Jump to content

AI never in attack.


Recommended Posts

 

I have uploaded 2 versions of one scenario -AI/H2H- and I realised there is no point in planning the AI to attack; it is stupid enough in defence.

 

I was reading and old post in CMFI... and yes, it is really stupid.

There was one scenario... maybe if it had 36 tanks... because if it has 40, the AI always win and if it has 30 the AI is always defeated. What's the point of this scenario? a shooting test?

 

Somebody was saying that with the "perfect timing", a "careful planning".. let's see.

The first platoon will take a position on the left flank to provide covering fire, 2nd pl will advance to "X", once in its vicinity -trigger- 3rd pl will move through 2nd pl to help clear the objective. Meanwhile the Human Player was moving against the AI left flank, attacking 1st pl and is now in position to fire on the flank of the 2nd pl while it moves forward, because 2nd pl WILL MOVE FORWARD regardless of the situation.

What has to do "timing" with the situation? Planning includes telling the HP not to attack the left flank?

 

Whatever the AI has deployed to defend the right flank will sit there, doing nothing, while the HP attacks the left flank; maybe some distant fire. If the AI launches a counterattack, it is going to move troops "somewhere", if, maybe, by chance, that "somewhere" has anything to do with what is actually going on... lucky you. The AI is following a script, is never going to react to whatever is happening. A trigger may be activated by a patrol or by a lonely AG supporting the infantry, while the armoured column is advancing 200m to its left or nowhere.

The AI will always do one thing and only that one thing, I think that's why the "counterattacks" I have seen in some scenarios are only there to give the HP something to shoot at.

 

On the other hand...

I haver never seen this behaviour or this code but I think you could provide the AI with, at least, some "reinforce position under attack" option.

 

A coy is on the left, around "X" objective, B coy on the right, around "Y"; C coy is in [support orders].

X = 100 points, Y = 50 points, -so, in case of a draw, X is more important- you add casualties, under artillery fire, enemy armour around, number of enemy units identified... After 30' the AI checks the situation, if the threshold of time, casualties... is triggered C coy will reinforced A coy. Meaning: C coy will move to the deployment area of A coy and adopt its orders, C coy could even leave one platoon behind, still under "support", that later on is going to join B coy.

 

Obviously, it is still not working, just a bit less stupid.

A HP could see that casualties in A coy are from long range fire, nobody is advancing: A coy take cover!, let's send reinforcements to B coy.

After 20' check casualties every 10': in the first minute A coy triggers support, in the next 9' B coy gets one pl wipe out. The AI is NEVER going to react, is not going to stop C coy, wait and decide according to circumstances.

Ii is just a limited, one time only, possibility to act based on the situation.

 

Is it possible to write this code?

 

As I read in another post...

In PC games there are two levels: easy is playing against the AI.

 

Link to post
Share on other sites
  • Replies 54
  • Created
  • Last Reply

Top Posters In This Topic

Top Posters In This Topic

Popular Posts

Given the current state of AI in games, if you are seeking to replicate how a human may play using the AI and specifically the AI in CM then you are on a road to frustration. There has been some

PERSISTANT MAP-DAMAGE might be a good cure for this decease...😁 If we could use large maps and play several shorter battles on the same map might make it somewhat easier on the AI...

I prefer to have the human player make his own decisions.  Decisions that are interesting because they have both benefits and consequences.  Also, in the editor, it's not possible or desirable to tell

"Old post"  Which scenario is important. The options available to someone building AI plans in the game now are much better, and a canny designer (looking at you @George MC, just to take an example), can provide an AI plan or 2 or 3, that will give you fits. Older scenarios did not have the ability to do so. With SF2 for example, ALL of the scenarios were gone over, and in quite a few cases, completely new AI scripting done.

Newer titles will have much more capable AI. R2V and upcoming RT provide more capable AI opponents than older scenarios. Lastly, it could be the designer didn't really intend to have it played in that direction and just added a basic AI capability to it.

Dave

Link to post
Share on other sites

Sorry.

The reference was to a topic in CMFI, I am talking of my scenarios in CMRT. I always use the 5 Plans, the "somewhere", "doing nothing" and "WILL MOVE FORWARD" still applies.

