removing cruncie-safe - another FAQ

[NEpi]

as much as i try to make those M2s survive the battle and fire their TOWs, i rarely do succeed. they usually wonder from the BP up the ridge and get blown to bits.

what to do to make them stop?


and another one: how do you use you mechs? do you prefer them as ATGMs or do you deploy them as infantry?

[BrokenArrow]

This is a tough area, but if you route them and give them a Hold or Defend BP at the end of the route, like any units or troops given these orders, they will move up to 200m looking for the best BP own their own, and will move the same now and then to avoid artty strike on their postions.

IF you route them and just leave it there.. no orders, they will stay there.

Now, there is no perfect order that makes them behave normal, in that there is an area, about 800m, that they fire those missles with good accuracy, and yet I have seen them almost run into a tank before taking aim and firing and it misses.

My best advise is route them to semi edge of forrest, give them "hold fire" orders when the get there, and when the spot target, jump and see what they see, and if flank or rear type shot, give them "fire at will" orders and generally they do well.

However there is no known "exact science" (that I know of yet) to get them to act rational all the time.

Kinda LOL :?

[NEpi]

i tried to use them as overwatch, and they failed miserablly.
can't jump in all the time for overwatches too. it diverts my attention from the more dangerous situations :?

and what about the tactical question?

[Hell_Hound]

If you have time in the planning phase, create some flank-shot BPs in various places. Then make an Engage/Slow route to each one from a hidden location, with the Embark-if condition "Trigger 1 set".

It's easy to do, and all the Brads will pop up when you hit Shift-F1. (Most of them will still get massacred, but at least you can sort of coordinate their efforts with the rest of your force.)

In front-on engagements, Brads lose to tanks regardless of any range advantage the TOW might have on paper.

[NEpi]

do you use them mostly as PCs, then?
do you use them for infantry assaults?

[BrokenArrow]

do you use them mostly as PCs, then?
do you use them for infantry assaults?

:arrow: Both

First, if you LOS a perfect BP for the brad, then once they get there, deploy the troops a min. of 1000m away, that way if you choose to move the brad, they won't "wait" for the troops to come back and get in. If troops are futher than 800m (rule of thumb) they will move right away to the given BP. But if you route the Brad closer than say, 500m (rule of thumb) to the troops, they will stop and troops will return. Terrible waste of time.

Now about the brads popping up and exposing themselves.

If you LOS a good BP route them there with "no orders", once there, move them (drive them) to the perfect HD BP and face them in direction you want covered. Then turn off the engine.

As far as troops, again, if you give them "hold fire" orders and use them as scouts/eyes, just route them and do not give them any other orders or they will move now and then and you know what that looks like (exposure, confusion, and death). And, to jump and see what they see only takes but 15 sec. max, and then you can give them orders to fire at will, and they will launch missles at given/seen target if less than 1200m (rule of thumb).

You can also hit the shift-key and Lase button at the same time (I think lase) at given target and a asteirok like star symbol will appear at or by the given target, and alerts the troops a/o a Brad, of target and decreases the time it takes for them to aquire the target and shoot.

One other thing, when you route anything to a given checkmark, once there, clk on the end chkmrk and "delete it" or, if you move any given unit from that mark and leave, it will move back to that chkmark if it is still on the map.

Hope this helps.

[NEpi]

it does, thanks.

i'm still annoyed by the amount of tweaking i have to do to have reasonable mech inf. force...


i think i've found a way to use the brads for inf. support efficiently enough. they're great at inf. assault, but lousy tank killers

[Ssnake]

If you have a specific suggestion for an behavior algorithm that makes better sense tactically, we'd be very interested to hear about it. I don't want to mock you here: I'm in for real. Since we haven't worked much so far on the computer control logic, I am hereby inviting you to invent a set of heuristics (=rule of thumb) how an IFV should behave in SB2. Keep in mind that different IFVs may require different rule sets (e.g. M113 with cal .50 vs M113 with 7.63mm machine gun, BTR-80, Bradley, Marder, BMP-1, BMP-2, BMD-2, MT-LB, ...) They all have different strengths and weaknesses, and ideally the CCL design would reflect this.
It may also be a good idea to have both rule sets for unobserved units and those where the player is in the observer's position.

[Hell_Hound]

Does the CCL know what aspect of the target it's looking at? Can it differentiate between when it's part of a platoon and when it's alone?

[Hackworth]

er, M113 with 7.62? Are we getting that too?! It would be kinda fun to rat-atat-atat with a M-60 or other mounted MG aside from the .50 cal.

When we get the M2 ODS and the M3....it will help to be ale to control them further than pushing them around and telling them what to do...sounds like what they do to me at work...

[Ssnake]

Does the CCL know what aspect of the target it's looking at? Can it differentiate between when it's part of a platoon and when it's alone?
Yes. Units can also determine whether they have a chance to do at least SOME damage, given the distance to and the aspect of the target. It can also -within limits- identify platoons and sections.

[Ssnake]

er, M113 with 7.62? Are we getting that too?!
I didn't want to imply that, although I suspect that at some time it will be a rather natural expansion. The question is, when will it be - but since we're just daydreaming here for the fun of it, why should we artificially limit our considerations about this point?

[Hackworth]

Consider away without limitation dear Ssnakie! That's how we get the good stuff done, the good fight won, the....ah hell, you get it ;)

[colin]

Jeeps/landrovers with Recoiless rifles?

M1113 mortars carriers?

Lynx track recce?

Saladin armoured car................swang!

I had a dream!

[Skip]

If you have a specific suggestion for an behavior algorithm that makes better sense tactically, we'd be very interested to hear about it. I don't want to mock you here: I'm in for real. Since we haven't worked much so far on the computer control logic, I am hereby inviting you to invent a set of heuristics (=rule of thumb) how an IFV should behave in SB2. Keep in mind that different IFVs may require different rule sets (e.g. M113 with cal .50 vs M113 with 7.63mm machine gun, BTR-80, Bradley, Marder, BMP-1, BMP-2, BMD-2, MT-LB, ...) They all have different strengths and weaknesses, and ideally the CCL design would reflect this.
It may also be a good idea to have both rule sets for unobserved units and those where the player is in the observer's position.

IMHO, if it's an M113 the only "rule of thumb" should be to get into the nearest hide position and pull it in after you. :)

[12Alfa]

A suggestion might be for the APC's when they have their grunts deployed and moving forward they them selves move to keep within range of the veh mounted wpn. As in real life they support the squad on the ground best they can. When the squad comes under fire the APC should be in wpn range and if not move right away to support them. Also the should be looking in the direction of the squad and if they have the sights/TIS to warn the squad they shoud open fire thous directing the squad wpns onto the target with their fire.

Does this make any sence? better yet is this do-able?

12Alfa

[Hackworth]

That sounds like engage on slow speed routing... That can be accomplished now. Or am I missing something?

[NEpi]

If you have a specific suggestion for an behavior algorithm that makes better sense tactically, we'd be very interested to hear about it. I don't want to mock you here: I'm in for real. Since we haven't worked much so far on the computer control logic, I am hereby inviting you to invent a set of heuristics (=rule of thumb) how an IFV should behave in SB2. Keep in mind that different IFVs may require different rule sets (e.g. M113 with cal .50 vs M113 with 7.63mm machine gun, BTR-80, Bradley, Marder, BMP-1, BMP-2, BMD-2, MT-LB, ...) They all have different strengths and weaknesses, and ideally the CCL design would reflect this.
It may also be a good idea to have both rule sets for unobserved units and those where the player is in the observer's position.

i have 2 problems answering your challange:
1. i'm a tanker, not a crunchie, damnit!
2. writing algorythms isn't my bread and butter.

on the other hand, i know quite a few inf. COs that might help with that. i'll give them a call.
also, they might be expert at using APCs as support, but not as overwatch, as israeli APCs do just that. let PCs be personel carriers and the tanks the big weapon platform.
that's why israeli APCs have low fire-power and great armor and manueverability. their armor can actually be equivalent to a tank's armor.
i might need some help deciding upon other roles the brads might have.

anyway, i'm picking up the gauntlet. i'll just have to have some help to make it.

here are some links about those APCs i was talking about

http://www.globalsecurity.org/military/world/israel/weapons.htm
http://www.geocities.com/joritwintjes/Neue_Dateien/APCs.html
http://idf-sp.tripod.com/frameset.htm

[Hell_Hound]

Writing algorithms is my bread and butter. :mrgreen:

I'll come up with a basic stupid set of heuristics and post it here, and we can kick it around and try to think of ways it would fail catastrophically - perhaps we can come up with something good by an evolutionary method.

[Sean]

I should be able to help with the logic too. How do you want to organize your thoughts on this, some kind of logic flowchart? Cause and effect diagram? Is this going to include the logic for the troops?

[Hell_Hound]

My first effort was going to be pseudocode, but a flowchart makes sense too.

Let's hash out some basic principles first - I expect these to get shot full of holes, then we can modify them to account for whatever I've overlooked:
IF target=tank AND range<500m AND targetaspect=front THEN (you're screwed, fire your best weapon or retreat if you have a route)
IF target=tank AND range<500m AND targetaspect=side/rear THEN (fire if kill prob is good, otherwise sit still and pray)
IF target=tank AND 500<range<2000 AND targetaspect=front AND enemydon'toutnumberyou THEN (fire if kill prob is good)
IF target=tank AND 500<range<2000 AND targetaspect=front AND enemyoutnumbersyou THEN (go turret-down, your missile will never get there before he whacks you)
IF target=tank AND 500<range<2000 AND targetaspect=side/rear THEN (fire if kill prob is good, otherwise go turret-down)
IF target=tank AND range>2000 THEN (fire if kill prob is good, otherwise hold fire and observe)

The above is obviously incomplete, and through most of it I was envisaging a Brad in a battle position. There would ideally be different behaviors for in a BP versus out in the open.

Anybody who wants, edit and contribute.

[Ssnake]

Take into account that the vehicle may encounter several threats simultaneously from more than one direction. Also think about the employment of smoke grenades - both in the offense and defense.

[Hell_Hound]

If(numOfTargets>1)
{
death.isCertain=TRUE;

dismountTroops();
popSmoke();
burstIntoFlame();
}

I'll work on it. :shock:

[NEpi]

i think they should fire anyway if they think they can kill, unless otherwise ordered.
there's a big problem with hull-down positions for the IFVs. they're standing way too high. first, let's concentrate at positioning themselves in the lowest possible height, which in brads with TOWs means DAMN LOW! (i wouldn't want to try hitting one in good BP).

ok, let's start fresh:
what are the parameters that will set the IFV's behavior?
1. threats
2. ranges
3. armament available
4. terrain
5. mission

anything else i'm missing here?
now what are the priorities?

