Under the Army’s Future Combat Systems (FCS) concept, the warfighter manning a Control Vehicle (CV) crewstation must maintain situational awareness and apply tactical decision-making principles in a heightened information-rich setting with distributed vehicles and sensors under his command. This paper discusses a proof-of-concept Intelligent Tutoring System (ITS) to provide scenario-based practice for the FCS soldier. In this context, a limited principle hierarchy serves as the instructional basis for the training system and the automated evaluation of student actions in an FCS scenario. Embedded training systems for this domain must be integrated with a variety of software packages using a common protocol. This system communicates with the OneSAF Test Bed (OTB) simulation environment, and the control interface for networked robotic vehicles under the student's command. In addition to the fundamental tactical principles, students are also monitored for their mastery with the task of translating tactical intentions to robotic commands correctly executed in the control interface. The ITS observes the student's actions and performance in a simulated scenario and produces specifically tailored feedback on principles executed correctly and incorrectly. Design issues for the development of an ITS for the FCS domain also include the need to facilitate scenario authoring, and the objective of providing a flexible architecture that can switch between real-time feedback during scenario execution versus strictly after action review. This proof-of-concept system aims to provide a foundation for future training systems based on the same architecture, but supporting team training on multiple scenarios with multiple simultaneous participants.

This paper is available on the 2003 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

AN INTELLIGENT TUTORING SYSTEM FOR REMOTE SENSING AND IMAGE INTERPRETATION

Aaron M. Bell

Stottler Henke Associates, Inc.

San Mateo, California

Remote sensing technology is playing an increasingly central role in military operations. The interpretation of satellite and aircraft photography has great utility in allowing forces to assess a situation from a secure, stealthy distance. However, performing effective interpretations of remote sensing photography is a complex science. Knowledge of a complex, highly-visual domain, such as that of remote sensing, cannot be learned from manuals. For an analyst to become proficient requires a large investment of training time from an expert.

Our paper describes an Intelligent Tutoring System (ITS) being built for NASA. The goal of this software is to provide the benefits of a one-on-one remote sensing instructor to teach Earth Science principles. The ITS integrates directly with the NASA Image2000 image processing software, the actual end environment in which researchers perform real-life tasks. Together, NASA Image2000 and the ITS provide students with a dynamic, hands-on educational experience that is cost-effective. It monitors and assesses the student’s actions, provides immediate feedback as well as contextual guidance and challenges. A separate authoring tool allows new tutorials for the ITS to be developed by non-programmers. Thus, instructors can readily produce interactive, adaptive tutorials for an area of remote sensing interest. With regard to military application, tutorials can be constructed to teach the wide array of interpretation tasks: image enhancement, object classification, target discrimination, damage assessment, weather prediction, etc. The tutorials can focus on a specific geographical location.

This paper outlines our research of a remote sensing ITS system, its general framework, the artificial intelligence technologies it employs to model the knowledge of the student and the expert, and its application to the Earth Science domain.

This paper is available on the 2003 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

TECHNIQUES FOR AUTOMATIC AAR FOR TACTICAL SIMULATION TRAINING

Richard Stottler

Stottler Henke Associates, Inc.

San Mateo, CA

One of the limitations which prevents more effective use of tactical straining simulations is the need for instructors or observer/controllers to observe the student’s actions in the simulation and then debrief him. Many tactical training situations require this After Action Review (AAR) to ensure that the student actually learns from the experience. Thus an instructor is required to observe each simulated scenario, limiting the number of such scenarios that can be played. An automatic AAR capability therefore greatly increases the number of scenarios each student can perform. But the development of this capability is very challenging, principally because in a free play simulation, there is an infinite number of possible outcomes, at least at the most detailed level.

Two techniques have been found to be very useful in the development of automatic AAR capabilities for tactical simulations. The most common type of tactical decisions are made in real-time during the execution in a simulated scenario. General and scenario-specific Behavior Transition Networks (BTNs) have been shown to be highly adept at determining the correctness of student decisions in real time free-play simulations in a variety of domains including Navy, Army, Air Force, and Marine Corps training applications.

Another type of tactical decision occurs during the tactical planning process that results from the receipt of an order to perform an upcoming mission. The result of this process is a tactical plan which is intended to be executed to meet stated objectives. It has been shown that evaluating the correctness of the various aspects of the plan can be accomplished through the comparison to previously entered and annotated plans specific to particular scenarios. These plans are typically both good and common bad plans.

This paper describes both techniques in detail as well as their application to Navy, Army, and Air Force tactical decision-making domains.

This paper is available on the 2003 I/ITSEC CD ROM. Order it from I/ITSEC'S Website

NAVAL GUNFIRE TRAINING WITHOUT THE TRAINING RANGE

James R. Cooley

AAI Corporation

Hunt Valley, MD

The political storm following the tragic live-fire training accident on Vieques Island in Puerto Rico resulted in the imminent closure of that facility and the urgent need for a replacement. One solution rapidly emerged that involves the integration of several existing technologies together into a system known as the Virtual At-Sea Trainer (VAST). The VAST system supports both live fire training into a synthetic environment and simulated gunfire into the same simulated environment. The VAST system consists of four major, independently developed subsystems that were combined into an integrated whole to provide this type of training. These systems are the Integrated Maritime Portable Acoustic Scoring and Simulator (IMPASS) acoustic sonobuoy system, the Battle force Tactical Trainer (BFTT), the portable BFTT system also known as the Carry-on Combat System Trainer COCST, a MK 45 gun simulator developed for the Device 204 pierside trainer, and a commercial, PC-based forward observer trainer. This paper describes the VAST system, the integration process and the technologies incorporated into the VAST system.

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