Slight concern

[JamesLxx]

From watching utube aars ive noticed that the AI has a tendency to keep on advancing whole formations of into heavy fire and get knocked out one by one. Surely on the first few losses a formation would pull back to cover, especially lighter vehicles when they come up against tanks.

[Veitikka]

ORIGINAL: JamesLxx

From watching utube aars ive noticed that the AI has a tendency to keep on advancing whole formations of into heavy fire and get knocked out one by one. Surely on the first few losses a formation would pull back to cover, especially lighter vehicles when they come up against tanks.

What should they do after they've pulled back and reached cover? Stay there and ignore the objective they're supposed to capture? Perhaps you can come up with an example of such a situation so we can analyze it and discuss how to improve the AI opponent.

[Werezak]

Well, ideally the AI would identify which areas on the map are dangerous, and avoid putting units there.

One way to conceptualize danger is to think of a heatmap, where the intensity of the map depends on the amount of firepower of enemy units that can shoot into that part of the map. As the AI learns of new enemy units, it updates its map. It should be trying to move its units so that they can shoot at the enemy units (i.e. get LOS on known enemy locations) while avoiding these dangerous areas. This just emulates what a player naturally does, and would naturally lead the AI to pulling back and instead focusing on an (implicit) intermediate objective: neutralizing the defending enemy units.

If a player attempted to probe an objective, and came under fire and started losing units, they would pull back before committing too many important units. They would then attempt to destroy the source of the hostile fire before continuing, right?

If we think about it, the goal of the AI isn't to advance into objective locations! Rather, it's goal is similar, but critically different: to clear out enemy defenses so that units can safely advance to the objective locations afterwards.

No player just drives their units into the objectives from the get go. Instead they first scout, to determine roughly which locations of the map are defended, then they clear out the defense while keeping their own units covered, then they advance to the objective once it's safe. It's complex but it can be broken down in simpler steps I'm sure.

EDIT: Just wanted to add that I really hope this is clear, because it is quite subtle: The goal is not simply to have units move into objectives. The goal is to ensure units get into the objectives alive and stay alive while they are there. As the AI encounters resistance, it's going to have to generate new intermediate goals for itself, and defer the final goal of actually putting a unit into the objective zone until later.

[varangy]

Can the AI be modded?

[philip wells]

ORIGINAL: varangy

Can the AI be modded?

Yes its called either a patch or DLC

[Veitikka]

ORIGINAL: dholedays

Well, ideally the AI would identify which areas on the map are dangerous, and avoid putting units there.

One way to conceptualize danger is to think of a heatmap, where the intensity of the map depends on the amount of firepower of enemy units that can shoot into that part of the map. As the AI learns of new enemy units, it updates its map. It should be trying to move its units so that they can shoot at the enemy units while avoiding dangerous areas. This just emulates what a player naturally does.

If a player attempted to probe an objective, and came under fire and started losing units, they would pull back before committing too many important units. They would then attempt to destroy the source of the hostile fire before continuing, right?

If we think about it, the goal of the AI (or players for that matter) isn't to advance into objective locations! Rather, it's goal is similar, but critically different: to clear out enemy defenses so that units can safely advance to the objective locations afterwards.

No player just drives their units into the objectives from the get go. Instead they first scout, to determine roughly which locations of the map are defended, then they clear out the defense while keeping their own units covered, then they advance to the objective once it's safe. It's complex but it can be broken down in simpler steps I'm sure.

I think one major issue is that most of the kills are made by units that are not seen by the opposing side. If you see the enemy then you probably can kill it. This is not a strategic level game where you can calculate the front lines, and then shift your resources accordingly. Was your extremely valuable tank killed by a cheap AT weapon 100 meters away or an ATGM 3000 meters away? In many cases you just don't know. Is the 'dangerous' area around the killed unit, or around the killer? Or somewhere else? Should the AI cheat and know the attacker's weapon and ammo parameters, so the LOS from that unit can be calculated? Your armored column advances in a forest, and the first vehicle explodes. Who killed it? Was it just a single unit in ambush or a company of units hiding there? What do in that case? In most cases the fog of war hides all parameters, so there's no data to process. How to 'clear out' units that are killing your units? Send formations to their positions, instead of the objectives?

The technology we're using doesn't understand abstract commands such as 'be more like human'. It understands zeros and ones. So, more detailed instructions are needed. Perhaps screenshots showcasing situations where the AI did something wrong, and what it should have done instead?