[Lupson]

This is interesting, I'll make a try at providing a generic model for unit behaviour. It'll probably will be total rubbish, but I'll try anyway. :)

I want a generic behavioural algorithm which should be applicable to any unit:

For any individual vechicle or squad - i.e. the lowest unit with a composite fire- and damage "score":

1. Get unit orders - The behaviour of any unit should be based upon it's current orders. A unit on recon wouldn't behave in the same way as a unit with engage orders. However, certain situations would of course supercede a particular behaviour.A "Kill och be killed" situation for instance.

2. Evaluate if any "danger" and/or "fire opportunity" exist -
first, evaluate wether there is any danger or fire opportunties at all. If there aren't any - just proceed with current orders.

3. Determine degree of "danger" and/or "fire opportunity" - Just compute a numeric value based upon a formula for each possible "danger" and "fire opportunity". These values should be based upon common sense things such as:

Offense: range, available arms, aspect angle a.s.o.

Defense: knowledge about enemy threat (enemy arms, aspect angle, range) and own armour. Fog of war and the element of surprise makes this even more complex.

Now I need to put some more thinking into this. The numerical values should provide us with a base of determining wether the unit should partake in offensive or defensive action. For example: If the difference between fire oppurtunity and danger is very large AND master orders are "Fire at will", offensive action should be taken IF the fire opportunity is large enough for the unit to have a credible chance of inflicting sufficient damage to the target.

Or, if we have a high fire opportunity number - meaning we certainly can kill a target - but at the same time, we can be killed just as easily, the unit could do some kind of engage and evade AI pattern.

The idea of these numerical value based rules is that they can determine a master behaviour for a generic unit in any situation which then could be refined in, for example, a lookup table. Though the idea is that one shouldn't need lookups since the computations in part 3 should provide us with values accurate enough.

Auhhmmm...It feels as this is totally off mark, I need to test it with pen and paper...I'll get back if I come up with anything remotely useful :)

[Hell_Hound]

In constructing mine, I was trying to do some cost-benefit analysis...an APC has to think about the consequences of firing in terms of what will happen if his launch is detected.

At close range, a missile can kill a tank if fired at the side or rear - but you're probably only getting one chance. If a kill isn't likely you're better off staying put and hoping the tank doesn't notice you. (If he's facing straight at you, you have less than 7 seconds to live, so you might as well throw your best shot at him.)

At medium range, you're still very hittable - so you fire only if the target isn't looking at you, or if you're part of a group and your massed fire is likely to be effective.

At long range, if you have cover, you might as well shoot if you have any chance of hurting the target. The target has to make a difficult shot in a hurry, probably while maneuvering - you're pretty safe.

[Ssnake]

Keep in mind, you need to formulate it as if you were talking to a brain dead zombie, or formulating a contract with the devil. Computers have no common sense, and they do as you TELL them, not as your WANT them to act.

[Olympian]

The problem I think is a function of too much insight: one can easily develop focus for AI behavior if it is assumed that AI units behave according to a very standard model in a given situation- only a lack of awareness for the nigh infinte possibilities in a decision tree hits the gavel with the hammer and makes definite conclusions about how a unit ought to clearly behave. A person on the other hand who sees an overwhelming array of possibilities must cope with the practical limitations of design, programming time and computer cycles, and so would arbitrarily draw the line somewhere: his decisions will reflect some intuition and educated guessing alternating with doctrine born out of established military principles- and even still he could never prepare for a vast amount of behavior patterns that conceivably would occur. When should crews scan to the left, when should they scan right, to the front and to the rear? The forest on the left might either shield a left flank or it might conceal an anti-tank squad or recoiless rifle. It therefore may be considered a dead zone so to free up crews' attentions to other areas- or it may be a mistake to do so. Should the TC anticipate possible future actions against tanks and order a Sabot round loaded during the next load cycle or stick with HEAT against PCs? Perhaps longer ranges for T-XX tanks might even prefer HEAT over Sabot, or possibly not.

Unless you can prepare for evey possible combination (which seems more or less impossible) of units interacting with one another, a model which simply presumes to explain AFV behavior remains very generic as we have not even specified what types and tokens are represented- for example, 8 Bradleys squared against eight T-72s may behave much more aggressively than 3 BMPs against 4 M1A1 tanks, or 2 Bradleys vs 6 T-80s or 3 T-72s against 12 Leopards. It has been said before that the infantry in SB begin experieincing a decay in morale when taking losses. How might this apply to IFVs, tanks, and PCs? Do we consider AI behavior for each unit seperately, as if operating independently and apart from supporting forces, insulated from victories and losses- or do they act according to the import of witnessing friendly units either taking losses or chewing up enemy units?

At close engagement ranges a single AFV crew facing a tank might:

Fail a morale check and freeze- or open up with everything or only some things, or attempt to slink away if it thinks it has been spotted, or not, or wait for an opportunity to fire that isn't too late nor too soon, or wait and hope another unit might kill the tank or distract its attention, or the crew might just panic and bail, or charge the tank at oblique angles at high speed to throw off the tank gunner's aim, or back the vehicle into a defilade position or woods, if present. And none of these scenarios takes into account what specific vehicles are represented, how much and what ammunition is available, if any, or damage already sustained by all parties- for example, whether or not a BMP-2 should maneuver for an ATGM flank or rear shot on an M1 when the M1 would appear to be immobilized or blinded- orjust harass it with coax fire or with the autocannon, or whether or not a T-72 would act similarly. Nor have I even begun to elucidate the problem of the tactical behavior transgressing into the boundaries of the global behavior and vice-versa- when for example should a unit stop behaving according to order assigned to its route or checkpoint in favor of improvising with built in tactical AI routines better suited to the current events- when precisely should a unit faithfully execute the orders given to it despite the realization that the tactical sitaution would appear to have changed and so that a kind of tactical judgment would at least temporarily put the primary goal on hold in order to respond to what is happening?

I wish everyone luck- and know that any solution is merely one representation of what could happen while ignoring infinitely many others.

[Olympian]

Keep in mind, you need to formulate it as if you were... formulating a contract with the devil.

It is an old story- the demonic pact and the Faustian bargain- the devil always seduces with knowledge; Don Juan fell to the seduction of the last knowledge that eluded him- the knowledge of Hell- an allegory to the notion that widsom makes weary, knowledge a disappointment when one is finally confronted with and realizes the trade-offs inherent in all decisions.

It is interesting to observe what trade-offs people would commit themselves toward when formulating their models, and whether those models will reflect dispositions of their playing styles and even their personalities: aggressive players for example may hold retreating in contempt unless it were to set up for an aggressive attack.

[Ssnake]

You can talk all day, but don't evade my question: What is YOUR suggestion. Be specific. :mrgreen:

[Werewolf]

THIS IS ABSOLUTELY THE BEST AND MOST PROMISING THREAD I'VE SEEN ON THIS FORUM IN - WELL - EVER!

Hope looms immense for SB2 if the community and the guys in it can contribute to unit control logic.

Combine the knowledge of real world tactics that some have with the algorithm writing skills of those who do that sort of thing for a living and WOW!

Ahhhhhh....
The possibilities!

I strongly recommend that this thread be made sticky and stuck up at the top of the forum.

[dejawolf]

engaging a target with M2A2

hmmm..
if several AFV's has been spottet in relatively the same distance, pick the one clostest to the reticle center,
second priority is alignment (front, side back), and third is distance, if it's not more than 900m from the target closest to the reticle..
when a single target has been nailed out,
use 1 secs to consider if threat is a tank or apc, and check what aligment it has to the player.
if only half profile is viewed and if it stands still, use 3 secs to consider if it's alive.
if it is, fire tow, if distance>2000m or if it's a tank.
while the TOW is in the air, the commander should NOT traverse to another target, but scream tank left 1200m for example, and if the tank is going to shoot, order the driver to go hull down.
if the target is too close to give the tow a chance to adjust,
aim low, fire, and quickly adjust up and onto target after 1 sec.
when both missiles are used, move in to turret-down position,
deploy troops in front of M2A2 in a line. 2 ready to fire RPG's(except for the ones needed to upload missile)
while commander scans left 90 degrees slowly, right 90 degrees slowly, 360 degrees medium speed etc. until
the missiles have been uploaded.

ok. that was a single engagement on a flat desert terrain type area where you're located in a good BP. in woods or areas with houses, things get a little more complicated.

and then there's how it should engage inside woods, troop engagement, and general movement behaviour, and actual tactics which i'll let you guys at.

[Hell_Hound]

and then there's how it should engage inside woods, troop engagement, and general movement behaviour, and actual tactics which i'll let you guys at.

Much of which is decided by the player; having some kind of aggressiveness setting for each unit would take a significant burden off the CCL. Of course, we sort of have this already with the Guard/Defend/Hold assignments.

As for engaging in woods - do units have any awareness of how good their cover is? Units buried in trees would of course behave differently from units on an infinite flat plane (or plain :) ) but I assumed the AI couldn't judge that.

I am assuming that the unit knows if it's at a BP or not, and as I refine my rules I'll start to incorporate that.

(I may not get to work at this for the next couple days, though...an officer delegated me a writing assignment last night which is going to devour a bunch of my keyboard time. :x )

[Olympian]

You can talk all day, but don't evade my question: What is YOUR suggestion. Be specific. :mrgreen:

I am much more comfortable with theory than with actually producing results- I have no specific solution that someone like Werewolf or Hell Hound could not also match and probably do even better, short of just telling everyone to play MP with military-minded folks if one desires a more accurate model for behavior. I could not even begin to develop a rule set that always places units into an ideal situation, given the available information to that unit, that also avoids automaton tendencies (witness the occasions where the infantry units run back and forth without purpose as if caught in a loop). Furthermore, I can appreciate the difference between knowing that and knowing how- I can possibly identify a situation whereby a recce vehicle ought to evacuate its immediate position- that is fairly easy. The methodology of executing that order however illustrates the challenge. We operate in real life with sketchy and ambiguous information that is peripheral to our senses; we sense things about our environments that reveal themselves in less than apparent and quantiable terms; we are sometimes privy to a larger perspective that may have some impact on the immediate picture- for example knowledge of the general location of enemy units, so that one doesn't have to scan apparently dead areas. These abstractions might very well be as least as important as any concrete fact when making decisions, but I admit my ability for replicating these things in even a crude way is less than desirable.

