[OPEN BETA] [SP] bCombat infantry AI Mod

[fabrizio_t 58]

Okay thanks for detail information, will readjust my config to provided settings and will report then :)

Ok, let me know.

[fabrizio_t 58]

UPDATE:

i'm making good progresses on the packaging of the mod, which would be distributed via Steam Workshop.

I think that we may retain direct config.sqf editability while still fully packaging the mod.

More to come.

EDIT: in the meatime bCombat reached 70 supporters out of 100 needed for "promotion".

I get dinosaurs are much more interesting ... ehm ... dinosaurs, really?? :)

Edited September 22, 2014 by fabrizio_T

[KaneSvK 1]

It's unacceptable that dinosaurs get more attention than bcombat - voted ;-)

[kremator 1065]

To celebrate my 7000 post ... I have supported bCombat!

[phronk 898]

So far the best AI addon, aside from ASR AI 3. It's too bad it isn't compatible with MP / Dedicated Servers, yet... regardless, I've given you my vote as well, instead of it going to the happy dinosaurs or "Life" map.

[gliptal 25]

IMHO the voting community is way off from the actual playing and supporting community, don't expect to find sense or logic in these types of voting contests.

Yay!

[SavageCDN 231]

voted.. although I do like those dinosaurs.... :p

[Ara6orn 0]

Voted!, the best AI MOD.

[Bouben 3]

Fabrizio, is there anything you can do about effectiveness of AI units against armored units? Would it be possible to make them useful again and make their decision making and reaction times faster?

[fabrizio_t 58]

Would it be possible to make them useful again and make their decision making and reaction times faster?

Making their reactions faster should be feasible, that's basically the same code currently applied for CQB, except that's for hard targets.

This as long as AT unit doesn't refuse to fire on target or to switch weapon , as it did in ArmA2 ... to be checked.

Fabrizio, is there anything you can do about effectiveness of AI units against armored units?

I tried addressing that within FlexAI for ArmA2, some raw code already exists.

Matter is more complex than you may think at first, though.

AT fighting is not usually a matter to force unit A to fire on tank B as far / soon as possible.

For instance AT infantrymen should try targeting hard targets from flank or back, but due to detection, movement and pathfinding issues i consider this hardly feasible within ArmA.

Then different AT units should be able to pick different targets, but this usually does not happen in ArmA.

Let's consider the typical case of an ambush against more than 1 tanks/APC, by a squad containing more than 1 AT unit:

in vanilla ArmA (checked in ArmA2) any AT units will open fire on the same target, as soon as it's detected and in range.

This usually means that ammo is wasted on an already destroyed tank, while following one(s) will cut AT units and squad to pieces.

Handling a successful AT attack is tricky, let me enumerate just a few basic things:

1) a squad should theoretically refrain from opening fire at armor if it does not stand a chance (e.g. less AT units or ammo than number of hard targets).

2) single AT units should target the maximum number of targets at once, in order to theoretically inflict maximum damage (even if they miss or just damage some vehicles) with minimum risk.

3) if an attack fails to destroy the hard targets and AT ammo is depleted or AT units dead, squad should withdraw to hard cover as fast as possible, as it's already compromised and vulnerable.

These are just basic points, but they already have complexities.

Increasing AT efficiency is definetely not an easy task.

Edited September 24, 2014 by fabrizio_T

[Bouben 3]

Making their reactions faster should be feasible, that's basically the same code currently applied for CQB, except that's for hard targets.

This as long as AT unit doesn't refuse to fire on target or to switch weapon , as it did in ArmA2 ... to be checked.

I tried addressing that within FlexAI for ArmA2, some raw code already exists.

Matter is more complex than you may think at first, though.

AT fighting is not usually a matter to force unit A to fire on tank B as far / soon as possible.

For instance AT infantrymen should try targeting hard targets from flank or back, but due to detection, movement and pathfinding issues i consider this hardly feasible within ArmA.

Then different AT units should be able to pick different targets, but this usually does not happen in ArmA.

Let's consider the typical case of an ambush against more than 1 tanks/APC, by a squad containing more than 1 AT unit:

in vanilla ArmA (checked in ArmA2) any AT units will open fire on the same target, as soon as it's detected and in range.

This usually means that ammo is wasted on an already destroyed tank, while following one(s) will cut AT units and squad to pieces.

Handling a successful AT attack is tricky, let me enumerate just a few basic things:

1) a squad should theoretically refrain from opening fire at armor if it does not stand a chance (e.g. less AT units or ammo than number of hard targets).

2) single AT units should target the maximum number of targets at once, in order to theoretically inflict maximum damage (even if they miss or just damage some vehicles) with minimum risk.