In every single Plan.

What capability can you add? To reinforce the flank actually under attack?

 

 

Because of all the men facing us, no one is call Gisco. Hannibal, at Cannae.

Link to post
Share on other sites
12 hours ago, semmes said:

What capability can you add? To reinforce the flank actually under attack?

This is an interesting topic.  I enjoy building scenarios and experimenting with the AI in scenarios.  The AI can't come close to reacting as a human but with the tools provided in the Editor it can provide a fun, interesting experience.  

There are a combination of friendly triggers, enemy triggers, friendly armor trigger, enemy armor trigger, orders trigger and timers.  These, along with scoring/objectives, TOE, terrain etc. can be used in a combination of different ways to make the AI put up a good, interesting fight. 

Reinforce an AI unit.  IMO the AI is best at defense.  In the CMBS scenario Tactical Operations Center I placed a friendly trigger behind an AI defensive position.  An AI reinforcement group was assigned to this trigger.  A few of the teams in the AI defensive unit were given poor morale.  If/when the player attacked the AI defensive position in strength at least one of the poor morale AI teams fled to the rear, hitting the trigger.  This triggered caused the reinforcing AI unit to respond.  @Sgt.Squarehead helped test this scenario and can speak for the effectiveness of this AI tactic.  If the reinforcing unit is not triggered it can still be useful sitting in keyhole positions covering an avenue of approach etc. 

There are several variations of this AI tactic that can be employed.  One is to place an enemy armor trigger in the AI defensive positions.  In this case the AI reinforcement groups will wait to respond until the player is starting to overrun the position with armor.   Then a platoon of IS2s appear and counter attack etc.

An orders trigger can also be used to reinforce the flank.  The AI defensive unit starts withdrawing when enemy armor (so enemy armor trigger) is closing with the defensive position.  Four orders into the withdrawal sequence is AI Group 11 order 5.  AI Group 11 order 5 is assigned as an order trigger for the IS-2 platoon to approach the original defensive position from the flank.       

If we are building a scenario for single player only (my preferred way) we don't have to worry about Head to Head TOE balance.  IMO this makes things significantly easier.  A company of T-34 tanks can sit on the AI side of the map, out of LOS, waiting on a trigger.  If the trigger is never tripped this company of tanks will keep the AI from a premature surrender but may never fire a shot.  The player will only become aware  of them during the AAR map review.  IMO this is okay in certain circumstances.  

In contrast that same AI group can be reused.  [A3] is a platoon of IS2 tanks on the east side of the map.  On a timer they attack across the map (through the player's positions) then hit an AI exit on the west map edge.  So most all of the [A3] tanks are destroyed or exited (an occasional one is still on the map immobilized).  Then 15 minutes later a platoon of AI controlled T-34 tanks spawn (show up as reinforcements) on the east map edge with the same AI group number [A3].  They again attack west or whatever the designer assigns them to do.  There are seven reinforcement slots, in addition to what starts on the map, to reinforce 16 AI groups.  I think I have used this in all my scenarios except one.           

IMO it is easy to make the AI very difficult to defeat.  However, I don't care for difficult battles and prefer instead to have interesting battles with interesting command decisions that have both advantages and consequences. 

As an example:

I learned this from @George MC.  Do I call on the extra platoon of Tigers from battalion reserve?  These Tigers are also a 200 Victory Point (VP) spot objective for the Soviet's.  Calling them up to the front should provide useful fire power but will also give the Soviets 200 VPs.  Also the ground condition is muddy.  How many might bog/immobilize during the fight?   Is it worth it?  Decisions......

Do I want to hit cease fire at one hour (as the scenario recommends) or do I want to fight on for an extra 30 minutes?  If I fight on for an extra 30 minutes the OpFor will earn 150 VPs (shortly after one hour an OpFor reinforcement will spawn on a 150VP touch objective).  But with the extra 30 minutes I could probably take the crossroads which is a 300 VP occupy objective.  Can I hold the crossroads against the expected OpFor trigger to counter attack until the extra 30 minutes is up?  Should I settle for a minor victory or go for the major victory?  Decisions...  

Lots of fun interesting stuff can be done with the scenario Editor and the AI. 