That does not mean that I have no useful perspective on the situation in general- I offer advice for would be programmers, if that should be of any consideration, since I have a stake in what comes out as I am exclusively a solitary player and I rely on the computer units to provide me with thrills- thus the reason that I say anything at all. There are of course theoretical assumptions in the design behind unit behavior reflecting the philosophies and psychologies of those who design them- and so I am addressing those theories, much like the theories of how wars are to be fought at military staff colleges are discussed and debated. There will always be critics among the customer base who second guess any design decision and think, "Hmmm, I would have done something differently if I were controlling that vehicle or squad," - it's too hard not to, given the challenge of modelling such complex behavior, and designers would do best to develop a tough skin for that eventual situation. I will also say that the more I offer my own criticism of SB's computer controlled units, the more I also appreciate how difficult this area is and that SB actually does a very decent job given the inherent limitations. I can compare SB very favorably with the drone like behavior of M1TP2 or the seemingly omnisicient perspective of the vehicles in Panzer Commander and Panzer Elite.

Take care all- we'll see you when SB2 debuts. And whether you think too much or hardly at all, you'll run into trouble either way.

[Hell_Hound]

IF nearesttarget=tank AND range<500m AND targetaspect=front THEN (you're screwed, fire your best weapon, dismount troops, pop smoke and retreat if you have a route)

IF nearesttarget=tank AND range<500m AND targetaspect=side/rear THEN (fire if kill prob is good, otherwise sit still and pray)

IF nearesttarget=tank AND 500<range<2000 AND targetaspect=front AND enemydon'toutnumberyourunit THEN (fire if kill prob is good)

IF nearesttarget=tank AND 500<range<2000 AND targetaspect=front AND enemyoutnumbersyourunit THEN (go turret-down, your missile will never get there before he whacks you)

IF nearesttarget=tank AND 500<range<2000 AND targetaspect=side/rear THEN (fire if kill prob is good, otherwise go turret-down)

IF nearesttarget=tank AND range>2000 THEN (fire if kill prob is good, otherwise hold fire and observe)

IF nearesttarget=apc AND range<500m AND (kill prob is good) THEN (dismount troops and open fire)

IF nearesttarget=apc AND range<500m AND (kill prob isn't good) THEN (dismount troops and hold fire)

IF nearesttarget=apc AND range>500m AND thisvehiclehasmissiles THEN (keep troops mounted and fire)

IF nearesttarget=apc AND range>500m AND thisvehicledoesn'thavemissiles AND thisvehicle'stroopshavemissiles THEN (dismount)

IF nearesttarget=apc AND range>500m AND thisvehicledoesn'thavemissiles AND thisvehicle'stroopsdon'thavemissiles THEN (go turret-down and observe)

IF nearesttarget=troops AND range<200m AND enemytroopsthisunitcansee>thisunit'sstrength THEN (pop smoke, open fire, retreat if route available)

IF nearesttarget=troops AND range<200m AND thisvehiclehasthermal AND it'sdaytime THEN (pop smoke, dismount troops, open fire with slugthrower)

IF nearesttarget=troops AND range<200m AND thisvehiclehasthermal AND it'snighttime THEN (dismount troops, open fire)

IF nearesttarget=troops AND range<200m AND thisvehicledoesn'thavethermal AND it'sdaytime THEN (open fire, retreat if route available)

IF nearesttarget=troops AND range<200m AND thisvehicledoesn'thavethermal AND it'snighttime THEN ( *** Ok, mech inf, help me out...your vehicle's blind and there are infantry crawling up your ass. Do you trust your horse's armor and attack? Do you bug out? *** )

IF nearesttarget=troops AND 200m<range<500m THEN (open fire)

IF nearesttarget=troops AND range>500m THEN (stay turret-down and observe)

[Ssnake]

Fine. When do we use the coax, when the Bushmaster (and if so, when the HE-I and not the AP-I rounds; an even more tricky question is when to use HE-I instead of the coax since both are area weapons against soft targets...).

Please specify target prioritization as well. What is more dangerous? The infantry squad nearby, or the tank at 1500m, or the moving APC in your flank?
:mrgreen:

[NEpi]

and there are not operating alone...
still working on something

[Hell_Hound]

Fine. When do we use the coax, when the Bushmaster (and if so, when the HE-I and not the AP-I rounds; an even more tricky question is when to use HE-I instead of the coax since both are area weapons against soft targets...).

Please specify target prioritization as well. What is more dangerous? The infantry squad nearby, or the tank at 1500m, or the moving APC in your flank?
:mrgreen:

Here's where I need serious help from mechanized infantrymen; I know approximately zero about ammo selection. My ranges probably need adjusting too; for instance my infantry range bands assume that the enemy infantry's bazookas become really dangerous at 200m, and that the APC has some kind of slugthrower that's highly effective on soft targets out to around 500m.

I'll work in target priorities in the next iteration. I'd prefer to have one series of conditions which identifies the vehicle's priority of targets, followed by the set I'm working on now which the vehicle uses to decide how to engage the target.

[Ratseal]

Practical engagement ranges for motorized infantry against armor vary according to variables already discussed in sufficient detail. However, based on my experience operationally, the following applies for US doctrine and in practice:

Day:
-If no AT weapons available, eat dirt or break contact and flee. Even if you have AT weapons, usually inf that encounter heavy armor will attempt to break contact long enough to make a call for fire. Against light armor and vehicles then aggressive engagement is more typical
-If you have AT4's or SMAWs, move to best firing position possible without exposing to enemy fire and engage. Max effective range on stationary targets 600m, moving targets 400m
-If you have Carl Gustav and AP ammo, add 100m to stationary targets and keep moving targets the same.
-If you have Javelin FGM-148C or later, then you can engage out to 2km against moving or stationary targets. This also allows the little kneeling crunchie sprite to go prone as soon as the missile leave the tube since it doesn’t need to be guided like the older Dragon missiles (now out of the inventory).

Night:
-Infantry will almost always break contact with vehicles at night and make their way to an improved firing position where combined arms can deal with the threat.
-Cut effective ranges of the unguided weapons in half. The Jav has a 4X thermal sight, and remains effective at the same ranges.
-In general the behavior of infantry/dismounted infantry that have night vision capability is very different than that of units that are limited to regular optics. EO provides much better SA. However, infantry rely very much on what they can hear too, much more so than I think vehicle crewmen do. This makes a big difference in fog and heavy rain, as well as wooded areas, or terrain (canyons, defiles) which dampen/channel the sound signatures of vehicles.

I don’t know how APV/AIFV crew prioritize targets, but I can make the same informed guesses that everyone else can. In general, I would expect crewmen under stress to shoot whatever they happen to have up the spout at the time. Add’ly, I would expect that a crewmen who is directing a weapon will engage targets with the heaviest weapon at his/her disposal, and that a vehicle commander will engage with as many weapons necessary to suppress/destroy the target. In response to Ssnake’s query about when you use coax or pintle mounted 7.62mm vice the 14.5mm/25mm, I would argue that you use both simultaneously. In the vehicles from which I have operated, SOP for targets was to employ all the weapons which would bear, often resulting in a combo of Mk19/M2 and M60E3 being operated simultaneously. This applies regardless of the target type, since even weapons that cannot penetrate tend to suppress enemy action or modify enemy behavior (for example, LAV’s which are taking effective, aimed 7.62mm fire will try to find cover or eliminate the threat – OIF/OEF). Main tank weapons may be operated differently, but I think this thread is focusing on mounted/dismounted infantry.

One of the things that isn’t clear about SB2 is whether or not the time period represented by the simulation is contemporary of if it will remain 1980s/90s time frame. This effects weapon selection a lot. If SB2 includes weapons likely to be deployed in the very near future, then we get include motorized infantry and scout vehicle weapons like this one: http://www.designation-systems.net/dusrm/m-166.html . If we are sticking to the mid 80s time frame, then obviously we need to keep Dragon/TOW and eschew Javelin/LOSAT etc. Likewise the proliferation of thermal and Gen III night vision devices.

[GaryOwen]

Some general Bradley traits to consider (IIRC):

* TOW launcher can only track one target at a time. If the TC overrides, the missile aborts.
* Gunner can scan and can engage with TOW from a turret down position.
* Range of TOW2A = about 3,750 meters.
* HE rounds self-detonate at either 2,300 or about 3,000 meters depending on version.
* HE time of flight makes it difficult to engage moving targets past 1,500-2,000 meters.

I think that FM 23-1 (which is available to unrestricted distribution through the Reimer library) has a section on priority of fires. I can't check it here at work, but will later. Priority of fires for M2/M3 may just be a matter of coding doctrine in the gunnery field manual.

[Hackworth]

I would think the first order of business on priority of fires would be 1. to engage any target engaging you, 2. to engage any target facing you and with range to use it's primary weapon against you, 3. then whatever FM 23-1 gives you :) And it might have some reasoning within it like this before going "tanks withing xxx meters, troops within xxx meters..." and so on.

And I have to place my complete agreement with Werewolf in regards to this thread. This is a great possiblity we have here to help influence the creation of our favorite game. We all have complaints about the way the AI behaves. Part of that is through improper use and understanding of the types of routing and BP orders, but a big part is our dislike of the AI logic at times. For instance, tanks that insist on engaging crunchies of no concequence 2km away while a company of T-80s are engaging you, or a battalion is rolling by your front as you sit in your ambush position. Hell, Im supprised more MP players don't use INF manipulating their influence on the AI in this way. Distract the AI and kick their teeth in with your IFVs or tanks.

Anyway great thread!

[Hell_Hound]

I would think the first order of business on priority of fires would be 1. to engage any target engaging you, 2. to engage any target facing you and with range to use it's primary weapon against you, 3. then whatever FM 23-1 gives you

By "engaging" do you mean "shooting at"? The system I have now is more about taking down the greatest potential threat, so if you have an APC facing you and a tank with its back to you, you go for the tank. That's probably a little more team-player-ish than real humans would be.

[GaryOwen]

The Reimer Library seems to be having one of its occasional down times. I really should have just saved the FM to disc.

[Hackworth]

What I mean is engage... First order of business should be attacking anything attacking you (or retreating if the threat is too great). My point is that the AI target selection should first try to keep you alive, i.e., not engage crunchies 2km away from you in the trees while a tank 2km away from you is taking shots at you. Then the second priority IMO should be anything that is a potential threat, i.e., a tank with it's weaon facing you, but obviously not firing on you. Did I just repeat myself?

[Olympian]

I would think the first order of business on priority of fires would be 1. to engage any target engaging you, 2. to engage any target facing you and with range to use it's primary weapon against you, 3. then whatever FM 23-1 gives you

By "engaging" do you mean "shooting at"? The system I have now is more about taking down the greatest potential threat, so if you have an APC facing you and a tank with its back to you, you go for the tank. That's probably a little more team-player-ish than real humans would be.