3) if an attack fails to destroy the hard targets and AT ammo is depleted or AT units dead, squad should withdraw to hard cover as fast as possible, as it's already compromised and vulnerable.

These are just basic points, but they already have complexities.

Increasing AT efficiency is definetely not an easy task.

Interesting but I would try to make things simple and at least a bit functional.

The biggest problem I have is that it takes AT guys so long to aim and fire while armored units aim and fire much faster and even spot AT guys immediately. That creates a huge imbalance in the gameplay. A vehicle will rush a squad's position and AT guys are unable to do anything. Before they prepare their weapon and decide what to do, they are already dead and the rest of the squad too. Those APC rushes are a complete game-breaker. Even if the vehicle has to rush for 200 meters, it is too fast for AT guys to do anything. Part of it is, in my opinion, because of unrealistic spotting and aiming abilities of armored vehicles.

Also, I would completely omit the number two point of your post. I personally don't find it important at all in the current state of the game because right now we does not even have a basic functionality right.

[fabrizio_t 58]

Interesting but I would try to make things simple and at least a bit functional.

The biggest problem I have is that it takes AT guys so long to aim and fire while armored units aim and fire much faster and even spot AT guys immediately. That creates a huge imbalance in the gameplay. A vehicle will rush a squad's position and AT guys are unable to do anything. Before they prepare their weapon and decide what to do, they are already dead and the rest of the squad too. Those APC rushes are a complete game-breaker. Even if the vehicle has to rush for 200 meters, it is too fast for AT guys to do anything. Part of it is, in my opinion, because of unrealistic spotting and aiming abilities of armored vehicles.

Also, I would completely omit the number two point of your post. I personally don't find it important at all in the current state of the game because right now we does not even have a basic functionality right.

I think it would be possible to:

a) prioritize hard targets in sight vs. soft targets for AT units

b) try to force them to open fire on target, when target is visible and in range (let's suppose unit won't refuse to open fire and it will switch weapon accordingly to commands).

This should probably suffice to make AT units more reactive to threats (and it was already scripted within FlexAI).

Then it would be possibile to reduce actual firing accuracy for vehicles (better dynamically constraining accuracy to vehicle speed: more speed, lesser precision) and maybe even reducing turret rotating speed.

That's what i suppose, at least. Theory and practice often clash in ArmA modding.

It would take some experimentation, i'm not excluding addressing this at a later stage.

EDIT: by the way, nice signature ;) Thx for supporting.

Edited September 24, 2014 by fabrizio_T

[Bouben 3]

I think it would be possible to:

a) prioritize hard targets in sight vs. soft targets for AT units

b) try to force them to open fire on target, when target is visible and in range (let's suppose unit won't refuse to open fire and it will switch weapon accordingly to commands).

This should probably suffice to make AT units more reactive to threats (and it was already scripted within FlexAI).

Then it would be possibile to reduce actual firing accuracy for vehicles (better dynamically constraining accuracy to vehicle speed: more speed, lesser precision) and maybe even reducing turret rotating speed.

That's what i suppose, at least. Theory and practice often clash in ArmA modding.

It would take some experimentation, i'm not excluding addressing this at a later stage.

EDIT: by the way, nice signature ;) Thx for supporting.

Excellent. Thanks, Fabrizio. Will be looking forward to it.

And the signature is a matter of course. You have done the best AI mod so far. I have to support it.

[tortuosit 486]

I supported you, but also had a look at MANW page... Looks like successful stuff is the kiddie supported stuff. Epoch mod etc. If you want to be successful, dump bCombat and create a vanity case mod. :( Indeed we're a 100% split community. But fortunately, the other part of the community is not here. I don't know where they are. Probably in Youtube comments and I hope they'll stay there.

[fabrizio_t 58]

I supported you, but also had a look at MANW page... Looks like successful stuff is the kiddie supported stuff. Epoch mod etc. If you want to be successful, dump bCombat and create a vanity case mod. :( Indeed we're a 100% split community. But fortunately, the other part of the community is not here. I don't know where they are. Probably in Youtube comments and I hope they'll stay there.

I was thinking about a zombie-dinosaurs survival game. Looks promising, isn't it?

[Not meant to be offensive to anyone, just joking - i respect anyone's hard work]

Seems bCombat is approaching promotion (86 outta 100 now), time for me to fiddle a bit with Steam workshop mod packaging, as promised.

However MANW is just a contest, let's not take it too seriously.

Edited September 24, 2014 by fabrizio_T

[Bouben 3]

However MANW is just a contest, let's not take it too seriously.

Sure. Moreover, it is written in the rules that number of supporters have no influence on whether the addon have chance to win or not.