[Veitikka]

Now here's something more concrete. Your tank platoons advance, follow the roads, cross the river, and they're ambushed on the other side. Bodies start piling up. What to do? You have a few mortars and some mech infantry following close behind.

[Veitikka]

The area there is burnt forest and mixed/leafless forest. You enter it, and you're blown to pieces. You don't know where it came from.

[Veitikka]

So please tell me, how should the AI function here so it would be like a 'coordinated WarPact attack', not 'piecemeal' but still 'unpredictable'?

[varangy]

The problem is that people who never programmed cannot imagine why its so hard to program an AI. (or any other complex thing)

[Scutarii]

The only problem i see with the AI is the lack of coordination in certain situations, it is capable to send to the death a company of BTR-60 and after human player kill them all appear the first MBTs with AT missile support.

Program an AI is not easy but maybe with the different postures AI can adopt (infantry-mech-armored-dinamic) you can made that in mech-armored postures AI mix better MBT-AFV/APC.

My suggestion is in mech posture made that for every company of infantry AI buy 1 MBT platoon or section (WP has platoons of 3 and NATO can have sections of 2) and in armored posture for every company of MBTs buy a section of mech infantry (NATO) or a platoon of mech infantry (WP) in the moment AI can buy individual platoons a balance of armor-mech infantry is interesting.

In combat maybe dont send solo AFVs-APCs to move behind scouts is necesary, they need MBTs and oposite to, MBTs need mech infantry, if you have more infantry than MBTs MBTs move with them and if is oposite MBTs move in first line with AFVs following them at certain distance.

I dont see bad AI push to far and lose a lot of troops, in the end is the only way to take an objetive, you need asume casualties are going to be present, number of casualties is going to be related in how AI attack and the specific tactical situation (same scen 2 different results based in the effect of arty VS enemy defender for example) i see the problem in send only 1 type of unit to push leaving the other type in rear waiting human player kill them all... i see a few coordinated attacks mixing MBTs and AFVs-APCs and they work better sometimes but simple send one type forward usually ends in a disaster.

PD: in the video German VS German you can see how human player advance with MBTs and take objetive and AI counter MBTs in city ATTACKING WITH BTR-60!!! after human player destroy them all appear some AT vehicles and MBTs (T-72M) and kill the human Leos relative easy maybe with a better mix in the AI attack human player has a less confortable defense at same time needs adapt earlier to the presence of enemy MBTs.

[Veitikka]

ORIGINAL: Scutarii

The only problem i see with the AI is the lack of coordination in certain situations, it is capable to send to the death a company of BTR-60 and after human player kill them all appear the first MBTs with AT missile support.

Sometimes it can be a good idea to advance with mech infantry in front. For example, if in the above example I posted you send expensive tanks to the forest they will be ambushed.

Program an AI is not easy but maybe with the different postures AI can adopt (infantry-mech-armored-dinamic) you can made that in mech-armored postures AI mix better MBT-AFV/APC.

The AI does make mixed groups of formations. However, I need more detailed instructions than 'make it better'.

My suggestion is in mech posture made that for every company of infantry AI buy 1 MBT platoon or section (WP has platoons of 3 and NATO can have sections of 2) and in armored posture for every company of MBTs buy a section of mech infantry (NATO) or a platoon of mech infantry (WP) in the moment AI can buy individual platoons a balance of armor-mech infantry is interesting.

The unit purchase algorithm is not related to this. The scenario designer can give the AI anything imaginable, and the AI must be able to handle it all.

In combat maybe dont send solo AFVs-APCs to move behind scouts is necesary, they need MBTs and oposite to, MBTs need mech infantry, if you have more infantry than MBTs MBTs move with them and if is oposite MBTs move in first line with AFVs following them at certain distance.

I dont see bad AI push to far and lose a lot of troops, in the end is the only way to take an objetive, you need asume casualties are going to be present, number of casualties is going to be related in how AI attack and the specific tactical situation (same scen 2 different results based in the effect of arty VS enemy defender for example) i see the problem in send only 1 type of unit to push leaving the other type in rear waiting human player kill them all... i see a few coordinated attacks mixing MBTs and AFVs-APCs and they work better sometimes but simple send one type forward usually ends in a disaster.

PD: in the video German VS German you can see how human player advance with MBTs and take objetive and AI counter MBTs in city ATTACKING WITH BTR-60!!! after human player destroy them all appear some AT vehicles and MBTs (T-72M) and kill the human Leos relative easy maybe with a better mix in the AI attack human player has a less confortable defense at same time needs adapt earlier to the presence of enemy MBTs.