I agree that concepts such as 'attack' and 'engage' are not really well defined- one could interpret them to mean that a hostile unit(s) is converging on one's position with the intention of destroying something that it finds, but actual firing has not started yet for a variety of reasons (maneuvering for a better firing position, a unit has not detected another, and so on-)

A defending unit witnessing the enemy closing in may have not been spotted yet or it is not under fire per se,- but its sector is being attacked generally- that is to say, the enemy appears to be heading in the defender's position; now the unit must decide if it is the case that it is 'under attack'- a unit might miscalculate and fire too early, only to reveal its location. This is probably difficult to ascertain at longer ranges.

[Olympian]

My point is that the AI target selection should first try to keep you alive, i.e., not engage crunchies 2km away from you in the trees while a tank 2km away from you is taking shots at you. Then the second priority IMO should be anything that is a potential threat, i.e., a tank with it's weaon facing you, but obviously not firing on you. Did I just repeat myself?


Infantry can cause quite a bit of trouble- we all know that units might tend to become fixated on them and so their mere presense makes them dangerous even when no other enemy AFVs appear to be present (but could be sneaking around the bend at any moment). Infantry can be difficult to deal with at long range when a computer TC or Gunner insists on directing the the gun at them, or when a TC rigidly slews the gun at troops hidden behind a ridge that a gunner cannot see or engage.


Here is another point of contention- sometimes a unit will stubbornly fire its full compliment of missiles and rockets into a treeline to get at units that are out of reach (but can still be seen) .

Finally, in a player controlled vehicle, it might also be an option for computer crew members to sometimes only call out and mark the presence of relatively less threatening units ("PC, left!" ) rather than actually take control of the gun and target them, thus not leaving the player exposed to more dangerous threats in a fight. The 'T' key sometimes keeps the computer TC at bay, but it often happens that the 20 second window expires at the most inopportune time, and it becomes distracting to always keep in mind to depress the 'T' key every 15 seconds or so (and it takes away a second pair of eyes to scan for threats).

[Ssnake]

You know that pressing "I" will make the TC select another target if he know one, right?

[richard]

Whats the big deal about infantry? Use your tanks to take the objective, then bring up the brads and infantry to hold it while you push on. :roll:

[Hackworth]

I?! I!? People have been complaining about this crunchie-loving AI for well, forever! And it can all be solved with "I"?! Damn it...just when I thought I knew everything. I knew about "T" in order to get control of the gun for 15 seconds, but I didn't know we could cause the AI to select and rotate though the targets. Can you make it select "last tartget"? Like, if your TC selects the crunchies then you make him selct the tank ahead of you, but then you want him to go back to crunchies, do you have to select all the units in between until it comes back to them or can you just hit say "alt+I" to get the TC to back step? Thanks Ssnake!

Oh and Richard, Im talking about the Tanks and IFVs engaging the enemy INF...their fixation with them inregardless of the hazards that may exist around them (like a battalion of T80's).

[Olympian]

You know that pressing "I" will make the TC select another target if he know one, right?


I never really made the connection to that command and the expected results since I never really perceived any consistent results produced- I have found personally the effect is far more immediately apparent with the 'T' command- but of course the 'problem' might just owe to my perception of things.

Sometimes the problem of perspective becomes compounded when I think of the player's role- he is nether any one crew member exclusively, nor the guiding spirit of the entire vehicle itself, but a mix of all things in various amounts- the player TC may spot something of interest, jump to the gunner's position, but then find that the computer TC has not picked up the target that the player had spotted when the player formerly occupied the TC position. There also exists a god's eye perspective that the player has as overall commander of the forces on the map that he can expolit even from the perspective of an individual crew: the few moments after a unit's death allows the player to pan the camera around in the external view in order to possibly pick up the enemy and so to either mark its location on the map or chase it down with his remaining forces, or call in artillery. Or the player can use the radio of a unit that could not have possibly known the location of the enemy (the observing unit has no radio to alert it to HQ for example) to call arty on its position, even though that unit's map shows nothing there (the player knows that there is something there; the other units do not- save for the one unit that has no way of communicating what is present). But I degress...

[Olympian]

Whats the big deal about infantry? Use your tanks to take the objective, then bring up the brads and infantry to hold it while you push on. :roll:

Is this a facetious remark or is it actually supposed to be a penetrating observation? I really wish the :roll: emoticon had never been invented.

[Lone*star49]

Whats the big deal about infantry? Use your tanks to take the objective, then bring up the brads and infantry to hold it while you push on. :roll:

Is this a facetious remark or is it actually supposed to be a penetrating observation? I really wish the :roll: emoticon had never been invented.
...

Have to agree with Olympian on this one Richard..

In all fairness, in the beginning and for a rather long time, I really didn't know what the grunts could do, let alone, if in the right BP, take out a tank or 3..

But, try and use the grunts first, as scouts.. I run mine either along a tree line or up a hill overlooking the area I want access to.. they are your eyes and are low profile targets if set right.

Great having them show you where enemy units are, so you can either artty them or make B-plans to flank them or set Sta-defence up against oncoming forces..

Also, speaking for me only, I try and use the Brads as long distance OW 2 -3500m distance of territory either I want to take or, supect enemy may use, and deploy troops min of 1000m ahead or to the flanks as said above, for scouts/eyes..

Really, they are the tip of the spear in my assults, and like I mentioned, they do on occasion find/by jumping in them, they/you "hear" a enemy unit/units and track it down either with Artty or go for the kill with missle launch..

Now granted, it doesn't go that smooth most of the time, but given 50/50 odds, I'll put my trust/hope in them along with the brads every time, cause the times they do it right, it is quite a show to watch, and it really makes one appreciate them..

LS

[Hell_Hound]

You know that pressing "I" will make the TC select another target if he know one, right?

It never seems to work that way for me. If the TC puts me onto a target, "I" seems to either do nothing (if the gun's approximately on target) or, if the gun is off target, "I" seems to mean "I lost him, put me on that target again".

[JayMan]

HH your right !
That's what it does with me and what I always thought it made for.

[Ssnake]

"I" ("Cannot Identify") technically means "out of range/can't see it", so the TC will present you with the target next in his priority list. That doesn't necessarily mean that you won't get another squad at 3,500m, but chances are he's going to give you the T-80 next.

[BrokenArrow]

"I" ("Cannot Identify") technically means "out of range/can't see it", so the TC will present you with the target next in his priority list. That doesn't necessarily mean that you won't get another squad at 3,500m, but chances are he's going to give you the T-80 next.

:arrow: My kind of odds 50/50 :cvcys:

[richard]

I basicaly use inf the way you suggest...But if i know what to expect in resistance and most scenarios tell you what to expect, I try to apply a liberal amount of ICM and HE to the target area and keep my units right behind it moving forward at engage speed slow. It works. ( I also play the mission several times following the briefing suggestions. Then i do it my way. :twisted: )

[GaryOwen]

I think that FM 23-1 (which is available to unrestricted distribution through the Reimer library) has a section on priority of fires. I can't check it here at work, but will later. Priority of fires for M2/M3 may just be a matter of coding doctrine in the gunnery field manual.

The Reimer library is back up. The FM doesn't have as much as I remembered it having, but this is what it says:

ENGAGEMENT DECISION
The engagement decision process is a series of progressive and interdependent steps (or actions): determining threat levels, selecting weapon or ammunition, and making confirmation. The Bradley commander is solely responsible for making these decisions.

2-4. TARGET DETERMINATION
Targets are categorized by the threat level that they represent. To defeat multiple targets on the battlefield the most-dangerous targets must be engaged first. Commanders determine target threat levels based on threat analysis of the mission area. All soldiers must know the engagement priorities of their unit; However, the BC is responsible for determining the immediate target threat level, based on the threat posture at the time.

a. Target Threat Levels. The threat levels are most dangerous, dangerous and least dangerous.

(1) Most dangerous. When the crew observes an enemy target with armor-defeating capabilities that appears to be preparing to engage them, the target threat level is most dangerous. This type of target is the greatest threat and must be engaged immediately.

(2) Dangerous. When the crew observes an armor-defeating target that is not preparing to engage them, the target threat level is dangerous. This type of target must be engaged after all most-dangerous targets have been destroyed

(3) Least dangerous. When the crew observes a target that does not have an armor-defeating weapon system the target threat level is least-dangerous target. This type of target is engaged after all most-dangerous and dangerous targets have been destroyed, unless it has a high priority of engagement (command and control vehicles).

b. Determining Factors. When multiple targets of the same threat level are encountered, the targets must be prioritized according to the threat they represent. The determining factors used to prioritize these targets are--

(1) Engage close-range targets before engaging long-range targets.

(2) Engage stationary targets before engaging moving targets.

(3) Engage frontal targets before engaging flank or rear targets.

2-5. WEAPON/AMMUNITION SELECTION
Weapon/ammunition selection is the logical selection of a weapon or ammunition that is appropriate for a given target, and it is a direct result of the acquisition process. Probability of hit and kill, as determined by threat type and range, determines which weapon and or ammunition to engage the enemy. (Table 2-2).

[NEpi]

good stuff, owen!

scouting and assaulting is 2 different missions, as for infantry assault and combined arms operation.
the inf. and IFVs should be capable of dealing with all of them, but on the other hand some of it must be the work of the player. the AI shouldn't do anything.

[colin]

I was watching some M113 last night fighting in a hull down position, they did quite well and manouvered to meet flanking threats and did not needlessly expose themselves, as opposed to the BMP1's which dropped off the troops and proceeded forward to hunt for targets. I think the weapon system dictates the performance a lot, the gun/missile on the BMP caused the TC to look for target to engage. the M113 .50 was more defensive and was used only to engage close in targets

[dejawolf]

seems like the crunchies currently deployed in steel beasts is the combined arms troops. terrorist troops would act a lot different, i'd believe.
and iraqi troops, which have no turning back.

[Ssnake]

Steel Beasts' behavior routines were develop to reflect high intensity combined arms combat tactics. We will need to create different sets for other purposes, that's for sure.

[Hell_Hound]

Thanks, GO - that's good stuff.

I haven't forgotten about this thread, I've just been waiting for the doctrine info (15%) and too damn busy with other things (85%). I'll try to fold your contributions into a new set of rules early this week.

Ssnake - I feel like what I've done so far is really primitive, and I'm not sure how much more refinement it requires. Would it be possible for you to post SB's existing set of algorithms, or a summary of them, so I can see what the standard is like? (Even cutting and pasting the code would be fine, if it resembles C++ or Java in any way.)

[Ssnake]

Sorry, but we can't make this kind of info public. Actually, we consider this the core of really unique code design. All other stuff is a matter of intense research and hard work. True creativity is in the behavior algorithms, and the semi-automation is what sets SBx apart from other (professional) training tools...

