CASE STUDIES

Team Coaching Assistant for Simulation-Based Training

T-CAST
Training scenario simulated within the OnLine Interactive Virtual Environment (OLIVE). Picture courtesy of Forterra Systems.

The Problem

The U.S. Army Research Development and Engineering Command’s Simulation Training Technology Center (RDECOM STTC) needed technology that automatically recognizes key events that occur during simulated exercises. They wanted a system that could support after-action review (AAR) in OLIVE (OnLine Interactive Virtual Environment), a multiplayer training simulation developed by Forterra Systems.

The Solution

Domain Analysis: Aptima conducted a domain analysis to develop training content. Field manuals describing basic operational goals and common tasks executed during urban operations were reviewed. Army subject matter experts (SMEs) were interviewed to identify key training objectives, focusing on decisions and actions associated with house-to-house search missions. Next, standard measures of performance and collected examples of errors that inexperienced soldiers often make were documented.

Model Development: Aptima used the analysis to create a framework that captures relationships between training objectives and operational tasks, and the typical situations in which soldiers might execute these tasks. Specific key events were identified — the main goal of the project — that signal when these situations are occurring and that provide information about how trainees are responding to them. Next, the OLIVE data model was examined to locate elements that provide reliable indicators of these events.

System Implementation: Aptima built an expert system called T-CAST – Team Coaching Assistant for Simulation-Based Training. T-CAST includes “rules” that infer the existence of key events from patterns of simulator data represented as time-stamped “data facts.” T-CAST determines that a given event has taken place when data facts match conditions stated within a corresponding rule. BBN Technologies developed the T-CAST spatial analysis module, which reports locations of “avatars” (simulated entities), identities of objects within avatar line of sight, and other space-based computations.

Integration and Testing: Working with Forterra Systems, T-CAST was integrated into the OLIVE system. T-CAST extracts data from OLIVE, transforms data into facts, computes spatial analyses, and then outputs a list of recognized events. OLIVE imports this list and uses it to “bookmark” the AAR simulation for replay during AAR. A small set of scenarios based on the content developed through the domain analysis was used to test the integrated system. T-CAST reliably recognized key events generated during test scenarios, verifying its accuracy.

The Results

T-CAST represents a general solution for recognizing key events in simulator data. Aptima is currently developing the capability to recognize events in near-real-time from data transmitted at runtime. A SCORM (Shareable Content Object Reference Model) interface that will enable interoperation with other SCORM-conformant systems is also under development.

For more information, check out our article Integrating team experiential learning into SCORM-conformant training from I/ITSEC 2007.