Below is a link to a new mini-campaign / scenario with the AI attacking.  I'm concerned the AI is to strong for the average player.  Waiting on partisans to finish it. 

 

      

Link to post
Share on other sites

Here is another way the AI can be very effective.  The scenario Coup de'tat features the “OpFor Room” located in buildings throughout the map.  The OpFor room is a ground floor room with no exterior windows or doors where AI teams might be located.  

The player will get tentative contact icons for some of the AI teams in these OpFor rooms.  Some of the AI teams will cause no problems and remain in their OpFor room the entire scenario.  Some of the AI teams will only come out if triggered and some are coming out (on a timer) to cause problems no matter what the player does.  The player has no way of knowing which one of the three situations a tentative contact represents and some of the OpFor rooms are empty.   

It is impractical for a player to try and clear all of the OpFor rooms especially since many will never cause a problem (in game intelligence may report something in a specific building).  The player is forced to conduct operations while surrounded by this AI population.  Just as in RL he is never sure which one of these rooms OpFor teams might emerge from or when. 

A lot of cool stuff can be done with this setup especially in urban terrain.  I originally used this technique in the scenario Tactical Operations Center.  However the urban terrain of Coup d’etat allows for much more widespread use.  Below are some screenshots.   

t56iOv9.jpg 

As an example:

1 = AI Group #1. This group will remain stationary (unless routed out by a nearby VBIED blast or airstrike etc.)  They will shoot at friendly units that enter the building but will not initiate activity. 

2 = AI Group #2.  This group will emerge from their OpFor rooms to attack a target if the player hits trigger X. 

3 = AI Group #3.  This group will emerge from their OpFor rooms and attack a target on a timer. 

4 = AI Group #4.  This group will emerge to attack a target if the player hits trigger Y or on a timer whichever comes first. 

Some AI groups will conduct operations, move to an exit zone and leave the map.  On a reinforcement turn the AI group will reappear with a fresh unit and conduct additional operations.  

 

5oAzTFPh.jpg 

Above, various AI groups are shown in "OpFor Rooms" waiting on a trigger, timer or combination.  Some will never initiate movement from their location.   

 

8Jt9sHuh.jpg

Above, exterior view of an "OpFor Room".

 

TNdNk4Ch.jpg

Above, a police unit and their Canadian advisers, conduct operations while surrounded by an AI controlled population concealed in OpFor hide rooms.    

 

6c5ztJ9h.jpg

USMC landing force exits the USS Wasp Assault Ship in order to conduct operations in the above mentioned urban area with OpFor rooms. 

Link to post
Share on other sites

I can't talk about CMRT basegame because none of my scenarios are in it, but I've got four scenarios in the coming Fire and Rubble module they lean heavily on the AI staging large complex assaults.

I've joked on the Beta board that the hardest part of scenario design is doing the stuff nobody is likely to see, because players tend to choose the active side against relatively inactive defenders. If they played the defender instead they'd see the AI mounting a formidable offensive.

Edited by MikeyD
Link to post
Share on other sites
1 hour ago, MikeyD said:

I can't talk about CMRT basegame because none of my scenarios are in it, but I've got four scenarios in the coming Fire and Rubble module they lean heavily on the AI staging large complex assaults.

I've joked on the Beta board that the hardest part of scenario design is doing the stuff nobody is likely to see, because players tend to choose the active side against relatively inactive defenders. If they played the defender instead they'd see the AI mounting a formidable offensive.

Excellent.  Looking forward to that.

Nice to defend against the AI once in a while.

Link to post
Share on other sites

Very good posts in this thread. I love those "OpFor rooms".  I'm trying to design my first scenario right now so this is interesting. I was having trouble figuring out how to get the AI to work.

I made a very small scenario for CMSF2 about an insurgent raid on an AI controlled checkpoint. All you get at the start is a small handful of guys in position to start shooting at the checkpoint to distract the enemy while another group of guys in a pickup truck and a taxi roll up from behind to try to kill as many as possible before driving away again.