[GaryOwen]

With regard to the Bradley, I don't think that it needs to be too complex. In fact the more complex the CCL gets, the more difficult it is to predict what the AI units will do and thus the more difficult it is to play the game. If the player/scenario designer wants to makes things complicated, he can always use the FIRE IF conditions. The general direction I'd take:

If unit is at BP then it stays turret down and scans unless it makes contact. Once targets are sighted it goes through target classification, prioritization, and ammunition selection routine. It only goes up to hull down to engage 25mm and coax targets.

If unit is not in BP and makes contact, it seeks turret down position. If unit is in contact with "Most Dangerous" targets, it lays suppressive fire with coax or 25mm while moving to turret down position. Once in turret down position, unit does quick scan and then goes through target classification, prioritization, and ammunition selection routine.

[Ssnake]

ah... turret down (or hull down) - with respect to what area?

[NEpi]

haven't forgot about this thread either, just had my comp die on me, and it's the holiday season, so no office computer for me either.

i think in regard to a portion of the seen terrain (hull-down can be 1/3-sky-2/3 land, maybe? turret-down=100% land from gunner and the TD percent for the TC? it's a beginning. then get higher a little and scan again)

i thought of another way to prioritize targets: "Threat Points". each sighted target is given a threat rating that is set by range, whether it's firing back, position, and. of course, the nature of the threat. an equation of some sort (still in the process) resolves the threat degree of that target in a more "streamlined" manner. then priority is measured by how more thretenning a target is compared to other targets.

now all i have to do is numerical analysis, a discipline i don't like that much. anyone wants to help?
i also need some way to describe "equithreatening" situations: in what conditions a tank is exactly as threatening as an inf. squad? (all the above parameters).

what do you think about the whole idea as a whole?

[GaryOwen]

ah... turret down (or hull down) - with respect to what area?

Obviously, for a stationary position, the orientation will have been determined by the player or scenario designer. For a unit that makes contact away from a BP, it should seek a turret down position relative to the area of the greatest known threat (do the classification routine). Once in position a quick scan to ID any other threats should be done. Appropriate adjustments to position should then be made.

[Ssnake]

...but what if the vehicle has not yet seen any enemy?
Or if it can see two targets with identical threat levels, but from different directions?

Remember: Not just any common sense bla bla - precise and specific step-for-step prescriptions to a brain-dead zombie who will do what you say, and not what you want.

[GaryOwen]

...but what if the vehicle has not yet seen any enemy?
Sorry, I mistakenly thought that we were only talking about actions on contact. If the unit is not in contact, then it simply follows its orders and continues on its route or occupies its BP and scans.

Or if it can see two targets with identical threat levels, but from different directions?

The FM 23-1 target classification heuristics are really nice because they seem specific while in fact they are actually rather vague. A human commander would have to decide whether to first engage a stationary dismounted ATGM team set up at 800 meters or the T-72 moving straight at him at a range of 400 meters. The FM provides no guidance as to whether a stationary far target is more of a threat than a moving near target. So I agree that's some fuzzy stuff. And we all know that every good armor crewman/cavalryman excels at fuzzy thinking. Personally though, regarding your question, I'd engage the target I thought I had the best chance of killing or suppressing. And yeah, anticipating your next question -- what if there are two targets of the same threat level and the same chance of being successfully engaged ---, then I'd flip the coin. I do however think that the FM 23-1 guidelines could provide a sturdy foundation for an engagement algorithm. (And be a good marketing point -- "computer controlled unit behavior is programmed to act in accordance with official army manuals.") As far as writing the code, I barely passed my C+ programming class 15 years ago.

Remember: Not just any common sense bla bla - precise and specific step-for-step prescriptions to a brain-dead zombie who will do what you say, and not what you want.
I woulda been really happy with a squad of zombies sometimes back when I was still trying to play NCO. And while I'm on the subject of zombies, the crunchie behavior is where most of the work is, IMHO, really needed.

[Ssnake]

I don't need code from you, just a step-by-step description of the thinking process of a Bradley commander. That alone is a challenge as most of you probably already realized. So far there has been exactly one attempt that at least came close to how the final result should look like. Brain dead zombies do not excel at "fuzzy thinking", they excel at not thinking at all with amazing speed, and using every loophole in your instructions to go on a silly and uncontrollable rampage. Fortunately it's virtual zombies on a virtual rampage, but still they can make YOU suffer if we set them free. ;)

This thread is an exercise in structured thinking. Only if you learn to think in algorithms you can hope to give us advice how to do things better. We certainly appreciate serious attempts to help us, but it should be obvious by now that it really is serious and time consuming work.

My primary intent in this thread was just to make you realize what exactly the challenges of developing behavioral algorithms are. I still don't believe that we will come up with a useful solution in the end (but I'd be delighted to see me disproved!). If you want to go on - fine. But keep in mind the two maxims: Giving orders to brain dead zombies, and as careful as if you were negotiating with Satan himself with your soul at stake. Then, and only then you can hope to come up with a solution of practical value.

[NEpi]

This thread is an exercise in structured thinking. Only if you learn to think in algorithms you can hope to give us advice how to do things better.

ok so let's get organized, and please, ssnake, don't insert more "yeah, but it doesn't answer this". that line of reasoning will never get us anywhere. i know how it works. i've done that too in the past.

first, let's talk about movement:
how should IFVs move
how should dismounts move
interaction between dismounts and their IFV (which is sometimes really idiotic)

then we could proceed to BPs, contact, engagement, retreat, and anything else i can't think of right now.

after we will have a good solution for each of the above, we can proceed to stitching them all together. then, we could come back to what we've achieved, and give it another look.

agreed?

i'll start conversing on movement:
is there any reason to break formation while moving? AI is great at keeping formation, but it might be overzealous about doing so sometime.

p.s. ssnake, i also defied the algorithmic approach presented here as being too rigid. there are other ways, like the one i described (i know it's still an algorythm, but it doesn't include an near-endless list of "if"s)

[GaryOwen]

Okay Ssnake,

Zombies and virtual soldiers do not excel with Auftragstaktik. Agreed, we all know that.

And since 1950, philosophers, psychologists, & engineers/programmers, et al., have been arguing about whether a machine that can trick an interrogator into thinking it is a person can be said to be intelligent. (the Turing test) So as far as programming the CCL being serious and time consuming, yeah, some of the best minds in the world have been working on AI for over 60 years. Maybe after SB2 or SB3, the brass ring won't be artificial intelligence but artificial cunning.

As far as my comments that complexity is not necessarily a good thing; I'm someone who defines eloquence as a measure of the ability to communicate more information, more clearly, using less words/symbols. Eloquence is rare but extremely powerful. I did not mean, by my use of the word 'complex', that I thought programming realistic behavior is simple, only that the CCL should make playing the game easier.

As far as this thread not having any value unless it produces an immediate set of practical algorithms, I disagree. There is room for improvement in the APC/IFV CCL. Brainstorming on which areas should be improved does not need to be an exercise in structured thinking. Coming up with a description of believable behavior can be a free form exercise. I, for one, am happy at this point to debate with peers what that behavior should be like. We might actually come up with something good, or at least identify areas with the current CCL that need work. Once such a description has been fleshed out, then we can work on the algorithms that can drive that behavior. It's best to identify a problem as completely as possible before you start the serious work of tackling it.

Finally, I have a dream. A dream that one day I'll watch a virtual Bradley pull up just short of the military crest of some good overwatch terrain. Two scouts will automatically dismount, run, and then low crawl to the top of the crest. They will scan and identify an armored recon patrol moving in the low ground. The Bradley and the rest of its platoon will automatically move to turret down positions and fire a volley of missiles at the tanks before moving to hull down positions finishing off the APCs with their guns. The Bradley platoon will then continue along its plotted route past the flaming hulks. In my dream, during the AAR I award Lt. DeFault a bronze star for his bravery and initiative in decisively engaging the enemy.

[Ssnake]

When writing about "a solution of practical value", this was from a strictly egocentric view. Of course you may gain insights for you personally from this thread, even if it doesn't produce an algorithm.

The point is however that for us the brainstorming phase is long over, and we do have a clear understanding how the virtual units should be have. The real challenge is to formulate the idea in a way that can be easily translated into code. You all have made contributions to "The List", and it is a mile long. Even if it were two or two or ten miles long, it would not improve the quality of the code. It is therefore that I say that ideas are cheap, but turning ideas into reality is what is precious. If you want to improve the code, we need a hard algorithm. And I totally agree that a short and sweet - an eloquent - algorithm is better. As long as it is functional, and complete. It therefore is better to have a complicated algorithm that is complete than one that is short and easy to understand and follow that doesn't cover every situation that must be covered.

[Ratseal]

I suspect (I hope) that SB2 is far enough along that changing the behavior algorythm code is out of the question. In order to accomplish a XMAS 03 release date, the code lock down on major components is probably underway and spiral integration testing is being performed. Since eSim is doing this on a relative shoestring, each phase will take longer. Changing code at this point would create cascading integration and test challenges which complicate timely release of the long awaited GOLD MASTER!!!!! Come on eSim, you can do it!!!

[Hell_Hound]

In the next iteration, nearesttarget will be replaced with prioritytarget, the determination of which will require a separate series of tests. My eventual goal is for any machine of 800MHz or slower to lock up when the AI spots a target. :roll:


IF nearesttarget=tank AND range<500m AND targetaspect=front THEN (you're screwed, fire your best weapon, dismount troops, pop smoke and retreat if you have a route)

IF nearesttarget=tank AND range<500m AND targetaspect=side/rear THEN (fire if kill prob is good, otherwise sit still and pray)

IF nearesttarget=tank AND 500<range<2000 AND targetaspect=front AND enemydon'toutnumberyourunit THEN (fire if kill prob is good)

IF nearesttarget=tank AND 500<range<2000 AND targetaspect=front AND enemyoutnumbersyourunit THEN (go turret-down, your missile will never get there before he whacks you)

IF nearesttarget=tank AND 500<range<2000 AND targetaspect=side/rear THEN (fire if kill prob is good, otherwise go turret-down)

IF nearesttarget=tank AND range>2000 THEN (fire if kill prob is good, otherwise hold fire and observe)

IF nearesttarget=apc AND range<500m AND (kill prob is good) THEN (dismount troops and open fire)

IF nearesttarget=apc AND range<500m AND (kill prob isn't good) THEN (dismount troops and hold fire)

IF nearesttarget=apc AND range>500m AND thisvehiclehasmissiles THEN (keep troops mounted and fire)

IF nearesttarget=apc AND range>500m AND thisvehicledoesn'thavemissiles AND thisvehicle'stroopshavemissiles THEN (dismount)

IF nearesttarget=apc AND range>500m AND thisvehicledoesn'thavemissiles AND thisvehicle'stroopsdon'thavemissiles THEN (go turret-down and observe)

IF nearesttarget=troops AND range<200m AND enemytroopsthisunitcansee>thisunit'sstrength THEN (pop smoke, open fire, retreat if route available)

IF nearesttarget=troops AND range<200m AND thisvehiclehasthermal AND it'sdaytime THEN (pop smoke, dismount troops, open fire with slugthrower)

IF nearesttarget=troops AND range<200m AND thisvehiclehasthermal AND it'snighttime THEN (dismount troops, open fire)

IF nearesttarget=troops AND range<200m AND thisvehicledoesn'thavethermal AND it'sdaytime THEN (open fire, retreat if route available)

IF nearesttarget=troops AND range<200m AND thisvehicledoesn'thavethermal AND it'snighttime THEN ( *** Ok, mech inf, help me out...your vehicle's blind and there are infantry crawling up your ass. Do you trust your horse's armor and attack? Do you bug out? *** )

IF nearesttarget=troops AND 200m<range<500m THEN (open fire)

IF nearesttarget=troops AND range>500m THEN (stay turret-down and observe)

I'll edit this post as I refine the algorithms. Ssnake - can you say anything about how a vehicle senses when it's hull-down or turret-down? Is it something to do with % of the view which is sky, or shooting out lines of sight in predetermined directions and counting how many of them extend more than a certain distance, or what at what elevation the gun must be to clear the nearest terrain feature to the front?

