Jump to content

Co-op vs. AI


Recommended Posts

Hey guys,

I'm thinking about creating a mini scenario for a two player coop pve (read: silver tower type enemy behavior) game. I'd like to use it as a small introduction type method for someone to get excited about aos. 


Dont want to play against them but rather make a small warband with 2-5 chars each and maybe go attempt to kill a giant or survive 8 waves of skaven in a tower. 

Id love to use actual heroes with many wounds and not regiments of renown type limitations. 

You, seasoned game masters and storytellers, would you give me tips on how to best balance such scenario? (I'm familiar with storm of sigmar contents)

Id love to play a few games on the side of my young friend so that he gets used to the smaller feel of aos and then increase unit sizes and go against each other when it's less intimidating. At this point he feels like it's unfair that I know more about the game and enjoys it less when he loses. Idc about that so would love to team up against a randomly behaving diced foes. 

Was thinking about us taking turns controlling the enemy forces (wqst runemarked) and maybe 1-2 = charge closest, 3-4=charge most wounded, 5=cover, 6=animosity. But it's still very basic and does not resemble things accurately. 


Link to comment
Share on other sites

If you're looking to fight against waves of stuff, try to keep the 'AI' warscrolls nice and simple (mostly rank and file stuff, no complicated special rules, maybe a single character leading them) and just say it moves towards, shoots and charges the nearest unit. Split attacks equally between models in base to base.

If you're planning on facing off against AI skaven, you could always make it fluffy and have them attack the smallest unit/most wounded character.


Please keep us posted on what you decide on and any results. I'd be really interested to see how this goes. :) 

Link to comment
Share on other sites

Will gladly do. I appreciate the thoughts you shared. Been sitting on the thought of this for a bit now. Made a few crooked houses to give it a more skirmish feel.

There's a few "disrupt the ritual" type battleplans in AoS. Could do something like: Break through the squad of whatever evil things to stop the ritual then fail to stop the ritual and have half a dozen heroes face off against a Bloodthirster or a confused giant squid.

Squad just stands there guarding until in range. Ritualists just stand there casting no matter what. Maybe don't even attack back but have a +1 save and ignore rend. Big guy summoned just stands there smashing my head into the wall. Would make for an easy AI mechanics.  

Link to comment
Share on other sites

I have a "AI chart" ive been working on. Its intended for larger games than the one originally mentioned and is very much a work in progress but it might be possible to ammend it to suit your needs.


It works on the AI having a number of "Stances" which result in the AI following a phase by phase plan (E.g. Aggressive is geared towards reaching any objectives (Enemy units being objectives in normal games) with commander and wizards buffing the closest to combat".

Stances are determined at the start of each round with a dice roll.

To help the AI decide you need to assign objective points (usually coinciding with victory points or unit kills in normal game) this effects the likely hood of the AI changing stance. (If it gains lots of objective points it will likely stick to its current stance.if not it will likely change and swap tactics.)

There is still a few holes but the few trials ive done seem promising.

Link to comment
Share on other sites

Anything helps. Bouncing ideas generates great games. 

So you are saying that there is an adaptive behavior implemented in your AI? If they feel like they are winning (read: getting objective points) they keep doing their thing but if throughout the round they fail to get any, they may change stance/behavior? At some point so much that they don't roll for a chance of stance?

I assume some formula is implemented where an X objective points gained correlates with the likelyhood of changing stances? 

This could be a really cool thing to try and perfect for an evening of "Kyle, we lost because you went left while we all went right!"

Link to comment
Share on other sites

Ive tried to keep it simple.

you roll a d6 at the start of the round if the result it 4 or less it will change its stance. however you add a +1 modifier for each objective point so essentially if it gains 4+ objective points it will keep the same objective regardless.

The only difficulty is the player needs to decide what an objetive is worth. If there is only one objective then you cant simply assign a single point for it otherwise the AI will never gain eneough points to follow a set path. So there is a little bit of a choice to be made there. I may add some battleplans once its finished with tried and tested objective point numbers.

Im also thinking of adding a automatic pass number to the Stance roll (E.g. 3+ OP mean the AI keeps that stance no matter what)

Just finsihing up an initial draft will upload it once its finished.

