is critical to ensure systems developed to run on these types of machines have the software architecture to complement the underlying hardware architecture.  Since the software architecture is derived from the problem space, it is important to understand the class of problem each type of HPC machine is designed to solve and how it can help solve the user’s problem.  We have chosen to focus on systems with limited or no user interaction to avoid the limitations most HPC system have dealing with a large number of concurrent users.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

USE SIMULATION TO DESIGN, DEVELOP, AND OPTIMIZE DISTRIBUTED SIMULATIONS

Cal Cluff and John Kirk

Illgen Simulation Technologies, Inc.

Goleta, California

This paper describes the philosophy, structure, and use of a simulation of a distributed simulation.  It has long been recognized that great engineering value is obtained by using simulations in place of systems before they are built or when the systems are unavailable, thus the current interest in Simulation Based Acquisition (SBA).  Now, many simulations are as (or more) complicated than the systems which inspired the use of simulation in the first place.  This raises the question: Why try to build these complicated simulations without the benefit of simulation?

Illgen Simulation Technologies, Inc. has developed a simulation of the performance costs for large-scale distributed discrete-event simulations.  The purpose of this simulation is to answer sizing and performance questions based on choices of: Processing and network hardware and topology, Runtime Infrastructure (RTI) management strategies, and load balancing and distribution of processing requirements.

This simulation is based on a simple assumption: For performance analysis purposes, the discrete-event simulation modeled is just load.  It is load on the RTI, network, and processor resources.  Performance, then, is measured by how well the modeled simulation architecture (both hardware and software) handles its load.  The model of a distributed simulation is thus the close coupling of three simulations: A simulation of the hardware, a simulation of the RTI, and a simulation of the processing load.

The simulation has been calibrated using a number of hardware platforms, RTIs, and federates used for distributed, large-scale, discrete-event simulations.  The paper will include calibration results and typical analyses using this approach.

This paper is available on the 2002 I/ITSEC CD ROM. Order it from I/ITSEC'S Website