How AI allocates force for secure task?

[hubee0]

Playing one of my scenarios I noticed, that the AI was using division and brigades HQs as assaulting units.
I discovered, that the enemy indeed run out of fresh units, because a lot of them was securing two targets deep in the area already under its control. The targets have completion and occupation points assigned and priority is set to medium (unintentionally).
Is it because of the priority the AI allocates and keeps so many units there even though the targets are not endangered anymore? I thought the priority doesn't matter when there are completion points assigned.

 

[jimcarravallah]

Playing one of my scenarios I noticed, that the AI was using division and brigades HQs as assaulting units.
I discovered, that the enemy indeed run out of fresh units, because a lot of them was securing two targets deep in the area already under its control. The targets have completion and occupation points assigned and priority is set to medium (unintentionally).
Is it because of the priority the AI allocates and keeps so many units there even though the targets are not endangered anymore? I thought the priority doesn't matter when there are completion points assigned.

Setting occupations points is what's driving your units to garrison objectives.

Occupation points accumulate over the time of the scenario starting from the time the objective becomes active. To obtain full credit for occupation points, a unit must stay at the objective for the duration of the time it is active.

Completion points are counted when a unit occupies the objective at the end of the game.

Setting both occupation and completion for an objective in effect double's its value so long as the objective is occupied from start of when its active until the end of the scenario.

Part of the art of scenario design is using the different criteria for objectives to replicate the tactical / strategic goals that existed at the time of the battle. The way I envision it is occupation being required over the longer term of the battle to achieve tactical goals (holding a river crossing for example) and completion replicating the span of control of the battlefield necessary to achieve the strategic goals.

Page 155 of the game manual describes this the differences in how objectives are valued.

 

[hubee0]

Thanks Jim,
I am aware of that in general. But in my particular case, AI allocates a LOT of fresh units to garrison on an objective with 1000 CP and 500 OP while there are more valuable objectives still to be secured and AI tries to get them by using whatever remained eg. attacking with corps and divisional HQs!
Why? Why not garrison HQs on already secured objective and use fresh units to assault next ones. Why it needs so many to secure? I set priority for the secured objective as medium. Can it be because of that?

 

[Bie]

Thanks Jim,
I am aware of that in general. But in my particular case, AI allocates a LOT of fresh units to garrison on an objective with 1000 CP and 500 OP while there are more valuable objectives still to be secured and AI tries to get them by using whatever remained eg. attacking with corps and divisional HQs!
Why? Why not garrison HQs on already secured objective and use fresh units to assault next ones. Why it needs so many to secure? I set priority for the secured objective as medium. Can it be because of that?

Yes I have noticed this as well. Up to a point sometimes that the side that is supposed to be attacking keeps hanging back and only attacks other objectives with a small force. I'll take a look if I can recreate it at some point and post a screen.

 

[simovitch]

Just set the objective completion timings to expire in conjunction with a reasonable rate of advance. That way the AI won't hang around the entire game in the rear area. In general, use occupation points on objectives that will be contested throughout the scenario.

 

[Bie]

Just set the objective completion timings to expire in conjunction with a reasonable rate of advance. That way the AI won't hang around the entire game in the rear area. In general, use occupation points on objectives that will be contested throughout the scenario.

Good to know I'll try it out.

 

[jimcarravallah]

Thanks Jim,
I am aware of that in general. But in my particular case, AI allocates a LOT of fresh units to garrison on an objective with 1000 CP and 500 OP while there are more valuable objectives still to be secured and AI tries to get them by using whatever remained eg. attacking with corps and divisional HQs!
Why? Why not garrison HQs on already secured objective and use fresh units to assault next ones. Why it needs so many to secure? I set priority for the secured objective as medium. Can it be because of that?

I was addressing the duration for the occupation.

Can't address allocation of forces to the task.

What you expect the AI to do (set HQ on the rear occupied objectives) is how I handle the occupy task(s) when I control the advance (I'll use HQs, Bases if the objective is at a road confluence, or long range artillery -- essentially units that can stay in one place and still accomplish a majority of their missions).

Regulating AI behavior has been and will continue to be an evolving process.

What may help is a nominal logic tree to define criteria for allocating appropriate units to such a task.

My criteria for selecting the units I control for the occupy task is one example of that logic.

Do you have other criteria that you think is pertinent to the task?

 

[GoodGuy]

Can't address allocation of forces to the task.

What you expect the AI to do (set HQ on the rear occupied objectives) is how I handle the occupy task(s) when I control the advance (I'll use HQs, Bases if the objective is at a road confluence, or long range artillery -- essentially units that can stay in one place and still accomplish a majority of their missions).

Regulating AI behavior has been and will continue to be an evolving process.

Correct, but timed objectives seem to work quite well with the enemy AI, and they can create a quite fluid AI troop movement and allocation. In order to keep a rear area group, a permanent objective with a way smaller objective circle should then be placed in the particular area. The AI will then allocate a percentage of the force to keep that area occupied (then just tweak by playing with different priority levels). I am guessing that once the number of troops decreases (due to the casual count), the AI may decide to dump some objectives (like the rear area obj. or lower priority obj.) and commit the rear area forces to the main fight, as well. In case the AI gets pushed away from all the other objectives, it may try to stage attacks to recapture one or another objective immediately, but it will also bolster the defense of the remaining rear guard objective. That would be my guess, at least.

 

[john connor]

No matter how much you try to tweak the weighting of points you may still find that the AI sometimes holds back too much of a 'reserve' for your taste - that is, fails to contest high value objectives when it has the force available to do so (whether it is actually 'thinking' in terms of a 'reserve' or just not sufficiently 'evolved', is a moot point). The clever weighting of objectives by the designer may only 'mask' this sometimes, it seems, by delivering the right AAR result (a draw or limited win, or whatever) whilst the AI behaviour on-map hasn't changed too much. This has been posted about before, so if you're noticing these behaviours you're not alone.

I have found that - in very carefully thought-out and designed scenarios (both official and community) sometimes the AI does exactly what you (as scenario designer, or as game player) would like it to do and really goes for the right objectives with the force it has available giving an excellent 'intelligent' fight. (Sometimes this behaviour is 'historical', sometimes not - that's a separate issue.) But sometimes the AI will hang back and fail to make a fight of things. This too is sometimes historical, sometimes not. I have seen the AI shepherd considerable forces - through to scenario end - on parts of the map that were not even objectives, whilst high-value objectives stood uncontested.

Luckily, I also see the AI putting in really co-ordinated, efficient use of resources to give a sound, realistic and thrilling fight sometimes. My experience tampering extensively with the objectives and the points allocations hasn't made much difference to these behaviours, I'm afraid, though that's largely in scenarios made by other people who had already spent a great deal of time balancing things.