allowFleeing default value?

madrussian · June 19, 2006

Greetings all. I'm creating some AI scripts, which need to disable certain AI features, like the following:

_group enableAttack false;
_unit disableAI "AUTOCOMBAT";
_unit disableAI "COVER";
_unit allowFleeing 0;

So far so good. My AI can do it's many amazing thing(s)! However later, I want users to be able to turn off my AI and completely revert the group / unit(s) back to original states.

So I am storing the original states like this:

_group setVariable ["MRU_DAI_RestoreEnableAttack", attackEnabled _group];
_unit setVariable ["MRU_DAI_RestoreAutocombat", _unit checkAIFeature "AUTOCOMBAT"];
_unit setVariable ["MRU_DAI_RestoreCover", _unit checkAIFeature "COVER"];
_unit setVariable ["MRU_DAI_RestoreFleeing", 0.5];

^ The problem is, the 0.5 for allowfleeing is a total guess on my part. So the question is:

Anyone know what the game's default value for fleeing actually is (set-able via allowFleeing) or how to obtain it? I've been wondering about since... oh about the days of OFP.

Description for allowFleeing specifies:

Quote

Sets the cowardice level (the lack of courage or bravery) of a group or unit. The more cowardice a Group or Object has, the sooner it will start fleeing. 0 means maximum courage, while 1 means always fleeing.

But not what the dang default value is! (Hmmm... If no one knows maybe a kind dev could check it out real quick and report back? Then we could update the wiki for all to know.) 😄

pierremgi · May 19, 2014

Try with:

_fleeing = 1 - (_unit skill "courage")

So probably depending on your difficulty settings (levelAI).

Note: the fleeing engine is more complex. A fleeing unit means also changes in speed mode and combat mode. Applying fleeing 0 on a unit doesn't mean you will change this "courage" skill value. But, for your aim, this little formula could be acceptable.

madrussian · June 19, 2006

Hey thanks for the idea! Let's see, I just found this related thread:

If I'm reading this right, looks like somebody tried this very thing (courage subskill) and didn't see a correlation:

On 2/27/2018 at 2:28 PM, AZCoder said:

The OP said he was messing around with courage and did not see a correlation. Sure it seems logical, but it doesn't appear to be related unfortunately.

Not to say I won't test this (courage correlation to fleeing) out myself, but seems fleeing is a pretty difficult thing to test. Would be nice to get a definitive answer from the devs. Anyhow two things would really come in handy:

Knowing default value for allowFleeing
A getter function for allowFleeing

Bit more rationale for anyone interested:

Spoiler

To me, fleeing is such an absolutely important thing to be able to reset back to original state. In my dynamic mission, I normally leave fleeing on, to increase mission dynamics. But now I'm writing my own AI, and certain things (like quick reliable movement as a prelude to forced movement through buildings) need a handful of Ai components turned off, fleeing chief among them. I humbly wish for the capability to turn my own AI completely off, and let 100% vanilla game mechanics take over again (when desired). Basically with my AI, it's going to need to switch back and forth from regular game engine cover and fleeing to quick reliable movement, and do this often. Anyhow, again knowing default value for allowFleeing will be incredibly valuable to my current (rather extreme) AI endeavor. I don't know how difficult it would be, but if an altruistic dev happens upon this thread and can check this out the game code (without too much trouble), that would be hugely supportive. Thanks again! 🙂

.kju · October 20, 2001

the skill is just a multiplier. its basically down to current percentage of health+armor(of transport) of the whole group - pretty sure it remembers also dead units (assigned vehicles?)

as when units die and the group gets small(er), they start to flee.

however more variables may play into it. and pretty sure helicopters may also flee (possibly also planes) when damaged too much

snkman · August 29, 2006

As far as i can tell there is nothing between 0 and 1 its like on /off switch.

madrussian · June 19, 2006

1 hour ago, .kju said:

the skill is just a multiplier. its basically down to current percentage of health+armor(of transport) of the whole group - pretty sure it remembers also dead units (assigned vehicles?)

as when units die and the group gets small(er), they start to flee.

however more variables may play into it. and pretty sure helicopters may also flee (possibly also planes) when damaged too much

Good info, yes I gathered that many factors affected when units break and flee, and assumed allowFleeing value was just one of the inputs. (Probably specifically having a big impact.)

38 minutes ago, snkman said:

As far as i can tell there is nothing between 0 and 1 its like on /off switch.

