Artificial Intelligence Techniques for SAF Based on Potential Tactics

The current Semi-Automated Forces (SAF) OPFOR does not provide a realistic interactive opponent for the Close Combat Tactical Trainer (CCTT). This problem exists because the SAF in CATT only encodes Soviet doctrine utilized in Central Europe. The U.S. Army does not anticipate meeting an army based on Soviet doctrine in Central Europe, therefore the U.S. Army requires a more varied SAF OPFOR. The unbounded domain of all potential infantry actions precludes an exhaustive state space search for deriving the optimal tactics, so a modern SAF OPFOR should be based on the varied doctrines and terrains that exist in the world. To do this, requires artificial intelligence techniques for the rapid implementation and merging or tactics from a large pool of possibilities. The best potential tactics are dependent on weather, equipment, terrain, and may other factors. The amalgamation and choice of tactics will give an effective an unpredictabe SAF OPFOR based on the conditions and the equipment OPFOR is given. Friendly SAFs must also behave in a realistic manner in a wide variety of situations for an effective training experience, so for all SAFs, a rapid easy method of producing realistic behavior must be achieved. In addition to CCTT, other simulations, including DIVE will require automated infantry agents. SHAI and RDC are already will acquainted with these projects and the capabilities necessary in SAF for an effective training environment. The prototype automated agents proposed could be interfaced to either simulation or any other simulation at the Army's request. In addition the mechanism for rapid production of agents proposed could be used for future simulations and future modules of existing systems.