I wanted to make it so you had to be fast when raiding the checkpoint, because AI reinforcements would come running to check things out once the shooting started. I put one AI squad in the nearby town, as if they were on patrol, then I put in an AI trigger for them to start running toward the checkpoint once the player reached it. They would never come though and it took me a long time to figure out why. They would just sit there at their start point and never move. At first I was just doing the AI orders and triggers wrong, but once I figured that out, they still wouldn't come. Turns out that the AI squad was simply getting freaked out about their buddies and their HQ getting blown away back at the checkpoint. They would get to a "rattled" state and then cancel the orders before they even began - as if they were saying "Screw that! We're not going over there!" It's kinda neat that the AI does that. I had to bump up the squad's quality and motivation just to get them to move.

I think I got the scenario working pretty well. I added in more AI reinforcements such as a mechanized platoon that drives to the checkpoint, dismounts, and then starts searching the area on foot. I'm not very good with the AI orders so it looks kinda clunky sometimes but it works. It's been fun designing it though. I'll probably end up making a lot more.

Link to post
Share on other sites
11 hours ago, Bozowans said:

I'm not very good with the AI orders so it looks kinda clunky sometimes but it works.

Cool. Thanks for sharing your experience and we all look forward to seeing something to play with.

Also, smart move starting small. It is so tempting to go big but getting started small really helps.

Link to post
Share on other sites

These tricks are nice and all and I agree that the quality of the scenarios have improved over the years. I like all the latest additions to the editor. Things like area-fire,  withdraw and face have all made the AI better.

But nothing is so good that it can't be improved...not even a Repsol-Honda apoarently 😉...

 

Any MAJOR remake of the editor within the CM2 timeframe is not likely and CM3 seems to be quite some time away...

The various tweeks/tricks the community have come up with works good enough in many situatoons but it is a bit of a shame that these tricks are neccesary. It makes scenariodesign more complicated and time consuming then it need to be imo...

Any further improvements to the editor would be most welcome...😊

One of the major shortcommings of the AI programing currently is the lack of any branching options...the 'one way forward or non at all' limitation is quite...

Limiting !

One fairly simple solution to allow the AI to have some more flexibility could be to add a SKIP-function to the editor...

Dependant upon the situation the AI group could be made to SKIP a number of AI-orders and move directelly to a selected one further down the cue...

This way the designer could program several paths forward and the AI could be made to pick the most adventagus one via the use of triggers.

Currently the timing options for the AI groups looks something like this...

- exit between option

- wait for trigger (or AI order)

What if we could expand this somewhat to look more like this...

- wait for trigger  XX

- wait for trigger  XX

- wait for AI order  XX

- wait for AI order XX

- exit between  XX

Now we have two trigger options AND two AI order options as well as the exit between option to choose from. That is FIVE different options that could decide when the AI group would move out.

But still only ONE way forward ! The next waypoint in the cue...

The game would 'check' this list of options from top to bottom...the first option that becomes true will be choosen...

We have more timing options...but what if we also add this...

- wait for trigger XX  skip to XX

- wait for trigger XX skip to XX

- wait for AI order XX skip to XX

- wait for AI order XX skip to XX

- exit between XX skip to XX

Now...if for example the first trigger becomes true the AI group would SKIP to the designated waypoint (AI order) and continue from there. If instead the first AI order becomes true then the AI group would SKIP to that waypoint and follow that path forward. This would allow THE SAME AI group to have several courses of action it could choose from at various points in the battle instead of only ONE or non at all.

 

If we for example have an AI group that has been designated to counterattack a player advance...we could now setup trigger one on the right and trigger two ln the left. If the player attacks into trigger one then the AI group would SKIP to the first waypoint in a path that is designed to counter a player attack on the right.

If the player instead decides to attack on the left...then trigger two will become true and the AI group will SKIP to a path designed to handle a left flank attack...

Or.. 

We have two AI groups attacking. Each consisting of a tank platoon. They need to cross a river...if the first platoon makes it across then the second platoon could follow. If the first platoon fails to cross at that location though then the second platoon could be made to move to another crossing-location instead of following platoon one and dying in a simular way...

To take this a step further 🙃...

What if we could also add this....DURATION

- wait for trigger XX skip to XX duration XX

-

-

-

-

Duration would specify how long the AI group would try to carry out the current AI order. If left unspecified no duration would be set.

This duration would not affekt the AI performance in any way...like forcing the group to run full speed for example...It would only be a timer for how long the current AI order will run...