Link to comment
Share on other sites

I have thrown this together quite quickly.

I am already going through the wording to get rid of logic loop holes but figured it would be quicker if i just posted it to get more eyes on it. Obviously the idea is to make it so there is little or no interpretation from the player.

Also ive only done a plan for aggressive and defensive so if people have some ideas on some other good stances that would be great.

Im trying not to make this any more complex that it needs to be.




Link to comment
Share on other sites

I've read it. Thank you for investing your time into putting it together. 

I agree with the less is more at this stage. Maybe there won't even be a need for more stances in the beginning. Have 1 3 5 and 2 4 6 roll mean either A or D stance. 

After getting used to it I'd try to incorporate a stance unique to the shooting units where they would attempt to move away from the enemy just enough to be in the firing range and then shoot. Would be cool to have gobbo archers do fluff like this. 

As for objectives - yes, it would have to be discussed in the narative. Agile development of objectives that are small enough to constantly be either achieved or failed yet significant enough for them to be considered an actual achievement. 

What is your thought on wounds being objectives? Wiping a unit can take a while and if my unit caused wounds to the enemy unit it might add vigor to my troops and make them want to push harder. At the same time if a bunch of skellies cant scratch my mega boss, it could be... nevermind... if a bunch of skaven cant scratch my mega boss it could affect their morale. Can wound difference in close combat be a modifier? 

Also, difference in battleshock tests can be used as modifiers to stance roll. Say I lost 2 models and the AI lost one. He killed more and feels good about himself. +1 to aggro. 

Link to comment
Share on other sites

After posting this i had a similar thought in that it may be better to stick to aggresive and defensive and instead have . . . .Sub stances? essentially just applying a little more details with the aggressive trait. (E.g. Bezerker: Just charges units blindly at the enemy. or Hunter: directs ranged attacks and spells against enemy heros)

I think i mentioned it briefly at the beggining although if cleaned it up now but i went for 15 wounds (Just pulled the number out of the air) gains an objective point. Also I feel adding the EXTRA section that adds to the objective points is effective as it plays directly to what the AI is trying to achieve. (E.g. if the AI goes for an all out aggresive attack but ends up losing more units than it wipes out it will effect the Stance roll and cause it to be more likely to change stance.

However iwouldnt want wounds to be overly involved as it would depend on the game objectives. I played a Narrative battleplan the otherday called "War of storms" from the generals handbook. Essentially there is a line in the middle of the board (the storm) which you  are trying to push all the way to the other players side. one way of doing this is by wounding their commander. This was my undead against his ironjaws and i won by tarpitting his forces while my mages sniped at his general. If i was an AI and based by decision to push my forces forwards "Aggressivly" largely on wounds then i would have changed stance long ago as my army was in tatters by the end while he still had loads. but i won because their deaths suited the objective.

So wounds should be a factor but not the factor me thinks. If i gained a couple of objective points each time the storm got pushed back im sure an AI could off pulled off a similar win. (Got me thinking about pitting AI V AIs now XD)

I feel like there would need to be a generic chart for these (Perhaps next to the stance one) that summizes these different modifiers to the AI choices. for simplicites sake it you could put battleshock loses under the wound count modifier (So losing a knight in battleshock would be the same as 2 wounds in the combat phase)

I would very much like this to be a pick u and play supplement however i may add some battleplans that incorporate a few unique stance modifiers to suit the objectives.

Link to comment
Share on other sites

I agree. 

Also, was wondering about that 15 wound mark. lol Thought you might've gone deep into calc and figured out the optimal number of objective generation via wounds. 

Here's what I'm gonna do for the first game: 

Inspired by William King's Daemonslayer book's skaven raid, I will have a fortified manor resembling the resting place of a group of characters. I'm torn on the amount of enemies so I'll go with double the wound count at first. So if 6 characters at 5 wounds each (2 human players, 3 characters per player, 1 character counts as 1 unit and moves separately), Ill have enemies (AI from now on) totaling 60 wounds and maybe a troll/rat ogre/chaos spawn to spice things up. 

