The need for human trainers to role-play the participants with whom a trainee must interact limits the effectiveness of both simulators and part task training (PTT) devices for individuals and small groups. The objective of this research is to design and prove the feasibility of our architecture for speech-enabled, synthetic role-players for a generalized PTT device or simulator for P-8A Tactical Coordinator (TACCO) tasks. We will adapt and build on speech-enabled agents we have developed for an on-going Joint Theater Air Ground Simulation training system. Our approach uses ORION"s revolutionary T-BORG simulation and integration framework, which allows disparate programs to intelligently use the data generated by one another with low computational overhead. In the Base portion we will develop standalone agents that play the roles of Sensor Operators to carry on doctrinally-correct conversations with a human playing the role of a TACCO. Scenarios, dialogs, language model and lexicon, and behaviors will cover a typical maritime patrol mission. In the Option portion we will add simple physical models of the aircraft and one or more targets and network them via high-level architecture (HLA). This will enable the agents to provide simulated sensor data on targets upon verbal request from the trainee.
Keywords: voice-enabled, voice-enabled, Architecture, role-playing, Simulation, synthetic agent, Training Device