Ones the duration runs out the AI group would move on to their next order.

A duration feature could be useful in shoot and scoot moves...area fire etc..

Or...

An AI group is set to assult an enemy possition in a village. The duration for the assult order is set to 5 minutes...a terrain objective is placed infront of the enemy possition...if the AI group makes it into that terrain objective within the assigned 5 minutes then the assult is considdered to be making good progress...the order after that assultorder checks of any friendly troops have made it into that terrainobjective...if they have then the next order will continue the assult...if they have not then the AI group will SKIP to a new waypoint and either pull back and wait for reinforcements...or try a new assult direction...

It may not be perfect but this SKIP to waypoint funktion seems to me to be a fairly 'simple' thing to add....and might be doable within the CM2 editor...

A completally new design with conditional/branching triggers and conditions would obviously be better...but how long will that wait be ? 

Would a skipfunction or something simular be a worthwile effort within CM2 ?

 

 

Link to post
Share on other sites
On 9/28/2020 at 5:42 PM, MOS:96B2P said:

IMO the AI is best at defense. 

Great, and great to see more people interested in the actual coding of the AI misbehave.

Now...

I did mention the trigger, the problem is that it can be activated by a scout or a T-70, instead of by an IS-2 platoon. Also, for some reason, your "trigger behind an AI defensive position" sounds more like a trap -but not an ambush- than a reaction and it can be triggered by an artillery round hitting the "right" squad; just a too elaborated script for my liking.

Yes, it is an improvement and you can do a few things with the tools at hand but I wonder if we are treating the symptom and not the disease.

"with the tools provided in the Editor it can provide a fun, interesting experience". I'm afraid I cannot agree, fun yes but not joy.

Interesting? ... If you move to the flank of the attacking AI it will still move forward, it will not wait and it will not deploy to provide covering fire. 

I cannot call that interesting.

 

Give me the guillotine and I will give you victory. Joffre.

Link to post
Share on other sites
On 9/29/2020 at 2:01 AM, MikeyD said:

I've joked on the Beta board that the hardest part of scenario design is doing the stuff nobody is likely to see, because players tend to choose the active side against relatively inactive defenders. If they played the defender instead they'd see the AI mounting a formidable offensive.

The reason people play the attackers against stationary defenders is because there is no AI that could mount a "formidable offensive".

 

Link to post
Share on other sites

Given the current state of AI in games, if you are seeking to replicate how a human may play using the AI and specifically the AI in CM then you are on a road to frustration.

There has been some useful, practical and workable advice regarding how to get the 'best' out of the AI given the current limitations of the AI.

So there are options.

1/ If you think the AI is rubbish and lacks then play against humans (just don't play against me because my playing in CM makes the AI looking positively ninja General like in comparison...). 

2/ Work with the AI tools we have and seek to make the best we can - @MOS:96B2Phas some excellent suggestions how this might work.

3/ Pound sand and bewail how crap the AI is.

Choices, choices, choices...

Link to post
Share on other sites

The good news is that it is possible for a good designer to create AI plans that emulate playing vs the intelligence of a good human player. 

In CMFB's "Mission to Maas v2" (which I highly recommend) I was amazed at how the AI plan (US) not only responded in a clever way but also initiated well-constructed attacks vs the human (German).  So, a very good AI attack can be created with current tools.  Since we don't experience this very often, presumably the effort takes work and a lot of playtesting.

Link to post
Share on other sites

I can understand that the guys that is associated with BFC in one way or the other at times gets tired with the pretty frequent complainig but...

When reading many of the replies in the various threads asking for improvments one could get the feeling that these guys have the opinion that NO improvements are needed...and NO further work SHOULD be done with the editor...

I hope...and belive that this is not true...

But it kind of sounds like it at times...

Come on ! 

It's not like we are asking for a fully functional cure for cancer or something...many of the suggestions in the various threads would likely not require more then some UI changes to the editor...

It ought to be doable without BFC having to put everything else on hold for a long, long time.

Some request are obviously more demanding though...and perhaps not realistic...or even desirable...

Fine !

 But this everything is fine attitude..."if you dont like it then stop playing" is somewhat strange...