Where possible, it will be a single wave (Think: 15 ogors) where not, several waves ending with a bigger fight (Think: gobbos, skaven, chaos warriors, using one of storm of sigmar battleplan mechanics) . Not sure how much bigger of a final fight, because outside of their rules (Bugman's ale or Grombrindals heal) characters won't regen hp. 

Inspired by both Double Misfire and KHHaunts's suggestions, I'll start off with basic:

Roll for who goes first/AI Stance Roll: 

Players always take turns controlling AI (Players familiar with runemarked player concept from WQ:ST will feel at home). If during the last round I controlled the AI, next round my ally will control the AI. However, you still roll for who goes first and choose between you, your friends or AI. Consider yourself playing a backstabbing goblin and an opportunity to make AI take 2 turns in a row and kill your ally while you escape. I'd leave it just to throw someone under the bus (...steam tank?) and make for a great story to tell here. 

Hero Phase: Will be very selective with mages I take to my games. Limited simple spells. Same with the boss AI and their command abilities. I like KHHaunt's idea to buff the engaged unit of AI with command ability and maybe roll a D6 if multiple units will be engaged to keep it simple. I'd remove them altogether but I am trying to keep all the AoS features in here to gradually increase the scale of the games and educate on much of what the game has to offer.
Movement Phase: Stronger AI moves to stronger character (possibly less wounded or closest if confused), weaker AI goes to wounded or closest. Outside of charging range, non-shooting AI will always run. Still, do consider case by case fluff scenario where stronger skaven would pick on the more wounded character.  
Shooting Phase: Shooting is strong in AoS. Ill go with no more than 25% of hostile AI can be allowed to shoot. Especially if Ill field an AI mage. However since I am trying to introduce most mechanics of AoS play, I will still be fielding shooting AI.
Charge Phase: AI attempt to charge the closest unit automatically if in range (12" if I'm not mistaken). Let's go with Total War:WH engagement. If you run into them - you run into them. 
Combat Phase:  Split attacks equally between models in base to base. Will be on a lookout for other tweaks that may come up. 
Battleshock Phase: AI - normal BS tests. Heroes - BS tests from AoS regiments of renown or none at all depending on your preference. I will assume that my guys will either be extra brave or have nowhere to flee.  

The Rule of Carnage: In Skirmish Play, where models can end up
fighting one-on-one a lot, a modified version of Battleshock is in
use. A test is taken when a model loses a Wound. When taking
a Battleshock test, count the number of Wounds lost rather than
models slain when making Battleshock tests.

Taken from here.

I like the idea of stances for AI and will try to modify the stance roll by the amount of models who flee during the BS tests for AI and the amount of wounds caused to characters. If, however the stance roll drops -6 (for now I chose negative D6. Will think on this after a game since a pack of 40 skaven vs 5 leaders can easily get -15) and more, the AI unit in question must immediately turn their back and run until the next round. If their next stance roll is defensive, the unit is removed from the table. Do not apply any modifier to a stance roll made after the AI unit runs.  

For instance, if 2 AI ogres flee as a result of a BS test - they will get -2 (More defensive/reserved/scared) to the stance roll at the beginning of the next round. If these same AI ogres cause 2 wounds to my character - they get +2 to it (more aggressive, reckless). 

Getting a 6+ on a stance roll (after modifiers) makes the AI unit berserk applying the "Inspiring Presence" command ability to that unit until next stance roll. 

I don't think I'll be using AI that can fly or artillery unless I feel extra kinky in the later games. 
Triumph: I do think triumph system is great for this kind of play since your heroes are always gonna be outnumbered by AI and gain more stats for future games would feel more engaging for me. Until I get bored with writing it all down. 
Allegiance Abilities/Artifacts: I do think I'll be going with artifacts and allegiance abilities (since I'll be fielding Dark Elves and Lizardmen on one team, it won't often be used though). Artifacts will be available for each characters but no 2 of the same artifacts will be allowed. I'll probably be choosing new artifacts every new game. 
General: Meh. I'll try using all command abilities available. Then not using any and see where it gets me. 