A winning addon will be chosen by a jury - not by community.

[froggyluv 2136]

Now the REAL trick would be in forcing the AT guys to fire at that TRex in an efficient manner without sh*tting their pants.

Not familiar with how the contest works, does Support = Vote or just to get it in the competition?

[Bouben 3]

Now the REAL trick would be in forcing the AT guys to fire at that TRex in an efficient manner without sh*tting their pants.

Not familiar with how the contest works, does Support = Vote or just to get it in the competition?

Support is not a vote.

See here: http://makearmanotwar.com/vote

---------- Post added at 18:27 ---------- Previous post was at 17:57 ----------

Fabrizio, one thing I have noticed.

When units are suppressed, they will go prone, right? Would it be possible to make them go prone only if any known enemy units are more then...lets say 100 m away?

I have found out that AI is getting killed a lot because they choose to go prone in CQC and try to hide to cover (in prone) which is a slow way of doing it. If they run to the cover instead, it will be much better IMO.

[sttosin 67]

Support is not a vote.

See here: http://makearmanotwar.com/vote

---------- Post added at 18:27 ---------- Previous post was at 17:57 ----------

Fabrizio, one thing I have noticed.

When units are suppressed, they will go prone, right? Would it be possible to make them go prone only if any known enemy units are more then...lets say 100 m away?

I have found out that AI is getting killed a lot because they choose to go prone in CQC and try to hide to cover (in prone) which is a slow way of doing it. If they run to the cover instead, it will be much better IMO.

I agree with this request!

How about Suppress means

A. Crouch

B. Get up and run to cover. Set speed = 100 or something.

[Bouben 3]

Also, Fab, units should not be running crouched when ordered to move a significant distance. It is painfully slow and also fatiguing. Right now they only seldom stand up when moving in combat.

EDIT: But I am not sure if it is problem cause exclusively by bCombat or something else is happening too.

Edited September 24, 2014 by Bouben

[bravo409 13]

Been away from arma for awhile what is this support Bcombat thing I'm hearing about? How can I support Bcombat, I mention bcombat to everyone in steam that ask about ai mods.

[Bouben 3]

Been away from arma for awhile what is this support Bcombat thing I'm hearing about? How can I support Bcombat, I mention bcombat to everyone in steam that ask about ai mods.

Here: http://makearmanotwar.com/entry/Z3gpQ2HmCy#.VBaiqvl_u-4

[fabrizio_t 58]

Support is not a vote.

See here: http://makearmanotwar.com/vote

---------- Post added at 18:27 ---------- Previous post was at 17:57 ----------

Fabrizio, one thing I have noticed.

When units are suppressed, they will go prone, right? Would it be possible to make them go prone only if any known enemy units are more then...lets say 100 m away?

I have found out that AI is getting killed a lot because they choose to go prone in CQC and try to hide to cover (in prone) which is a slow way of doing it. If they run to the cover instead, it will be much better IMO.

There's already a minimum threshold for prone switching, which is actually 30 meters. I think 50 would be reasonable, but 100 meters seems quite alot to me.

---------- Post added at 22:09 ---------- Previous post was at 22:05 ----------

Also, Fab, units should not be running crouched when ordered to move a significant distance. It is painfully slow and also fatiguing. Right now they only seldom stand up when moving in combat.

EDIT: But I am not sure if it is problem cause exclusively by bCombat or something else is happening too.

Within bCombat crouched position is always temporary, like prone.

Units do crouch for some seconds after being fired upon, or when some enemies are known to be close.

Unless they've been repeatedly suppressed units won't crouch for more than about 30 seconds.

[Bouben 3]

There's already a minimum threshold for prone switching, which is actually 30 meters. I think 50 would be reasonable, but 100 meters seems quite alot to me.

---------- Post added at 22:09 ---------- Previous post was at 22:05 ----------

Within bCombat crouched position is always temporary, like prone.

Units do crouch for some seconds after being fired upon, or when some enemies are known to be close.

Unless they've been repeatedly suppressed units won't crouch for more than about 30 seconds.

Thank you for the explanation. The crouch thing sounds very logical.

About the prone thing: I consider 50 meters to be a really short distance to go prone, really. It is a distance around which combat is still really fast and aggressive. Going prone is slow and therefore hindering unit's performance, in my opinion. 50 meters is a distance where it is very unlikely to miss a prone target with full auto fire. Fast reflexes are needed for such a distance, in my opinion.

[phronk 898]

"-Surrendering behaviour"

How can you try to force the AI to surrender? I've been setting up random battles in the editor to see how the AI fight differently and all that, and don't think I've seen AI surrendering to AI yet.