I mean...the severe shortage of community scenarios ought to be an indication that everything is indeed not...all fine

Yes...some of the more experienced and skillfull designers can design very good scenarios...but at what price ?

Hundreds and hundresds of hours...maybe even 4 didgit numbers...that is fine if you have the time and intrest in spending most of your free hours for many months doing this...some will...and that is obviously great...but many more will not. 

I for one still hope that many improvements will still be made to the editor...both big and small. Hopefully these will make the editor accesable to more people...increasing the amount of community scenarios being made...

As well as also improving the quality of the scenarios...EVEN FURTHER 😊

 

Edited by RepsolCBR
Link to post
Share on other sites
2 hours ago, RepsolCBR said:

Yes...some of the more experienced and skillfull designers can design very good scenarios...but at what price ?

Hundreds and hundresds of hours...maybe even 4 didgit numbers...that is fine if you have the time and intrest in spending most of your free hours for many months doing this...some will...and that is obviously great...but many more will not. 

You are right imo.  But, for major improvements it's likely we'll have to wait for a CM3 - assuming the CM system survives that long given the increasingly slow rate of development. 

Having been a "Prime" govt software developer am familiar with the allure of easier and larger money rewards when doing govt contracts tempting even the most ardent wargame design house.  

Link to post
Share on other sites
17 hours ago, semmes said:

I did mention the trigger, the problem is that it can be activated by a scout or a T-70, instead of by an IS-2 platoon. 

Maybe I'm not understanding but there are different trigger types.  A enemy armor trigger cannot be activated by an enemy infantry scout team.  Even an enemy jeep, truck (any soft skin vehicle) will not activate that trigger.  It will only activate when enemy light armor or armor touches the trigger.  It is fair & accurate to say an enemy armor only trigger will activate for both a single enemy T-70 or a platoon of IS-2s. 

I think it would be useful if there were more trigger types.  One for light armor only , one for armor only, one for soft skin only , one for infantry only.  There are currently design ways around this but it is always nice to have more and improved tools.  I'd definitely find a way to use an expanded trigger system.  More terrain objective / trigger slots is high on my wish list. 

Working with the tools we have:  The example of a single T-70 tripping an enemy armor only trigger when you want to wait for the platoon of IS-2s can generally be handled by the placement of the trigger.  Place the trigger very close to the AI defensive position or even in or behind the position.  A few player controlled T-70s will probably not survive to reach the trigger.  The panzerfausts, shrecks and AT guns will stop them.  However the player controlled IS-2 platoon may survive until it hits the trigger.  Then the AI controlled Tiger platoon responds etc.  Of course if the player has bad luck / skill and the IS-2s never make it far enough to hit the trigger then the Tigers were not needed and do not respond.  The Tiger platoon can however still respond on a timer even if the trigger was never tripped.   Testing on scenario author test mode will allow adjustments until the desired result is obtained most of the time.          

 

17 hours ago, semmes said:

I wonder if we are treating the symptom and not the disease.  

I guess we are treating the symptom.  As hobby scenario designers / players that is probably the only practical choice we have.  I'm getting a lot of fun and enjoyment out of "treating the symptoms" in both the games and editor.  In fact working in the editor is kind of addicting. 

This forum is generally a pretty friendly, helpful group.  If you need help, feedback etc. in the editor or in gameplay many would be happy to help.   In fact you may get more feedback, advise and opinions than you actually need / want. :D :lol:     

Also, BFC has talked on the forums about a new game engine.  So, not only are they continuing to upgrade, improve and support the current engine but are also planning for the future.  That may be something to look forward to.          

 

8 hours ago, RepsolCBR said:

I for one still hope that many improvements will still be made to the editor...both big and small  

One of my top requests is more terrain objectives since they are used for both terrain objectives AND triggers.  More AI groups would also be useful.  Also if the AI infantry could use Move so they could be made to walk down a trail or street etc.  As it is now they run almost everywhere (unless suppressed or exhausted).   Also more reinforcement groups and later reinforcement times.  So reinforcements can show up later in the scenario.  :) 

 

5 hours ago, Erwin said:

Having been a "Prime" govt software developer am familiar with the allure of easier and larger money rewards when doing govt contracts tempting even the most ardent wargame design house.  

