Multi option AI orders - a small suggestion.

[Glubokii Boy]

One major shortcomming with the current  ai plans is that each individual ai group only have one option avaliable for their next 'action' in the ai plan currently used in any specific scenario. Some options exist as to WHEN this 'action' (next ai order) is to be taken via the use of triggers and the gameclock but regardless of WHEN this action is taken it will always be the same. That is...whatever is specified in the next ai order in the list. Ones set free the ai groups will move to the location of that ai order in the mannor specified regardles of how the battlefield is evolving.

Each individual ai order may currently use only one trigger. Resulting in only one condition that can become true to release the ai group from the - wait for - and allow it to carry out its next objective. The only other option would be the gameclock. The gameclock takes no account of what is happening on the battlefield though. It is simply a clock...

These limitations are far from ideal...

The best way to solve this lack of addaptability on the side of the ai would probably be to add multi-option conditional triggers but this is most likely a to big a change to be at all something to considder for CM2.

I have been thinking a bit about a somewhat smaller change. One that would allow for atleast some degree of flexebility on the part of the ai and hopefully be somewhat feasable within CM2.

My suggestion is that BFC might add a function to the current triggersystem that would allow us to set a - SKIP TO ai order number (XX) - in the list of orders for a specific ai group.

This way the designer could place waypoints (ai orders) for several different movement options in the same list of orders for that ai group. 

An example...

The ai is defending. Ai group 1 is a counterattacking force. Ai group 2 is a group of reinforcements. Where will the player be attacking ? Center ? Right flank ? Left flank ? Where should the counterattacking force go ? Where should the reinforcements go ?

Currently the designer has ONE option as where to send these troops. If he guess the players intentions right and send them in the right direction then the scenario will probably work ok. But what if the player does not do 'as he should'...then the ai will not be quite as impressive.

With what i'm suggesting the designer could plot the first 10 to 15 orders or something like that for ai group 1 to be a respons to a left flank attack by the player. Orderlocations 16 to 25 could be a respons to a right flank attack etc, etc...

If a right flank attack is made by the player he will tripp the trigger designed to start the movement of the counterattacking force (ai group 1) in respons to a right flank attack. This trigger includes the new feature of SKIPPING to ai order 16 in this case (the first one in the 'react to a right flank attack' (16 - 25)). Simularely the reinforcement group (ai group 2)  could have different movement paths to handle different attacking/penetrationpoints made by the player and 'skipping' to the right one.

I understand that this will not be as simple as flipping a switch or something to implement and that some additional changes would need to be made...

Atleast these i guess...

In the timing section of the ai-order UI there would need to be space to have multiple 'wait for trigger' options as one extra wait for trigger is needed for each movementpath of that ai order.

A 'remove trigger' function would also needed to be included in the triggers to make sure that multiple movementpaths for the same ai group is not triggered. Ones ONE of the options is triggered the other triggers would need to be deleted.

An increase in the number of ai orders might be needed to allow for multiple movement paths as well as perhaps a higher number of triggers avaliable.

What do you guhs think ? Would something like this be to much work (on the part of BFC) for to small of an improvement or could something like this be useful ?

It would allow for some degree of flexibility atleast...

I hope you guys understand what i'm trying to explain

[A Canadian Cat]

Ah GOTO statements awesome /s

Back to being serious: The ability to add switch or even if then capabilities to the AI planning would be wonderful. Having the ability to do that is desired. Hacking it in with a GOTO type approach is not a good idea. You might think it is easier than doing it right but it is not. On top of which it would make writing and deciphering AI plans much more difficult.

1 hour ago, RepsolCBR said:

With what i'm suggesting the designer could plot the first 10 to 15 orders or something like that for ai group 1 to be a respons to a left flank attack by the player. Orderlocations 16 to 25 could be a respons to a right flank attack etc, etc...

Which also means you need a GOTO statement at order location 15 to skip over 16-25 if your system chooses to go to order 10. Add too much of that and or get that wrong and good luck figuring that plan error out.

1 hour ago, RepsolCBR said:

What do you guhs think ? Would something like this be to much work (on the part of BFC) for to small of an improvement or could something like this be useful ?

I think that adding a multi branch programming system to the AI would be awesome. I wold love to see it designed into the game. I think your suggestion is not a good solution at all - sorry.