If you can't share, can I use a test like "IF this vehicle is hull-down with respect to target X"?

[delaney]

It's very quick (takes few CPU cycles) to test whether a vehicle is hull/turret down with respect to it's target. (The target is the object that the TC has decided to fire at, or the object that the human gunner actually IS firing at.) The reason for this is that these calcs are already done for the target as part of the crew AI, and this information is stored, and so can be easily retrieved. It takes a lot longer to test whether a vehicle is hull/turret down with respect to an arbitrary point, since it would require new calculations that are not currently done.

Keep in mind that any algorithm has to be able to run without slowing down the frame rate too much. LOS calculations are costly, and the cost increases with the square of the number of units. (Assuming roughly equal numbers of enemy and freiendly units.) Hull/turret-down calcs are similar to LOS calcs, so they're costly too, and I try to use them very carefully.

One interesting thing that I noticed is that your presciption is that for troops > 500 meters, you stay turret down and observe. Sounds reasonable, but I get the feeling that people are complaining now about the AI TC being fixated on distant troops. This would, of course, lead to a similar fixation. Any suggestion about how to improve this? Is it something that really CAN be improved? (I mean, what would you do in real life? If you ignore the troops and go your merry way, you could present a nice flank target to some troops who have ATGMs. Not a pleasant thought!)

Al

[Hell_Hound]

In the next iteration (version 0.00000003, if you will ;) ) that will get cleared up. My intent is for "If the most threatening target you're aware of is troops > 500 meters", then get down and observe. The instant that observing APC notices a tank, it will probably become the new "most threatening target" regardless of its range, and the APC's behavior will change as it reacts to the tank threat.

The use of 500 assumes that the APC's slugthrower won't be super-effective on troops outside that range; it could probably be stretched to 800m or so but I wanted to be conservative. Likewise, the 200m number assumes that the troops' bazookas aren't much of a threat beyond 200m. Again, highly debatable but I want to get the qualitative details right before I start fine-tuning.

What Nepi and I are discussing is some system which assigns a numeric "threat level" to each sighted target, and then selects the highest-rated threat as the one the vehicle must react to next. (Hopefully we can lay out the responses so that dealing with the primary threat doesn't make the vehicle horribly vulnerable with respect to one of the lesser threats.)

[delaney]

Well, SB already sorts all detected targets by threat level. (The exact formula for threat level is probably different than yours, of course.) It also discounts those threats that it can't do much damage to. Troops, however, are a particular nasty case because they *can* be an armor threat at a great distance (if they have ATGMs) but they are very hard for a tank to kill at a far distance. This results in the "fixation" that the AI TC has for troops. Since this seems to be the biggest complaint, I wonder if anybody has any ideas about how a TC or Plt Ldr in real life would handle a situation with troops at 2000 meters. Lob HE rounds at the troops for a while? For how long? Sure, call in arty, but should the unit stay put until the troops are disposed of? It seems crazy to hold up an entire company because of a few troops with ATGMS, but do you risk getting hit in the flank if you move on? Do you move close to the troops to get within MG range (which is also probably RPG range) ?

Al

[Hell_Hound]

To correct the fixation problem, my preference would be for the AI to assume that infantry don't have ATGMs, and make it so infantry outside about 500m have a lower threat level than any other unit.

Can the threat level of a unit be adjusted because of its observed behavior? If so, it'd be fantastic - infantry units could be classed as "mostly harmless" until they're seen to fire a missile, whereupon they get upgraded to "actually rather dangerous" and start getting lots of attention from opposing tanks.

[Chaplain]

Let me put my 2 cents in:

If stationary troops in SB were properly concealed, as true infantry would be, tanks would not see them at 2,000 meters, unless they fired an ATGM. It should only be when troops move in a conspicious way or fire missile weapons/rockets that tanks focus on them.

[MatsW]

Most combat situations are (meant to be) solved on company level with fire and manoeuvre. In a case with a ATGM at 2000 m, most APC/IFV can´t fire with effect. In a company, the MechPlt that discovers the ATGM (how do you know that is ATGM unless you already been engaged?) would probably stay turret down and call for CO/artillery and the company commander have to manoeuvre the rest closer into the enemy. Everything depending on the mission, purpose of the mission and overall situation of course. It´s very hard to know what´s hitting you and were from. A small arms fire could be followed up by ATGM or RPG.

A single platoon fronting this ATGM? enemy would probably have to do the same - halt, turret down, observe, call for artillery and then manoeuvre closer to the enemy or just choose another route.

The big problem would be if the enemy infantry do the right thing and by surprise engage the MechPlt in the side in a situation were the platoon can´t finding any turret down positions. In this case the platoon return fire, pop smoke and stop and starts to fight or just puts in a higher gear to get out of that -backwards or forwards. In a company situation you could probably solve also this by manoeuvre but this perhaps don´t helps the platoon that got the first hits.

The solution for SB? I´m not so sure that a real life Platoon Leader have the possibility to react on a spectrum of different types of enemy fire or ranges. There must be one or two standard behaviours regardless to type of enemy just to survive and then be able to decide what to do next. SB have already good opportunities in the “Route options” but perhaps they can be improved and adjusted to MechPlt.

One way could be that the unit in a combat situation “search” for other friendly units within 1000 m (company). If there is a company (minimum of 1-2 platoons), he would return fire, pop smoke, look for turret down within 200-400 m backwards (stay out of enemy artillery) and then stop and dismount the infantry to be able to observe and engage and support the next step for the company.

If he his all alone - he would return fire, pop smoke and retreat backwards along the route or into nearest hidden position so far that is needed. Then try to find the route he had order to go.

After this (both scenarios) it would be for the player to act as company commander and decide next step.

Another solution could be to add another option on the “Route option” (Engage, Assault, March, Retreat, Scout) that describes the behaviour of “MechPlt Engage” that is not so “Tank platoon”. If there is a company situation (searching for plts in 1000 m radius) the MechPlt should act as the tank platoons so it will not be left all alone taking cover.

Well, this is some thoughts about it.

/Mats

[Ssnake]

I think the key in the question is suppression. In Steel Beasts suppression only works after a squad has a few men down. What we need instead is suppression that can start even by sight of AFVs (if, for example, there is a whole number of them coming at you at high speed, all guns blazing). It should happen in many cases if the squad finds itself under fire (still, there should be no guarantee that the unit would indeed take cover and not do anything else, and there should be a chance to recover from suppression).

Once that it is possible to suppress dismounted infantry you can resort to bounding overwatch techniques - one section observes and pins down the infantry, the other section closes the distance and moves to a more favorable battle position.

Finally we may need to take into consideration the intelligence picture. If the S2 reports that the enemy is equipped with Milan ATGM, then we know that 2.5km distance are pretty safe (save for incoming artillery). If the enemy dismounts have ATGMs of 4 kilometers, this will have to affect our considerations. If it is a slow and comparatively inaccurate ATGM (say, a Sagger with 100m/s and MCLOS guidance) the AFV crew in question might even be willing to take a risk in the hope of being able to evade the missile, or suppress its operator before the missile strikes home.

[MatsW]

Agree about suppression.

If you also add "Bounding-Overwatch" (with AI commanded 2+2 or 2+1 move/observe or fire) to the route options, it would certainly add a standard tactics to be used in SB.

/Mats

[CommC]

Here's a set of heuristics for ya, Ssnake: how about fire missiles at spotted AFVs within a time frame equivalent to the typical firing time of the reasonably well trained western armies of the world such as the US army or the German army?

How about fire at AFVs before infantry????

Is that so hard? Is that so complicated????

It is ironic that you, Ssnake, should ask for help with heuristics of IFV behavior when you know full well that you deliberately dummed down IFVs and infantry in SB to make the tank the king of the battlefield.

I have just about lost all patience with you and the SB team.

It is further astonishing to me that there were posts in another thread about the SB team wondering why SB wasn't more of a "commercial success". How can you expect "commercial success" when you have deliberately broken the IFV algorithm such that SB is neither an accurate sim of the modern battlefield, nor is it a fun 1st person shooter because you have insisted on taking out all the elements that might make for enjoyable game play such as a campaign, random mission generator and settable friendly and enemy AI troop quality. You have not even bother to devote programming resources to mission creation, instead relying on us "amateurs".

[GaryOwen]

I wonder if anybody has any ideas about how a TC or Plt Ldr in real life would handle a situation with troops at 2000 meters. Lob HE rounds at the troops for a while? For how long? Sure, call in arty, but should the unit stay put until the troops are disposed of? It seems crazy to hold up an entire company because of a few troops with ATGMS, but do you risk getting hit in the flank if you move on? Do you move close to the troops to get within MG range (which is also probably RPG range) ?

Al
I remember boresighting, zeroing, and taking headshots on silohuette targets at 800 meters with my coax. The ranges of 500 meters to 1000 meters are fine for the coax. For troops in that range, the coax is probably more effective than the 25mm, due to time of flight and round dispersion. I wouldn't really use the 25mm on troops in that range unless they were stationary and in some type of hardened shelter.

With regard to an emplacement of troops at 2000 meters, if I had at least a section, I'd have one vehicle suppress with the 25mm while the other manuevered to within coax range. I'd dismount troops from the vehicle closest the target and have them assault while both vehicles support with suppressive fire. If I had a platoon, depending on METT-T I'd manuever the other section to support the assault or provide security against reinforcements.