Now, that's just a scary thought.......

Unless of course they reinvest the profits into Combat Mission for more coders, games, modules, newer engine etc.  

 

Screenshot of an AI controlled flame tank in the attack against a human player. 

F3kHowrh.jpg

Edited by MOS:96B2P
Link to post
Share on other sites

The one comment I find interesting is, why do we not have many user made scenarios anymore, the answer is easy.

As they made the AI programming more difficult to allow the designer more ability to make better AI plans, Less players have spent the time to learn to use the tools.

Now adding more tools and increasing the complexity is good for those who want to design better AI plans, but keep in mind, as you create more complexity. Those who master it will be fewer and fewer.

So what you get from BF design team will be better in quality, but on the other hand, you will be seeing less output from the casual player who is just into it for their personnel enjoyment.

 

Keep suggesting concepts for better AI - its always good for that type of imput.

As for getting more content from the fanbase, I am sorry to say, the answer will never be by adding more features, unless you can also suggest on features that take away the need for the person to program the plan and that somehow the machine just automatically can do it.

 

 

Link to post
Share on other sites

If there was one thing I would love to see AI programming do, this is it.

When AI units attack, they will follow the path that seems correct to their programming to get to their destination. The problem is, they do not take into account if previous units have suffered losses taking that path.

Thus a human player can kill enemy unit after unit going through some obvious movement location choke point.

If some programming could be written to force the AI to stop using said path once losses has been taken to a certain extent and then recalculate and try other routes, it might be a easy way to make AI on the offensive much more competent.

Edited by slysniper
Link to post
Share on other sites
18 hours ago, semmes said:

If you move to the flank of the attacking AI it will still move forward, it will not wait and it will not deploy to provide covering fire. 

I cannot call that interesting.

Sorry, almost forgot about this one.  Yes, the AI controlled unit will continue to follow its order and move to the next orders location.  However, if that is not the desired result other design elements can be added.  IMO this is part of the fun of designing and testing a scenario.  

The original AI group (we'll call it [A7]) that is moving forward can be given different types of orders and stances.  If it is given Dash it will prioritize moving fast over shooting.  If it is given Assault or Max Assault it will move Quick with alternating team/vehicle bounds, one pass the other.  The AI prioritizes shooting over moving.   There are Stance Orders (Fire discipline) such as Active (shoot often), Normal, Cautious, Hide & 10 different ambush stances.  AI controlled vehicles can be ordered (in 3D mode) to open up so they have better situational awareness.  Soft factors can be tweaked etc. 

So if [A7] is given Assault & Active, open hatches, it will bound and shoot during the flank attack.  They will basically try to fight their way through the ambush (instead of staying in the kill zone).  They won't ignore the ambush.  But yes any survivors will ultimately move on to the next assigned orders location.  Also many players would give their player controlled moving unit an over-watch unit.  Maybe AI controlled [A7] should also have an over-watch unit?  That is a designer decision. 

Below is a screenshot of an AI controlled unit after breaking through a player controlled position.  In the background, where the columns of smoke are rising, is the breakthrough location.  The AI unit was on Assault, Active, Open up during the breakthrough.  In this screenshot it is on Advance so the AI stops bounding and moves to exploit. 

PBoTkHgh.jpg    

 

The AI is not as adaptable as a human player.  However it can do surprisingly well.  :)                   

Edited by MOS:96B2P
Link to post
Share on other sites
23 minutes ago, slysniper said:

If some programming could be written to force the AI to stop using said path once losses has been taken to a certain extent and then recalculate and try other routes, its might be a easy way to make AI on the offensive much more competent.

+1.  This would be useful. 

Link to post
Share on other sites
40 minutes ago, slysniper said:

The one comment I find interesting is, why do we not have many user made scenarios anymore, the answer is easy.

As they made the AI programming more difficult to allow the designer more ability to make better AI plans, Less players have spent the time to learn to use the tools

I guess this could be true...to a degree 😊

But that is part of the reason why i would like to see the editor expanded/improved...

Simply because we get more/new features this does not mean that it will HAVE TO  get more complicated...i'm rather hoping for the opposit 🙂...

That we get new and better features that will make the scenariocreation faster and easier...as well as better.

