BGForums

What things should the AI consider or how should it handle disjointed regions?

Any thoughts here? Picture a map that includes Japan as one region (which has two separate land masses). Then Russia attacks Sapporo (the Northern island I think). How should the AI protect this region in the first place? How should it respond to an attack? There are rules already in place but I'm curious to see if anyone has additional thoughts on the subject.

Threat Analysis and deploy troops accordingly! The AI should conduct some manner/method of threat analysis against all potential enemies and deploy/build/reserve its troops accordingly.

1 - Threat Analysis: Russia conducts an airborne/air assault strike.
Japan's Counter Action: Improve its air defenses along potential air routes onto the island by means of Naval, Ground and airforce units.

2 - Threat Analysis: Russia conducts an amphibious landing.
Japan's Counter Action: Deploy surface and subsurface units to counter this threat. Deploy ground/air based assets with anti-ship capabilities.

*** This also applies to a region looking to go on the offensive.

As dipomacy deteriorates then steps should be taken to react against the possible threats by pulling troops from reserve, increasing DEFCON, etc, etc.

The AI should use Radar Stations, AWAC, emplacements and air defense structures to further improve its defensive positioning.

Diplomatic possibilities include intel-specific treaties that would allow established partnering countries to exchange such information.

Tony

Next question from the Goats will be:



It is not easy to teach the AI to do such things...

I think there it is easier to teach the AI to react to air/amphibious assaults rather than to prevent them...

In a first step, the AI must try to achive numeric superiority in the area to prevent more landings and to close the breach... Of course... That will just make airborne and amphibious divertionary attacks more effective... ... But at least it would be a more active defense.

Matter of fact, the AI should respond to any breach of its borders the same way instead of just throwing a few units and wait for the enemy to advance.... If it sees 1 enemy unit, it should send 3 ...

When the AI does this first step, well, I'll worry about the second then

Draken, I like your ideas. I was asking if the AI wants to do an amphibious assult on another region what path should it take, or how should it determine which way is safe? Should it send a patrol boat along a path and if the patrol boat doesn't die, then it's ok to try the landing? Those are the sorts of things that will help knowing.

Instruct the AI to design its force (ground, amphibious or whatever) based on intel and geared towards the mission it wants to accomplish; then determine what security it would need to support the mission with ground, air, naval support; then accomplish mission.

In this discussion the AI would have to realize the vulnerability of its amphibious fleet and determine what major factors his enemy could through against the mission.

In this example the threats are: Enemy Air with anti-ship capabilities, Enemy Surface ships, Enemy Subsurface Ships, enemy ground Anti-Ship assets, enemy ground indirect fire assets.

Once these threats have been established the AI should then design his amphibious flotilla to counter those threats. Countering the threats with numbers is an approach, however technology would provide better protection and the AI should work towards those means.

The AI should determine initially and then again as the game progresses the possibilities of having to conduct an amphibious, airborne/air assault operation. If it is likely then the AI should set itself a goal and work towards that end.

While the AI is building its forces it should begin/continue extensive intel operations to further improve its chances of success. This intel will begin to represent a target list for the AI to take/destroy/damage during its assualt. The target list will need to be prioritized and worked into the AI's operation.

As the mission is being created the AI needs to have success/failure criteria. If during the AI's amphibious assault it notes that 50% of its air superiority aircraft have been destroyed then the AI may decide to redirect forces from other objectives onto the main objective. If Air superiority is not achieved at all then the whole mission might be scrapped. Success/Failure criteria should be a part of this games engine!

Returning to the intel aspect of any mission, the AI should be made to conduct as much intel as possible with available assets. A single submarine creeping around before the attack begins and in front of the AI's amphibious during its movement to the objective would make an enormous impact on mission success.

Tony

Sorry I though you were asking how to prevent/defend against amphibious assaults...



Amphibious assaults are not easy to plan... Since you are asking about how to reach your assault point I'll assume that you already know where you will storm the beaches, and why and with what ...

Simple put, don't assume any route is safe! You just have to secure the route and landing area and make then safe! And do not forget to escort the convoy there! You must have naval and air superiority or a total tactical and strategic surprise on the landing zone. Gather as much data as you can about enemy forces in your target area so you can calculate your force package accordly (if the enemy have subs, send ASW ships, if it has planes send AirDefense ships, etc). How much units? Well, I guess you can come with a formula of strenght by type of attack/defense to compare both forces

Another tip: Stay away from the coast as much as you can (depending on your convoy supply levels).

Edit:
Tony posted while I was typing....



Interesting.... Making the (military)AI mission based... Each mission with a goal, a time table and forces assigned to it... hhhmmm.....

After some recon with satellites and recon planes Ive came across some big problems with AI:

1. In most cases - without land borders - AI have NO general strategy. They just sit and wait. Ive came across literally hundreds of russian units sitting near their bases in the black sea. - almost every nation (turkey and romania too) in war with russia and there was not even a single patrol out there. Only damaged ships (i wonder how they got damaged) or ships waiting near landmasses - no active units. Thats kinda strange, cause if there IS any damaged unit, shouldnt it be directly routed to reserves for repair? How does AI handle this?

2. I think and im almost sure about it, that AIs main war strategy is just to produce units and fight with em. There is no diplomatic connection with further unit moves, there are no pushes how to make these connections, there are no attempts to make expeditionary forces. I know its very very hard to make AI brain to make something like amphib invasion which is very complicated but what about the most simple type of support - expeditionary air forces for example.

main problems are:

AI is trying to make ALLIES which are too at war with current enemy.
Thats ok - thumbs up.

Unfortunately AI is not tryin to support these allies even if there is no big danger near home borders.

Example : Germany at war with russia, Poland at war with russia, Belarussia decimated and captured - Poland fighting russia with all forces, German units are just chaotically wandering bavaria. Whats the point of this? ?? AI really have no strategy to support allies ?? and im not talking about numbers like how many units should move there and how many of em should wait and defend im talking about absolute disability to help.

Another problem is general BG approach for strategic game concept. I think u guys are trying to make this game fully fuctional even if world will look completely different. Ure trying to make some kind of AI calculations to solve any problem with any land or naval invasion anytime for every region. ... but this is not really the problem for today since AI is not even able to effectively attack on landmasses... not even use air force effectively and not even support their allies even if they DO have path to the enemy (and i tihnk thats the main problem of long time wars lasting forever)



sorry for long post

I hate to bring up another game, but if it is so hard and difficult to program the AI to conduct diplomacy and amphib/airborne operations then why is that the CIV series can do it and many others????

I think it's because the devs what to keep a "realistic" feeling to the game, meaning that you don't have 200 countries with tens of thousands of units all fighting and doing invasions and etc . I guess it would be mayhem on a computers resources. I would like an option where all the countries go beserk on each other, or even gang up on the "human". Pure lunacy, yea!

It would be great to see US and Europe allying together as China takes Southeast Asia with partner Russia eying parts of east Europe and the middle east. Africa would be mayhem with all the countries going after each other, or maybe forming a great alliance. the possibilites are endless, even if they aren't "realistic" And if you put in good working treaties to start with and with sea invasions possible, you might be suprised to find there may be more realism in it then previously thought (North/South America, Europe,Israel,Australia vs. China,Southeast Asia,Russia vs. The Rest of the World).

Or to hell with realism and let each country start with a clean slate (no treaties) and just go nuts on each other. With randomness and luck it may not boil down to US vs. China. Especially if the human is not playing either one of them to begin with.