The scenario designer may have given the AI zero tanks. Please show some concrete examples how this 'better mix of units' is built, and when/where it should be used.

[garga3]

To reply to the example given with river crossing.

A) The AI needs to evaluate the need to cross the river. What will happen if it tooks defensive posture and just hides in its own woods? If it is a least a draw than this is a valid option. Even if its a sure loss it can try to gamble(rarely) and play the def posture.

B) If it decides to cross it needs to do it with cheap/recon units until it knows there are no enemies that shoot in that woods.

C) If it loses its cheap units in that wood than this wood is not safe. Usually this will be due to enemy inside but it can also be due to enemy observing the forest from some location(this should be checked using LOS as the players does, without cheating) - Everything in LOS of the location of ambushed unit and out of your own LOS should be marked dangerous with different coefficient. If it is assumed there are enemy units in the wood, then one way is to bomb the hell out of the forest, to prepare a cheap force that excels in wood ops (basic infantry + some recon) and to send them in using smoke cover.

From the movies i saw i can say the following:
concerning point A. The AI is usually too aggressive. It takes defensive positions too rarely. The evaluation where to advance and where to defend needs to be reevaluated. The AI does not respond properly to problems after making its initial plan. It should make a new plan according to the new info it posses during battle. When you send in a major trust and it is ambushed your initial plan is void, you need a new one, and only try to bruteforce as a small random factor option(to still be unpredictable).

concerning point B. It seems it does sent cheap units, but usually not in good formation. The way they are sent could be improved, they seem to follow same path and this is wrong, recon units should try to identify maximum territory with minimum squads, this is not done when they are going one after the other over the same route.

concerning point C. The AI does not use artillery in good coordination. It seems to use far less arty than the player as a whole, this needs to be improved - the AI should try to use the arty more. When it does use its arty we can give comments if it does it right.


So the main things for me: over-aggressiveness and minimized(under-aggressive) arty usage. Having some way to adapt this would be great. As a side note a question: how much of its arty ammo does the AI use in a battle (as percentage)?? Note that as a player is usually more static oriented more arty will really hurt him more and encourage him to be more aggressive himself instead.

can the ai be modded (a config file or something)?

as a question for force composition: the ai needs to have sufficient recon(or to assign other units as recon if player does not give it recon units), infantry(non open terrain) and tanks(open terrain)(again it should virtually assign some other units to perform these tasks if it lacks resources), and sufficient arty/support.

For example if you have no recon(lost or not assigned) you should assign some other units(preferring cheaper with enough visibility and low profile), as in your example it is better to assign a mech unit on a recon mission then a tank in certain terrain, in open even a tank could do(but moving in open dangerous terrain is bad idea).

so if you have infantry only, you should assign 5-10%(or your pref number) of them for recon duties, some percent for defensive inf duties, some for offensive(objective storm), and some for virtual support anti-air, anti armor, anti-infantry and other specialized duties and possibly some in reserve. the defensive/offensive percent should obviously vary if you are stronger or your opponent is evaluated stronger.

[Veitikka]

ORIGINAL: gargamor

To reply to the example given with river crossing.

A) The AI needs to evaluate the need to cross the river. What will happen if it tooks defensive posture and just hides in its own woods? If it is a least a draw than this is a valid option. Even if its a sure loss it can try to gamble(rarely) and play the def posture.

In this example the AI is advancing and the player is defending. The only way to reach the objectives is to cross the river.

What should a 'defensive posture' be like? If the advancing group has tank platoons, mech platoons, AA/AT assets and mortars, advancing in a column formation, what happens when it takes a defensive posture? They just stop where they are?

Currently the AI can leave formations to defend objectives it has captured.

B) If it decides to cross it needs to do it with cheap/recon units until it knows there are no enemies that shoot in that woods.

C) If it loses its cheap units in that wood than this wood is not safe. Usually this will be due to enemy inside but it can also be due to enemy observing the forest from some location(this should be checked using LOS as the players does, without cheating) - Everything in LOS of the location of ambushed unit and out of your own LOS should be marked dangerous with different coefficient. If it is assumed there are enemy units in the wood, then one way is to bomb the hell out of the forest, to prepare a cheap force that excels in wood ops (basic infantry + some recon) and to send them in using smoke cover.

A lot of assumptions are made here. The AI doesn't necessarily have recon units, artillery/mortars or dismounted infantry. All this must be taken into account. It may have a few though, but not enough to include them in all maneuver groups. The map can be huge (15x15 km), so it takes time for units elsewhere to travel to these hotspots.