[Glubokii Boy]

On 2/21/2019 at 9:18 PM, IanL said:

 

Which also means you need a GOTO statement at order location 15 to skip over 16-25 if your system chooses to go to order 10. Add too much of that and or get that wrong and good luck figuring that plan error out.

 

I see your point...My idea was that you should design the ENTIRE lenth of a movement path kind of 'conected'...orders 1 to 15 would be one path from start to finish of the scenario and orders16 to 25 would be an alternative path etc, etc.

There would be no jumping back and forth in the list...only forth 

But i can see that it might be a bit complicated to figure out excatelly how many orders each movement path would need for the entire scenario. Will path 1 require 15 or 17 orders ?

Where should path 2 begin ? order 16, 17 or 18 ?...this could be a bit tricky maybe. I guess that an other problem could be if the paths end up at the same location at some point in the scenario it would indeed require some skipping back and forth to be able to use the same remaining orders for the different paths...or else the same remaining orders would need to be made for each individual path...Not the best solution maybe 

Oooh well...this was just a small suggestion....Hopefully we will soon se some sort of branching in the AI plans.

 

[Glubokii Boy]

On 2/21/2019 at 9:18 PM, IanL said:

sorry.

No need to appologize 

[Heirloom_Tomato]

1 hour ago, RepsolCBR said:

I see your point...My idea was that you should design the ENTIRE lenth of a movement path kind of 'conected'...orders 1 to 15 would be one path from start to finish of the scenario and orders16 to 25 would be an alternative path etc, etc.

There would be no jumping back and forth in the list...only forth 

But i can see that it might be a bit complicated to figure out excatelly how many orders each movement path would need for the entire scenario. Will path 1 require 15 or 17 orders ?

Where should path 2 begin ? order 16, 17 or 18 ?...this could be a bit tricky maybe. I guess that an other problem could be if the paths end up at the same location at some point in the scenario it would indeed require some skipping back and forth to be able to use the same remaining orders for the different paths...or else the same remaining orders would need to be made for each individual path...Not the best solution maybe 

Oooh well...this was just a small suggestion....Hopefully we will soon se some sort of branching in the AI plans.

 

I would like to see the triggers carry a separate side branch set of orders. For example, you plot out a set of orders for a meeting engagement based on the assumption the enemy will be coming up the left side of the map. All the orders are plotted out for the whole battle just like normal. A terrain trigger is placed on the right side of the map and can only be used once per AI grouping. All orders for this trigger are placed in the trigger pop up menu section. As the scenario designer, it will be up to you to decide when to have the potential for the trigger to be tripped, say between minutes 10-20. The AI group will wait for either for the trigger to go right, or if the trigger isnt hit, go left according to the original plan.

It could also be used when designing an attack plan. A and B company are given to you for this battle with the main objective being a large village, and a secondary objective of a small forest near a crossroads just past the village. A company has the potential to sieze the village alone, leaving B company to take the woods and crossroads. Yet, if A company runs into trouble, the village is far more important and the crossroads objective should be abandoned in favour of taking the village. 

B company is given orders to move to overwatch positions and join the attack on the village from the right flank 35 minutes into the battle, as well as a Terrain Trigger with plans to bypass the village and seize the crossroads. If A company reaches the terrain trigger on the far side of the village before 35 minutes, B company takes the crossroads. If A company doesn't trip the trigger before 35 minutes,  B company joins the attack on the village.

Editted to add: I would envision once a trigger is tripped, all other orders not in the trigger section would be forgotten and only the trigger plan would be followed. If the trigger is not tripped and the timer passes the preset time, the trigger orders are forgetten and only the orginal AI plan is followed. Since each AI group could be given only one branch, it should help keep things relatively simple. 

Clear as mud?

Edited February 23, 2019 by Heirloom_Tomato

[A Canadian Cat]

5 hours ago, RepsolCBR said:

I see your point...My idea was that you should design the ENTIRE lenth of a movement path kind of 'conected'...orders 1 to 15 would be one path from start to finish of the scenario and orders16 to 25 would be an alternative path etc, etc.

There would be no jumping bac

Right making it block based is the better design. Case 1 do this block, case 2 do this block, case three do this block, optional default after x minutes do this block.

To do that you would really want to change the UI for the AI authoring quite a bit. The small column space for the order flow doesn't really lend itself to branching and cases. Don't forget any of those blocks could have other cases with more branching. Not sure where the limits would be.

I think that this is the way to go at some point. Hopefully it will happen.