[Ssnake]

I don't think I could win you back, but for the record, I'd like to present my view of the story.
Here's a set of heuristics for ya, Ssnake: how about fire missiles at spotted AFVs within a time frame equivalent to the typical firing time of the reasonably well trained western armies of the world such as the US army or the German army?
It is not as if we specify engagement times in our algorithms. We calculate first whether there actually is a line of sight, then assign a chance to actually spot the target (which varies with the direction of the LOS with respect to the vehicle and its primary focus (forward)). Once that the target actually has been spotted, the algorithms kick in to prioritize the targets, and to engage them in order of their priority (if there is a decent chance to actually kill or at least seriously damage the target in question).

It may very well be that the final outcome is not what you would suspect. In which case a more constructive feedback would be helpful. Yours is not.
How about fire at AFVs before infantry????
Is that so hard? Is that so complicated????
It is, because, like Al just pointed out a few posts above, infantry potentially can be a very significant threat. It is OK to request AFVs be given a higher overall priority. Fine. The question however remains what the best possible balance is.
I think the secret is to introduce better suppression algorithms for dismounts to tone down the inhumanly combat effectiveness of current infantry algorithms. If we just upped AFVs on the priority list, all we would observe in our scenarios would be a shift of IFV losses from MBTs and other IFVs to infantry.

I am willing to debate if that is enough and all you want. The easier, the better for us. :) I have my doubts, but I'm open to suggestions.

It is ironic that you, Ssnake, should ask for help with heuristics of IFV behavior when you know full well that you deliberately dummed down IFVs and infantry in SB to make the tank the king of the battlefield.
This statement simply is not true.
We have NEVER dumbed down anything. There has never been a point where we said "Whoa! The game balance is infavorable for tanks, and makes them look bad! Let's dumb down something to make tankers appear like supermen!"
We are guilty of the following crimes: Starting to work on a combined arms simulation in blissful ignorance of the amount of work that was waiting for us. Starting to work without financial backup (because nobody would finance such an endeavour), and being forced to cut costs wherever possible. Working seven years straight with about three holidays in total, four years with sixteen hours a day, seven days a week, with private money at stake and a second job to pay the bills, paying for a minimum amount of artwork largely by offering royalties instead of hard cash Concentrating on tanks first to get them working better than in any other commercially available simulation game before, and then, in the face of a choice to either add infantry or anti tank helicopters, opting for infantry (which probably was the bigger can of worms). We added infantry with decent anti tank capability, and made AFVs respect that AT capacity in behavioral routines, instead of leaving infantry as the cardboard dummies you will find in about any tank simulation game released prior to Steel Beasts.
It that's your charge, yes indeed, we are guilty of all that, and more.It is further astonishing to me that there were posts in another thread about the SB team wondering why SB wasn't more of a "commercial success".
I don't remember such a statement, but even if I made it, it's probably out of context. Overall, we are satisfied with the commercial success of SB1 given the circumstances of its troubled release history.How can you expect "commercial success" when you have deliberately broken the IFV algorithm such that SB is neither an accurate sim of the modern battlefield, nor is it a fun 1st person shooter because you have insisted on taking out all the elements that might make for enjoyable game play such as a campaign, random mission generator and settable friendly and enemy AI troop quality. You have not even bother to devote programming resources to mission creation, instead relying on us "amateurs".
This is really ridiculous. There has never been an attempt to make IFVs look worse than they are, even if that may be the results of our efforts. We did not spend as much time on IFVs as we did on tanks, true, but I seriously question your basic business knowledge or suspect that you simply are badly informed about the circumstances under which we had to start our work, and have spent most of the now 25 man-years in the development of SB and its successor versions.
We have not "taken out" the elements that you may miss in our software, we never found the time or an efficient way to put them in. That's a big difference. We work on those issues that have the highest priority for us, and just because our priorities are not identical to yours it doesn't men they're stupid or chosen to further an evil scheme to deprive you from your rights for the ultimate combat simulation game.

If the majority of gamers were willing to buy three copies per person (or to pay three times as much per copy as they actually do), we might have the cash to expand the developer staff. The truth is, we still are essentially a three man team with subcontractors and a lot of volunteers. Without the volunteers, there would not be this web site. Without the volunteers, there would not be the continued stream of new missions. Without the volunteers, there would not be a Steel Beasts community as we know it (there probably would be none at all). Al and I and Volcano certainly know what we owe this community. And we also know that many of you "amateurs" still are active soldiers in many armies, or have been there for some time, so I think that it is not foolish to ask you for suggestions, but a very smart and clever idea.
Al and I have to consider WAY more in the design of our software than just the behavioral routines of vehicle classes. We need to consider the design principles of thermal imagers to model them properly, we need to consider the mobility properties of ground types and certain vehicles, we need to consider armor and ammunition statistics, hardware and driver questions, marketing and PR issues, write the manual, integrate the work of contributing artists, handle the localization, find army customers for additional funding etc.

Maybe I'm guilty of having to handle too many tasks, and being a workaholic, and not being able to delegate work. The other truth is, I don't have employees to whom I could delegate work, and I don't have the bucks to hire them. As long as this fundamental constellation doesn't change, this is the best with what we can come up. Your move.

[smogover]

Even though my view is that you shouldn't have to answer that sort of post Mr Ssnake, I congratulate you on the way you handled it. Well done fella, & please don't let one angry puppy annoy you too much. I think you guys should be damb proud of what you have achieved.

[dejawolf]

MG ROF has to be increased to make it more effective at longer ranges.

[Bluewings]

Yep !

That 3 man team did beat HANDS DOWN any other Tank sim on the market , including some from BIG names where ressources are not a problem .

I trust so much that 3 man team than I am ready to cash out 2 or 3 copies of SB1 right now and will do the same with SB2 .

And another thing : Steel Beasts IS a community . Why that community would not help the best it can ? In regard to the MONUMENTAL effort the 3 designers did put in SB series , I am asking this community to make a little gesture and send whatever they can afford as a little donation .
Ssnake , I am willing to donate £30 . How can I do that ?

[Ssnake]

Thanks for the offer, but No. I'm not asking for charity donations. We're not short of money for the way we're doing business. Thirty bucks could buy us a couple pizzas, but are by far not enough to hire an employee. We get along, don't worry. :)

The decision whether or not to hire employees is none that would depend on community actions. It would be irresponsible to hire employees from dontations, even if the donations were substantial. Employees must be financed through regular business activity, and that means establishing other sources of income. I'm working on it, and making progress, but it's not something that could be achieved in a few months.

We at eSim Games are confident that the market demand for what we're doing is big enough for sustained business operation. And we wouldn't want to have it any other way than how we're doing it. I am happy as a self-employed entrepreneur, nursing a company from tiny to, well, moderate (hopefully, one day). I do believe in the business model that we have, and I think that it will work. And I want to make it under conditions of a free market: By working profitably and growing organically through acquisition of more and better jobs because we provide our customers with a product of real value - be it entertaining, informative, or for training purposes.

The beauty of the free market system is that if you provide real value, you will get money in return, and if you can handle money in a responsible manner and invest it without slander and serious considerations, you may eventually build a small fortune. That's what I want to do. If we accepted donations, it would be like subsidies, the economic equivalent of drug addiction. You start wasting money, become accustomed to it, and in the end you lose initiative and fighting spirit and become dependent on it. No way.

Thanks for the offer, but I pass. :)

[CommC]

SSnake,
Venture capital is available if you seek it. SB is such a high quality sim, and the market is proven, that you could get capital if you wanted it. Its your choice not to, as you have said, and for that we hold you accountable. My constructive suggestion to you is that you obtain sufficent financing to get the programming resources necessary to make this a commerically successful product.

We all want SB2 to succeed and be the great sim that SB1 is and even better.

By way of a more specific constructive suggestion, what I am saying is that you already have much of the code for the IFV and infantry operations, all you need to do is improve the spotting slightly for IFVs and somewhat shorten the time from spotting to missile launch. Thats all.

Regarding spotting, as you know, the more recent patches to SB deliberately worsened the AI TCs performance in tanks. With the later patches friendly AI TCs were not as adept at spotting targets in the rear and side positions. I further believe, as I have stated, that the IFV model deliberately built in tweaks to make the spotting and firing rate slow. I don't know the full reasons for this, but I do believe that it could be corrected with some simple adjustments to the code and not a huge rewrite. Thats all I am asking for.

I agree fully that the infantry and IFV model could be improved in addition to the suggestion I have given, and I am fully supportive of that effort. I applaud this thread and the effort you are making to improve this for SB2. We are all rooting for you and want to do whatever we can to help make SB2 successful. If I am frustrated and angry, it is because SB is such a great tank sim, its a shame to see that wasted due to some relatively simple extra tweaks that would make it both an accurate sim of the modern battlefield and a fun and commercially successful product. It is very close to being there. Its a shame to see the greatness of the tank sim wasted due to some misguided design decisions, i.e. the limitation is not in the writing of the code, it is the decisions that are made in the design process. If in doubt about a design decision, lets put in both ways with user settings to let each person set it how he likes.

Thanks for your tireless efforts to make such a great sim, and keep up the good work.

[delaney]

Venture capital is available if you seek it.

Two quick points:

1) We tried for over a year to get a publisher with money to pick up SB1. We even hired an agent to do just that. Result: many meetings, but no offers. The market size for hardcore sims (especially ground sims) is so small that every publisher that we met didn't want to risk their own money on it. (Our current publisher was interested in SB only after we started selling it on the web ourselves. They turned us down when SB1 was in development, and we get no money from them for current developement.)

2) *If* we were now able to find a publisher (or venture capital) with a lot of money to fund further development, you might very well be unhappy with the final product. SB would have to be changed quite a bit to appeal to a mass market in order to recoup the investment.

Now, if you know of a venture capitalist or angel that would like to invest in SB's development without trying to change it into a shoot 'em up game, please send him our way! I would love to talk to him (or her).

Al

[CommC]

OK, thats different. If you tried to get funding and couldn't, I can understand that.

But your second point, you don't want funding because you're afraid to lose design control, that is a different issue. I agree with you in that, but the point I am trying to make is that it is possible to have it both ways. If I were an investor, I wouldn't want you to focus on just a hard core sim, either, I would want a return on my money. But you can add elements that make the game more appealing to a wider audience, while keeping the hard core (i.e. highly accurate) parts of the sim.

Just put in user settable switches to select the different modes, i.e.:

1. Keep the current tank sim model, but call it the "advanced" mode. Then write a simplified tank gunnery model and call it the "normal" mode and let the user decide which he uses.

2. Let the user set the quality of the enemy and frienly AI: recruit, regular, veteran, or elite.

3. Keep the scenario editor with advanced features, but include a random mission generator, a campaign, and a variety of single player missions.