If this is true, then when I set to 0.5, should be exactly the same as setting to 1. (If I understand you correctly?) In fact, in my dynamic mission, I do set allowFleeing to different values (like 0.3, 0.2, and 0.1) based on the enemy order and thought I could tell a difference. But it wasn't a scientific test or anything. Anyhow, should be fairly easy to definitively prove or disprove.

.kju · October 20, 2001

allowFleeing is a threshold. so how early the AI is to flee - it overwrites the threshold set by skill courage otherwise

madrussian · June 19, 2006

OK, I did some independent investigation. Two very different tests, with somewhat different results.

Goals:

Test kju's hypothosis (see post immediately prior to this one)
Attempt to determine the default value for allowFleeing

Btw - We've managed to reverse engineer the mystical skillFinal formula, so why not this too?

Test #1 - Fleeing resulting from systematic unit deletion only

(This test was actually performed 2nd and yielded more specific results.)

First, two input values are set:

_setCourage (sets every unit's "courage" skill to this value, when nil does nothing)

_setAllowFleeing (sets every unit's allowFleeing to this value, when nil does nothing)

On the VR Map, automated test repeatedly does the following:

1. Creates 16 west groups, each with 10 units, equally spaced around the map in a grid pattern (4 rows of 4 groups per row), each group so far away from each other that they should have absolutely no impact on each other. (160 units are created each round.)

2. Applies the _setCourage value to every unit via:

if (! (isNil "_setCourage")) then {
    _unit setSkill ["courage", _setCourage];
};

3. Applies the _setAllowFleeing value to every unit via:

if (! (isNil "_setAllowFleeing")) then {
    _unit allowFleeing _setAllowFleeing;
};

4. Once all groups are created, the groups each have 1 man deleted, then for 0.2 seconds fleeing units are counted, then 1 more man deleted, then for 0.2 seconds fleeing units are counted, and this process repeats until only 1 man per group remains.

5. The total number of fleeing units is counted and reported. (Note that upon a unit being counted as fleeing, he is flagged and not counted again.) Result is between 0-160 fleeing units.

6. All units and groups are deleted, followed by meticulous (automated) checking to ensure all the test units & groups actually got deleted. (i.e original clean state is achieved).

Observations & Results:

1. When courage is set, the units' courage has no bearing on the number of units recorded fleeing. This is true when allowFleeing is also set, and also true when allowFleeing remains completely untouched. Specifically:

Quote

When _setCourage = 0 and _setAllowFleeing is UNTOUCHED, result is 80 men fleeing.

When _setCourage = 0.1 and _setAllowFleeing is UNTOUCHED, result is 80 men fleeing.

When _setCourage = 0.2 and _setAllowFleeing is UNTOUCHED, result is 80 men fleeing.

When _setCourage = 0.3 and _setAllowFleeing is UNTOUCHED, result is 80 men fleeing.

When _setCourage = 0.4 and _setAllowFleeing is UNTOUCHED, result is 80 men fleeing.

When _setCourage = 0.5 and _setAllowFleeing is UNTOUCHED, result is 80 men fleeing.

When _setCourage = 0.6 and _setAllowFleeing is UNTOUCHED, result is 80 men fleeing.

When _setCourage = 0.7 and _setAllowFleeing is UNTOUCHED, result is 80 men fleeing.

When _setCourage = 0.8 and _setAllowFleeing is UNTOUCHED, result is 80 men fleeing.

When _setCourage = 0.9 and _setAllowFleeing is UNTOUCHED, result is 80 men fleeing.

When _setCourage = 1 and _setAllowFleeing is UNTOUCHED, result is 80 men fleeing.

2. When allowFleeing is set, it has a direct impact on the test. Specifically:

Quote

When _setAllowFleeing is 0, result is 0 men fleeing.

When _setAllowFleeing is 0.1, result is 0 or 16 men fleeing.

When _setAllowFleeing is 0.2, result is 0 or 16 men fleeing.

When _setAllowFleeing is 0.3, result is 0 or 32 men fleeing.

When _setAllowFleeing is 0.4, result is 0 or 48 men fleeing.

When _setAllowFleeing is 0.5, result is 64 men fleeing.

When _setAllowFleeing is 0.6, result is 80 men fleeing.

When _setAllowFleeing is 0.7, result is 112 men fleeing.