In this particular "Survive the Night" match:

If they win - they set off on a quest to kick some teeth. Field fights and tbc.

If they lose - they wake up in the dungeons of the attackers surrounded by foes and having to save each other before they attempt escape. You'd start with 1 character each and then gather the rest from cells throughout tunnel fights and sneaky stuff. All triumphs and artifacts will be lost. I almost wish for them to lose now. 

In the games following, the story of the band of Groomy's Guild of Heroes will meet with new unlikely ally characters who would join them in future games, fight friend and foe, face ghost dwarfs, inquisitive stormcast, endless swarms of chaos and greenskinz on their way to safety, treasure hunt or just plain attempt to survive. They will awaken the dead and monstrous behemoths from slumber and at times achieve what gods couldn't. Meeting death would not be their worst of fears. 

I welcome all corrections and feedback as well as ideas and thoughts. I do not assume that this covers all possible outcomes or balances things out but to me this makes up for an intriguing series of small Co-op vs AI battles. 

Link to comment
Share on other sites

I made an AI system a little while ago and posted it on Dakkadakka. It uses the keywords to decide if a unit is "offensive" or "defensive" and then gives them action points for each phase in the turn. By coincidence I was in the process of updating it and will post it up soon. You guys might be able to take one or two usually ideas from it :-)

Link to comment
Share on other sites

Kind words @MacDuff Cheers. 

By all means, please share what you fancy with us. May it be a further inspiration to KHH and I and may we all praise what you've done together. 

I still need to read your Hinterlands scenario. Saw it this morning @bottle. Does it assume you take a small number of battleline units like RoR or can I play with heroes such as the almighty White Dwarf too? 

Link to comment
Share on other sites

Whatever works for you @KHHaunts. I find my efforts are incomparable to the quality that you or @bottle put into your rulemaking so Ill just be glad to have more ways to play with my hobby. 

@bottle Would you educate me for a second? What's the most common reasoning behind forbidding named characters in the settings like ours? Is it because I can take Archaon and no matter what is thrown against him will be sat on? I mean, I have a warband of: Gotrek (Dragon Slayer/Unforged), Felix (Knight Questor w/o the shield), Snorri (Unforged), Malakai Makaisson (Cogsmith), Josef Bugman (Named Char) and Grombrindal the White Dwarf (Named Char). I'd very much like to use them all in whatever twisted reason behind them appearing in the Mortal Realms may be. 

Josef is comparable to a non-named toon. WD on the other hand I assume is extra strong if you go along with his rules. Will I struggle finding an interesting opposition (read: amount of AI models) to challenge this bunch? 

Edit: This thread has no pics. Mine ain't pretty but adds some character :D Still a wip on a few things.


Link to comment
Share on other sites

No worries. Just wouldnt want to bog your thread down with generic AI stuff if your looking for infro with regards to the particular scenario your looking to put together.

Is your question with regards named heros asking if their stats are to OP for what your looking to do?

Link to comment
Share on other sites

Its hard to tell which direction AOS is going at the moment as nearly all the named characters are VERY powerful. There is no new small scale named characters so its hard to tell if all this OP is just because of how grand the models are.

However many of the old compendium heros arent much stronger than the generic counter-parts (With some exceptions obviously). I think in this type of scenario if you put a limit to what sort of heros can be involved. Like no dragons allowed (I know your playing with dwarfs. just an example) i cant see any of the smaller character being to OP for such a game.

As for the particular dwarf list youve gone for. looking at there profiles they seem very well suited to a scenario like this. even the named ones.

Grombrindal is definately the most powerful. Perhaps thinking him as a leader choice.

Surely a few additional elite units or rat ogres will balance it out without to much effort?

Link to comment
Share on other sites

I appreciate your thoughts on this. Decided on the same, that if the first game with this bunch turns our to be a 5 minute slaughter, I'll just bring an AI giant/dragon/daemon/Lord-Prime next time and we will see who will have the last roll. :D

I had Bugman facing a unit of 5 ogres and he only could take out a couple before they stomped him into the ground. Beer or no beer. Very strong folk.  

Link to comment
Share on other sites


This topic is now archived and is closed to further replies.

  • Create New...