About checking the LOS to mark the 'dangerous areas': See the attached image. If the attacker hasn't been spotted, in many cases there's no way knowing if he's kilometers away or 100 meters away. On the other hand, if he's spotted, then he'll probably be killed very soon, especially if spotted by tanks.

From the movies i saw i can say the following:
concerning point A. The AI is usually too aggressive. It takes defensive positions too rarely. The evaluation where to advance and where to defend needs to be reevaluated. The AI does not respond properly to problems after making its initial plan. It should make a new plan according to the new info it posses during battle. When you send in a major trust and it is ambushed your initial plan is void, you need a new one, and only try to bruteforce as a small random factor option(to still be unpredictable).

concerning point B. It seems it does sent cheap units, but usually not in good formation. The way they are sent could be improved, they seem to follow same path and this is wrong, recon units should try to identify maximum territory with minimum squads, this is not done when they are going one after the other over the same route.

I agree. However, it's a long way from these abstract ideas to having it in the code. Is there a tactical wargame that does this without scripting or cheating?

concerning point C. The AI does not use artillery in good coordination. It seems to use far less arty than the player as a whole, this needs to be improved - the AI should try to use the arty more. When it does use its arty we can give comments if it does it right.

The AI does use artillery, but it's true that sometimes it doesn't use it much. In these cases there's a reason for it, and each case should be analyzed separately, without generalizing. If the AI has DPICM it can destroy your dug-in M1A1 platoons in minutes.

can the ai be modded (a config file or something)?

Factions do have a few AI attributes that can be adjusted, but nothing complex.

[Searry]

I think the AI has always just been the warm up and the human opponent is the real challenge because an AI can never be as fun of an opponent. AIs usually become predictable as you figure them out and then it just gets boring from there. Some games have an AI which can be fun even for a few hundred hours but those are really rare. The worst example of an AI related to a game like this is probably in any Close Combat. Completely unenjoyable.

[Mindfield]

The videos I have seen so far have not been fair for the AI.

AI attacking is something very hard to do.

Human player playing defence side or meeting engagements is not good.
In a meeting engagement scenario, the human player simply gets into a defensive position and waits for the AI to come and exploiting it.

The old trick in such cases is to give the AI a bonus, to make it somewhat challanging or to let it cheat.

A strong AI you can only expect in turn based games (IGOUGO/WEGO) and still it is not easy.
Here it is further limited by real-time constraints.

It is better to let the AI defend and the human player to attack to make for a more interesting match.

Another idea is to restrict yourself is to only give commands to formations, like the AI does, and stop microing.

But we can ask, why is the AI not microing?

[Hexagon]

My point with how AI deal with combined arms is that many times you can see AI sending APC/AFV formations without MBTs support... for me AI is over agressive and lacks a little in mix better vehicle types... for example are videos where AI send mechanized mortars or missile AT vehicles towars enemy defensive line after MBTs and AFVs/APCs were destroyed and they die for nothing.

My suggestion is when AI buy mech and MBTs try made MBTs move in first line, behind scouts, with a few AFVs as support when AI has enough MBTs compared with AFVs/APCs (i think a ratio 1 MBT per 2 AFV/APC or better) when AI lacks MBTs made the few MBTs avaliable (if they are avaliable) move between the AFVs/APCs and not in rear to far to provide the support they need to deal with enemy units.

The point is prevent send AFVs/APCs ALONE to their dead VS enemy line of fire, is impossible for a BMP-BTR rush pass a defensive line with MBTs (even VS older MBTs) because die very fast, i dont see BMPs and other missile vehicles using them in movement, if they need stop to use missiles they are wrecked in seconds, if they are behind MBTs or with a few MBTs mixed between them in the attack MBTs can attract enemy fire and if is a modern MBT fire on the move and at least they can work as shield to give AFVs time to use the missiles and/or to made vulnerable infantry dismount.

In battle generator maybe if is possible you can add an option to made AI buy a little more MBTs, if i dont remember bad this was an option in SPMBT, and try adapt the buy list to the posture, for example a soviet armored division had 3 tank regiments and 1 infantry regiment in motorized divisions oposite, 3 infantry 1 tank, i dont say use this ratio but do something similar to adapt buy list to posture.

I dont say is bad at all as is now but AB offer maybe an extreme use of vehicles in AI area because is a little over agressive and sacrifice to much vehicles in a role that is not their role even when i see videos with AI doing a great job selecting the enemy weak areas and sometimes after a very good use of support (arty, helicopters...).