4. Let the users select the difficulty settings in the MP assembly area screen, such as are found in many other on-line real time first person games.

If you add these features, you can still have the hard core sim mode, while still getting financial backing of investors, and a good chance of greater commercial success.

Just my two cents.

[Bluewings]

I tend to be a Hardcor-ist , but the idea deserves a bit of thinking . At least .

Even if the point " Let the user set the quality of the enemy and frienly AI: recruit, regular, veteran, or elite. " seems a LOT of work to me ...

Ssnake , My idea WAS if everyone of us would give (let say) £15 (20$) , it would probably be enough money to hire an "artist" for a month or two who could be in charge of the eye candy tricks , or else , giving the 3 of you time to focus on the game logic , or else...

I think the main problem is you ARE a workaholic and you get your buz from it . :lol:
I know .... :yltype:

Cheers , and ....don't work too hard :shock: !!!!!!!!!!!!!!!! :lol:

:cvcys:

[GaryOwen]

I think the key in the question is suppression. In Steel Beasts suppression only works after a squad has a few men down. What we need instead is suppression that can start even by sight of AFVs (if, for example, there is a whole number of them coming at you at high speed, all guns blazing). It should happen in many cases if the squad finds itself under fire (still, there should be no guarantee that the unit would indeed take cover and not do anything else, and there should be a chance to recover from suppression).

Once that it is possible to suppress dismounted infantry you can resort to bounding overwatch techniques - one section observes and pins down the infantry, the other section closes the distance and moves to a more favorable battle position.

A 'suppressive fire' or 'area fire' fire command that could be used to keep an opposing infantry unit's heads down would be a great addition for the APCs. Especially if it could be somehow connected to a 'dismount and assault' command. Something like that could even make the M113s relevant.

[delaney]

Just put in user settable switches to select the different modes...

Good ideas, and we've been wanting to implement most of them (except for the random mission generator, which probably can't be done with SB) for quite some time ;) Of course, everything is a matter of priorities, but we hope to get to them eventually. To be honest with you, though, I doubt that those changes alone would be enough to encourage the kind of investment that would make a big difference in SB's development.

The brutal truth: If you want to get an idea of how many investors (publishers) are willing to invest in tank sims (or sims in general), just take a look around and count all the sims that are currently in development.

BTW, we do have enough money to hire a few contract artists, so we're not so poor that we need to take donations from fans. (Although the offer is greatly appreciated!) However, to make a major change in SB's developement would require funds on the order of hundreds of thousands of dollars. (Keep in mind that the typical game being made today costs about a million dollars to develop, and some, like half life 2, cost much, much more. IIRC, America's Army cost about $7 million to develop.)

Al

[dejawolf]

i would think the "lesser" vehicles (m113 and humvee) would be good for newbs, to get acquainted to the game.
they automatically make the game more easier, since it doesn't take much brain to pull a trigger.
some missions to introduce the newbs to handle troop management, before going onto the tanks,
and there would be no need for any realism switches.

[NEpi]

i want to concentrate on the original issue of that thread:

this thread was dealing with two different problems:
1. using IFVs
1a. as ATGM platforms
1b. as support for infantry assaults
2. using infantry
2a mount/dismount
2b tank killers
2c better coordination with their APCs

anything wlse?

[colin]

I would like to see the IFV show more caution, they take to many risks in order to get a good shot, the infantry is pretty damm good all things considered and I use them a lot. Presently the IFV's have to high of a casuality rate and I generally leave them futher back to avoid losing them. I prefer using tanks and infantry in the woods and tanks with hummers for long dashes, the hummers draw fire from the tanks, suppress infantry and are good at hunting BRDM-AT.

[dejawolf]

this is not about what you want, but how you think it should be done, colin.
tell me, HOW do you think the IFV's will get a smaller casualty rate?
how should they move? attack? defend?
you have to describe every little twitch of their fingers

[CommC]

Now, if you know of a venture capitalist or angel that would like to invest in SB's development without trying to change it into a shoot 'em up game, please send him our way! I would love to talk to him (or her).

Al

Al,
I find it curious that you would express the fear that SB might turn into a "shoot 'em up" game, given that you have already turned it into that yourselves by your (perhaps inadvertent) poor design decisions.

You have made the friendly AI of supporting tanks, IFV and infantry so poor that anything beyond single tank vs single tank battles in MP are essentially a joke. Nothing resembling real world tactics involving platoons of tanks or combined arms battles are possible in SB. This leaves nothing but single tank engagements, which are not only unrealistic, but also turn SB into a first person shooter, devoid of platoon or combined arms tactics. This has turned SB into the "shoot 'em up" game which you have said you want to avoid. (Right now SB MP does a great job of modeling one scenario: lets see what happens when a moderately decent tank crew goes into battle with supporting tanks, IFVs and infantry that are raw recruits, whose reaction times are dulled into oblivion through lack of sleep or other causes)

I remain confused, just what kind of game do you want SB to be?

For SB to be a realistic and successfull product, we need to come to terms with the reality that the full threat environment as well as the full support environment must be modeled. This includes other tanks, IFVs and infantry (this is not to mention air, but that is another topic). The current movement model is fine, all that needs to be adjusted is the time to fire and the spotting of the AI of friendly supporting units. The enemy AI in single player is fine. You already have the code.

[dejawolf]

by shoot-em-up, he means plain point and click, with vastly reduced realism in the gunnery simulation aspect.
remember, steel beasts has the most realistic gunnery model of tanks, ever seen on a PC this far.
there's enough people which have trouble getting used to this model, and therefore, publishers would rather see this model simplified...

[Sean]

I agree with commc in that there should be some "reward" or advantage to keeping your tanks in platoons. Right now, a single tank can easily wipe out a whole AI platoon without them even noticing where the shots came from. If you don't believe me, just ask the 1st canadian :-) Since the AI suffers from big time tunnel vision, theres no advantage to leaving your tanks in platoons since you lose control over firing arcs.

I do hope some adjustments can be made in the future.

[ShermansWar]

How about 2 man inf counters
for MG teams and ATGM teams.Purpose deicated set of algorythyms or eheuristics ot whtever they are makes sense, especially since in a real battlefield your gonna find purpose dedicated ATGM assault teams that would perform far better than what we get in game.a real MG positon is gonna render open ground impasable to inf, and i think a set of rules where in inf starts game in a certain pos and doesnt move it is considered entrenched.how about new AFVs?
like the LAV, the LVTP-7(large amphibious troop carrier), and thehummers with tows.
All these things would be present on a real battlefield, along with inf units larger than 6 men, regular line inf.would like to see a scen where you gotta assault entrenched inf, in numbers, cplt and comp size,a real platooon, say 35-50 men,they could only carry zooks,would have purpose dedicated atgm teams,2 man teams.
How about a Palladin, or onboard artillery, mortars
40MM grenade launchers, etc?

we'd wait
and I'd still like to see those 5 tank platoons
, and 3 tank scout platoons, as used in 54 tankl battallions during period covered by game
Wanna be able to pick unit names, not choose from a list
As far as inf getting confused when facing tanks, they do, in real life

make inf better, yes, but dont make a game where morale, confusion, fear and fog of war are not considered factors in AI unit behavior heurythym thingys


.


.

[JayMan]

I just found that if you creat a route for the crunchies and once they have started moving you set the speed to halt, they will stop and not wander around (at lest 20 times less than at a BP)

[Hell_Hound]

I agree with commc in that there should be some "reward" or advantage to keeping your tanks in platoons. Right now, a single tank can easily wipe out a whole AI platoon without them even noticing where the shots came from. If you don't believe me, just ask the 1st canadian :-) Since the AI suffers from big time tunnel vision, theres no advantage to leaving your tanks in platoons since you lose control over firing arcs.


Wholehearted agreement. IMHO the simplest and most realistic perk would be if, when a single tank in the platoon spots a target, the others become aware of it after a very short delay (2-3 seconds).

One perk that exists, as you may already know, is that when SB determines who has control of a certain area it gives a formed unit (2+ vehicles) greater value than those same vehicles would have if separate.

[dejawolf]

how about a setup where the commander tells you he has identified a target, it's approx distance and orientation, and you press a button that tells him you're not busy, whereupon he swings you onto the target?
could give you time to pull the trigger to fend off that APC you have perfectly leaded up in front of you.
"tank, 300m, left!"
i think it can replace the cannot identify button.

for ammo selection, the same setup could be applied.
for example, an engagement where you got your eyes on an APC, you got sabot loaded, the commander screams load HEAT!
you got a single button which is equivalent to for example "negative, tank next" or "too far away, need sabot" and TC cancels order for sabot.

[Hell_Hound]

All you'd really need to do is have all the TCs in the platoon become aware of the target shortly after one member is. That's even realistic, because (at least in the CF) the commander's actions-on-contact are...

1. Initial contact report
2. Fire order
3. Detailed contact report

So, "One, this is one three, contact, wait, out! Sabot, lasing tank, on! Fire! One, this is one three, contact, grid [gun goes off in the background] three one four, one five niner, T-72s on ridgeline, am engaging..."

would/should come out of the commander's mouth in rapid succession.

The absolute first priority is to make the rest of the team aware of the contact so they can react to it. If they have more urgent contacts, fine - the game already knows how to prioritize those, give or take some adjustments that we're all well aware of.

To implement this, you'd define the location where the spotting tank sees a target. Then, for the other tanks in the platoon, pretend that the player has Shift-Lased on that point, drawing the TC's attention to it. If he looks that way and sees nothing, oh well, sucks to be him. If he looks that way and sees a target less urgent than others he's aware of, fine. If he looks that way and sees a priority target, he acts according to his route/checkpoint logic, or goes into "E" mode if he has nothing better to do.

[NEpi]

only one problem: the whole platoon is going to shoot the same unlucky sod and waste a lot of ammo doing so. it's more complicated than that (or i might have not understood you correctly).

maybe letting the Plt. CO (player?) decide whether all the platoon engages or to order only some vehicles to engage (difficult to achieve)

[dejawolf]

how about a setup where the commander tells you he has identified a target, it's approx distance and orientation, and you press a button that tells him you're not busy, whereupon he swings you onto the target?
could give you time to pull the trigger to fend off that APC you have perfectly leaded up in front of you.
"tank, 300m, left!"
i think it can replace the cannot identify button.

for ammo selection, the same setup could be applied.
for example, an engagement where you got your eyes on an APC, you got sabot loaded, the commander screams load HEAT!
you got a single button which is equivalent to for example "negative, tank next" or "too far away, need sabot" and TC cancels order for sabot.

well, ok, this is more of an interim "quick fix" idea, until the TC can be made more dependable, than a solution.
the solution would be watertight AI.
which Nils has asked for your help.

[Floydii]

With regards to infantry behaviour and that of the '