I don't think that it currently is the amount of features that is holding the average scenariodesigner back...but rather the userfriendlyness of those features...

 

 

Link to post
Share on other sites

Some great tricks in this thread.

Part of the issue concerning the paucity of user made scenarios in my opinion is that aspiring scenario designers too often feel that they absolutely have to include stuff like AI plans or fancy briefing graphics in their scenarios.

The truth is you don't really need to bother with any of that stuff.

Overhead map with graphics for objectives on the tactical screen part of the briefing? Nope, not needed. You get to see objectives just fine in game with the ALT J shortcut or whatever it is. Graphics for reinforcements, artillery support or a breakdown of the point values of the various objectives using the official Battlefront Photoshop template? Nope; not needed. Just write it in plain text in the briefing. AI plans ? Nope, not really.

If you make it so your scenario has the human player attacking, then not every battle needs an AI counter-attack or a withdrawal.

The point of counter-attacks is that it encourages the player to keep honest, use real life tactics by covering his flanks for instance, if he doesn't want to be caught with his pants down. But what triggers this behaviour is the possibility of a counter-attack. And this possibility doesn't come from having an AI plan with an actual counter-attack in the battle at hand. It derives from the fact that among all the scenarios you are going to play, some of them will have one and you got burned once. But it doesn't have to be this one particular scenario. As long as some scenario designers include counter-attacks, then players will take this into account and play accordingly. Because really the threat of a counter-attack is the determining factor here. Often when it actually materializes, if you have taken the proper precautions, it can be fun but it is rarely a huge problem.

So it is perfectly fine and acceptable to have the AI sit still in defense.

The only two things you really need for a scenario, for it to be decent and enjoyable is 1/ a good looking realistic map and 2/ good defensive positions for the AI.

Now will all these extra elements make your scenario better if you go out of your way to include them? Yeah sure they will.  But this is by no means a must.

Link to post
Share on other sites
35 minutes ago, Zveroboy1 said:

Some great tricks in this thread.

Part of the issue concerning the paucity of user made scenarios in my opinion is that aspiring scenario designers too often feel that they absolutely have to include stuff like AI plans or fancy briefing graphics in their scenarios.

The truth is you don't really need to bother with any of that stuff.

Overhead map with graphics for objectives on the tactical screen part of the briefing? Nope, not needed. You get to see objectives just fine in game with the ALT J shortcut or whatever it is. Graphics for reinforcements, artillery support or a breakdown of the point values of the various objectives using the official Battlefront Photoshop template? Nope; not needed. Just write it in plain text in the briefing. AI plans ? Nope, not really.

If you make it so your scenario has the human player attacking, then not every battle needs an AI counter-attack or a withdrawal.

The point of counter-attacks is that it encourages the player to keep honest, use real life tactics by covering his flanks for instance, if he doesn't want to be caught with his pants down. But what triggers this behaviour is the possibility of a counter-attack. And this possibility doesn't come from having an AI plan with an actual counter-attack in the battle at hand. It derives from the fact that among all the scenarios you are going to play, some of them will have one and you got burned once. But it doesn't have to be this one particular scenario. As long as some scenario designers include counter-attacks, then players will take this into account and play accordingly. Because really the threat of a counter-attack is the determining factor here. Often when it actually materializes, if you have taken the proper precautions, it can be fun but it is rarely a huge problem.

So it is perfectly fine and acceptable to have the AI sit still in defense.

The only two things you really need for a scenario, for it to be decent and enjoyable is 1/ a good looking realistic map and 2/ good defensive positions for the AI.

Now will all these extra elements make your scenario better if you go out of your way to include them? Yeah sure they will.  But this is by no means a must.

I  agree with this post fully !

But the sad fact is that this suggestion has been mentioned...oooohh, i don't know how many times...but it comes to pretty much nothing...

We don't see many such scenarios uploaded...no matter how 'true' this suggestion is...

Many have suggested it...including myself.

Many years ago i tried to get a new type of scenarios going...take a QB map, place some static defenders on it and call it finished.

Such scenario could be good enough for others to play...in a quick sitting...

I made a few of these but there where very few takers...something like 2 guys made one or two...no more.

 

 

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...