Does anybody know of any good texts (online articles or books) on military intelligence at the strategic, tactical, and operational level? I''ve done searches at amazon.com, google.com, yahoo.com, and ask.com and can''t find anything on actual tactics or strategy. Most of what I find is ''find the enemy weakness and exploit it''. Not specific enough to be helpful for programming a tactical AI at all.
"I believe; therefore, it is." -True Perception
"The Requested Information Is Unknown Or Classified" -Anonymous

The Art of War?

(Hey, it had to be said!)

[ GDNet Start Here | GDNet Search Tool | GDNet FAQ | MS RTFM [MSDN] | SGI STL Docs | Google! | Asking Smart Questions ]
Thanks to Kylotan for the idea!

Well, I guess "Find an opponents weakness and exploit it" is a generalization of Sun Tzu''s strategy. Basically you want to get as much information as possible so you can figure out how your opponent operates. Then you take an offensive stance that is capable of defeating that mode of operation. Also you have to attack in an unexpected way through use of decoys or surprise tactics. If you read the book you''d probably be more at ease.

While ''find the enemy weakness and exploit it'' is a good idea, it is very difficult to express in a programming language, especially for something as free as an FPS bot. I plan on making an FPS wargame with one side being entirely AI controlled and the other side being actual human players, and I need more specific strategies that aren''t so conceptual.

I''d alike to read some actual info as well so I can understand the higher-level ideas as well, so is there any particular translation of "The Art Of War" you would reccomend?


"I believe; therefore, it is." -True Perception
"The Requested Information Is Unknown Or Classified" -Anonymous

Do a search for ''influence mapping'' and then wait for Ferretman(Steve Woodcock) to do his rounds. I''m sure he''ll be able to point you in theright direction .

Steve has archived a thread on influence mapping at his website (gameai)

If its a FPS bot you want to program see the thread "Hunting mode in realtime games" for some tips/links



Stimulate

quote: Original post by Extrarius
Does anybody know of any good texts (online articles or books) on military intelligence at the strategic, tactical, and operational level? I''ve done searches at amazon.com, google.com, yahoo.com, and ask.com and can''t find anything on actual tactics or strategy. Most of what I find is ''find the enemy weakness and exploit it''. Not specific enough to be helpful for programming a tactical AI at all.

I feel your pain.

Since I''m working on a tactical combat simulator, I too have been looking for information
relative to specific tactics (company level on down) on the web, and from looking for over
a year, I can tell you that stuff is out there, but it is fairly well buried.

As mentioned, Sun Tzu''s The Art of War is a great place to start. Understanding what
is being said takes some patience and thought, even in the best of translations. There are
ton''s of sites for you to choose from. You need to check out a few and find the one that
works best for you.

I found more luck doing specific searches and researching specific areas, than I did by
searching on general terms. Instead of searching using "strategic military intelligence"
or some other higher level concept, look for specific operations like "Black Hawk Down",
"Battle for Hunger Hill" or "Battle of the Bulge" that reflect the tactical level you are
interested in learning about. The tactics deployed are usually discussed within the
documentation about the operation.

You also might want to read up on the "lessons learned" material put out by all the services.
These are typically unclassified accounts of past operations recounted with an eye towards
what was thought to be done right and what was thought to be done wrong.

FWIW, I don''t think you are going to find a neat, consise presentation of what you want,
anywhere. (If you do happen to, then by all means tell us about it).

Good luck,

Eric

Try here. It's really devoted to REAL wargames of a military nature, but you might get some kind of info out of it. This is the only link which may provide some kind of useful programming tips.

Other main links at www.strategypage.com are good too. You might also want to try the defense and national interest page for some good info. This site is excellent as a designer's resource and covers things like psychological warfare, and covers Operational warfare in some detail. Mind you both links are for modern warfare not ancient or historical warfare.

The last link I can think of is a government site.

What's interesting to me though is the concept of 4GW, or 4th generation warfare. Basically it says that most wars will no longer be fought by large standing armies. They are necessary because of the possibility of another large army, but basically evolution and technology have made standup fights like WWII obsolete. I've tried to factor this flavor into my own game, which was why I had my post about extrapolations on sci-fi warfare. It's still a 3GW game at heart, but with some 4GW elements to it. I highly suggest going to all the links above.

One word though, if you are making a FPS game, then you really needn't (is that a word?? ) concern yourself with Strategical or Operational level details. The very nature of a First Person game lies in tactics, not strategy or operation. I might suggest that you read "Book of Five Rings" by Miyamoto Musashi. He was Japanese greatest kenshi...a "sword saint". This book is right up there in military classics with Sun Tzu, but it is on a much more personal level than the ARt of War. So you might glean more information on how to code for a FPS AI bot by reading this.



[edited by - Dauntless on April 13, 2002 12:02:44 AM]

Its not actually an FPS, its a wargame in first person perspective. There will be two teams, AI vs humans, and I want the AI to simulate the entire command structure as it will actually exist for the human team. Different players will be different ranks and each will command those immediately under him until the orders get to the individual troops. It will be a massively multiplayer game with a persistant world, being ''reset'' only when one team entirely annihilates all the bases of the other which would give new players for that team no place to spawn.


"I believe; therefore, it is." -True Perception
"The Requested Information Is Unknown Or Classified" -Anonymous

I can''t help but think you''d be better served by having 2 human teams, if you''re looking for it to be ''massively multiplayer'' anyway...

Personally I think that many of the points in "The Art Of War" could be translated quite effectively into an expert system of some sort, since they tend to come in the "If X then Y" form.


[ MSVC Fixes | STL | SDL | Game AI | Sockets | C++ Faq Lite | Boost | Asking Questions ]

I tend to advocate influence maps. Here''s how I''d use one:

- Enemies add to the influence map
- Allies subtract
- Use A* to find the path with the lowest cost. The cost of a node is given by the influence map
- Follow this path

Now, for the details of adding to and subtracting from the influence map. Find all within a certain radius visible from the enemies location and from points near him. This influence propagates using a flood-fill algorithm in an inverse-square fashion, and is proportional to firepower. You may want to change the falloff function for different weapons: use higher powers for shorter range weapons, and lower powers for things like sniper rifles. Influence will be higher in an enemy''s FOV than outside of it.

Each timestep, each unit adds (or subtracts) from the influence map, and a fixed value is subtracted (or added) to all values in the influence map (to make influence fade over time).

The details of how you determine influence will of course be determined by the details of your game. These are just some suggestions.

This influence mapping and pathfinding system works well for intermediate level strategy. Higher level strategy (what the objectives are to begin with) would be handled by a different system (though you could also use a similar system for this: "strategic influence mapping," as I call it). Lower level tactics would also be handles by something else - probably a simple state machine.

I mentioned how influence maps can apply to even higher level decision making. This is how they do: Important things (enemy leaders, enemies that are causing significant damage to you, a flag or exit, etc) add to a second influence map similarly to what I described before - except such things as FOV are irrelevant. This map can be interpreted as the expected benefits from capturing an area. Of course, the enemy influence represents casualties and therefore subtracts from this. The cell with the highest ''strategic_map - influence_map (if enemies are positive)'' value is the cell you want to go to. This becomes the destination for the previously described pathfinding system.

You can extend this type of thinking to deal with most types of units (both friend and foe) and to most types of objectives. It''s easily-understandable, and tweaking is straightforward. The down side is that influence maps can cost memory. With 256 MB and 512 MB of RAM becoming commonplace, I don''t think that'' an issue. Also, you can always experiment with different resolutions in different axes and different numerical precisions to strike the perfect balance.