Voice-Enabled Agent for Realistic Integrated Combat Operations Training

ABSTRACT: ORION will develop and demonstrate voice-enabled intelligent agents to improve the realism of integrated combat operations training and rehearsal. We will implement them in the Joint Theater Air Ground Simulation System (JTAGSS), an Air Support Operations Center (ASOC) simulation, but they are adaptable to a wide range of military and commercial simulations. We will build upon the voice-enabled agents we developed for JTAGSS using a sparse primary lexicon with a statistical language model that gives highest probability to words in a primary lexicon of mission-specific keywords and phrases such as callsigns. A two orders of magnitude larger secondary lexicon with lower probability n-grams limits false positives. Agent behavior uses the same method we previously used in an earlier project for AFRL The voice output uses a multi-step process, adjusting the phone and intonation symbols with regard to context to produce more natural speech. Our open architecture facilitates substitution of other speech recognition, communications, behavior, and speech synthesis modules and we provide utilities for user tailoring of lexicon and language models. Our agents will be invaluable enhancements to training systems for today"s asymmetric combat operations in highly integrated. interoperable battlespaces. BENEFIT: The utility of a synthetic training environment for Air Support Operations Center (ASOC) personnel is constrained by the personnel requirements needed to staff a training session. This system will allow some of the human role-players to be replaced by voice-enabled synthetic agents, drastically reducing the cost and logistics. The use of agents will also facilitate using the simulation in ways other than straight procedural training, thereby enhancing the quality of the experience. Agents will initially play roles external to the ASOC, but further development of our technology will allow agents to play internal roles as well. The agents use industry standards and can be incorporated into other training and rehearsal simulations. In conclusion, the benefits are better training at lower cost. In the global commercial markets , the development of realistic synthetic agents that can engage in logical dialog, albeit in limited realms, can be incorporated into dialog-interactive intelligent (DII) games and inspirational educational/tutorial applications for children . These represent nascent market sectors that have great global benefit for mankind.