ATF and Raging Tiger AI

[Patrick Proctor]

I have had this question more than once, in varying forms, so I thought I would post the response here, for posterity. Enjoy

********************************

>I have noticed that the enemy forces seemed to
> ignore my troop movements, sticking to their original game
> plan despite
> my actions (and we got right in their faces). Isn't the AI supposed to make them
> react to my movements and behaviors?

I would have to know which scenario you are talking about to give a more detailed answer, but I can give some general observations that may clear up any confusion...

There are four levels of enemy AI in the ATF Engine (and, by extension, Raging Tiger).

1. Unit-Level. This is the lowest level of AI. This governs how individual units and teams react to fire, suppress targets, select targets for engagement, and react to obstacles and other forms of contact. It can be customized to an extent, by the scenario designer by using "SOPs" such as "engage on contact", "bypass obstacles", "suppress", and "hide on contact". The player also has limited access to this AI layer, via these SOP functions.
2. Hierarchy-Level. This is identical to the "Mission" and "Formation" orders that the player can give during game play. This AI, given input such as objective locations, desired formations, and desired routes, uses complex, branching scripts to replicate the way companies and platoons in real combat execute tactical missions and formations. The AI uses the guidance it is given and a series of decision point triggers to do its best to execute hierarchy-level instructions.
3. Global Fire Support. This is a special, enemy-and-neutral-side-only, AI layer that uses complex engagement rules and tactical rules to mass enemy artillery against the player's forces. It attempts to determine, by use of enemy unit locations and enemy hierarchy missions, what the "focus of fires" should be, and then selects and engages targets that are appropriate to this focus. This AI can be overriden for individual units or hierarchies by changing the status of units so they will not fire (using the "Do Not Emplace" SOP), or planning "TOT" targets.
4. Global Faction AI. Both enemy and neutral factions also have this AI layer. This is a global "Script" with branches and sequels that allows the scenario designer to coordinate his forces. It is designed to replicate the way that battalion and regimental commanders plan operations for their forces, by using operational phases and triggered decision points. The AI is not "god-level" or "omniscient". If the player, for example, is successful in defeating the enemy's reconnaisance effort, it will impair the enemy faction's ability to make decisions about player disposition or intent (just as in real life).

Again, it is difficult for me to say what you are observing, as I do not know what scenario you are playing, or what situation is occuring. But here are some reasons that the enemy Global Faction AI might not seem to behave "intelligently":

1. You have outsmarted the Enemy commander. The scenario designer, just as real planners, in real military conflicts, must make plans based on the "most likely" opposing course of action. If you have found an innovative approach, or an original course of action, you may have out-witted the enemy battle plan. A motorized rifle regiment is somewhat like a battleship, in that it cannot "turn on a dime". On the highly lethal modern battlefield, a surprised force is usually wiped out before it can adjust. Congratulations!
2. The enemy does not consider you a threat. It is possible that the scenario designer has decided to "accept risk" at the point where you have made your "breakthrough". The battleplan may call for simply "fixing" forces in the area where you are "right on top of him". If you stop, even if you wipe out the force you are facing, that force has accomplished its task (which was to get you to stop). This is a common doctrinal task for a support force in FSU (former soviet union) tactics.
3. You have run "off of the page". The Global Faction AI is designed within the constraints of the tactical problem. That is, the enemy force battle plan is designed with YOUR forces restrictions in mind. That means that he expects you to stay in your boundaries (as, notionally, he has forces on his left and right to protect his flanks). If you move outside your boundaries, you can completely flank the enemy force and he will ignore you until you are behind him (at which point it is too late). IRL (in real life) you would never be able to do this because a) you would run into another enemy force, in your adjacent unit's sector, b) you would likely be a victim of fratricide from your adjacent unit, and c) you would get your *** chewed by your commander after the battle for not following his orders (of which the boundaries are a part).
4. The faction AI models a "stupid" enemy. Perhaps stupid is too harsh a word, but some countries have militaries that are not as well trained, or cannot command and control their large formations as well as the US Army can. The Iraqi Army (pre-OIF) was such an Army. As such, the enemy forces in the Iraqi campaign feature less complex battleplans, with less flexibility to react to palyer actions.

I hope this helps.