When _setAllowFleeing is 0.8, result is 128 men fleeing.

When _setAllowFleeing is 0.9, result is 144 men fleeing.

When _setAllowFleeing is 1, result is 144 men fleeing.

^ Note that for these particular _setAllowFleeing tests, these results are consistent no matter what courage is set to.

^ Note that for _setAllowFleeing 0 through 0.4, I indicate that "0 or X" men were counted as fleeing. Turns out tweaking one small detail of this test produces one result or the other. For _setAllowFleeing 0 through 0.4, when a specific (non-leader) is selected for deletion (see step #4 above in the test steps) like last in the group's units array or first non-leader unit in the group's units array, this (apparently) always results in 0 units fleeing. However, when a random group unit is selected for deletion, this (apparently) always results in the higher number of listed units fleeing. I know, this looks (& sounds) like some straight up quantum-entanglement-double-slit-observer-quirkiness or the like. Importantly though, I assure you these results are very consistent. Also note that for _setAllowFleeing 0.5 through 1, the results (apparently) always exactly match what you see above (thus I didn't mention "or" for these).

3. Based on this test alone, we can conclude that the default value for allowFleeing is 0.6! (because running the test with allowFleeing left untouched results in exactly 80 men fleeing, and also running the test with allowFleeing 0.6 also results in exactly 80 men fleeing ... along with all of this being completely independent of "courage" skill)

... Not so fast though! We have another test.

Test #2 - Fleeing resulting from Actual Battle

(This test was actually performed first, and the results are unfortunately a bit muddier.)

On the Stratis Map, automated test repeatedly does the following:

1. Creates 8 west groups and 8 east group (8 units per group, using "Rifle Squad" composition). So 64 vs 64 men, for a total of 128.

2. Applies the _setCourage value to every unit via:

if (! (isNil "_setCourage")) then {
    _unit setSkill ["courage", _setCourage];
};

3. Applies the _setAllowFleeing value to every unit via:

if (! (isNil "_setAllowFleeing")) then {
    _unit allowFleeing _setAllowFleeing;
};

4. Gives all groups a move orders towards opposing enemy group. (Note at this point every time I increased game speed to 4x manually via the provided key.)

5. The total number of fleeing units is automatically counted. (Note that upon a unit being counted as fleeing, he is flagged and not counted again.) Result is between 0-128 fleeing units.

6. The groups are allowed to battle until 1/4 of the total number (32 men) are killed leaving 96 alive, and at this point the results are automatically recorded. (A 180 sec timeout is provided for cases in which they won't fight, which in these cases all or too many units are fleeing.)

7. All units and groups are deleted, followed by (automated) checking to ensure all the test units & groups actually got deleted. (i.e original clean state is achieved)

I ran this test over 1000 times, trying to establish the correlation between courage and allowFleeing (if any) and to hone in on the elusive default allowFleeing value! Here are the inputs & results:

Spoiler

[Note - You will see gaps in the test numbering. That's where I deliberately reset the numbering]

Observations & Results:

1. Very early in these test runs, I could tell that (in actual battle) both courage and allowFleeing impact a unit's propensity to flee, and that allowFleeing has a far greater impact.

2. I made a spreadsheet with multiple tabs to establish average fleeing values at various levels. It's pretty messy, and I'll spare you the gory details (unless someone really wants to see it). Instead if you simply scroll down through the results spoiler above, you'll see me honing in towards the default allowFleeing value.

3. Early in the testing I was doing 11 battle instances per game run (checking out courage over 0-1 values in 0.1 increments while keeping allowFleeing consistent -or- conversely checking out allowFleeing over 0-1 values in 0.1 increments while keeping courage consistent). However this test is apparently somewhat imperfect, as I could tell equivalent units (assigned the same courage and allowFleeing combos) were fleeing somewhat less over a single game run instance, as the number of battles increased. Once I noticed this, I started testing 50 battles per game run with consistent courage and allowFleeing combos. You'll see this starting with test run #300.

4. Based on this (battle) test alone, (at least when courage is 0.5) the default value of allowFleeing is ~0.525!

So we have competing default allowFleeing results of 0.6 and 0.525... I really wish BIS could give us an allowFleeing alternate syntax that resets unit's allowFleeing to untouched, default state. Pretty please? 😄

Spoiler

In the mean time, I think I'm just going to use allowFleeing default value of 0.525 (as it seems to be a better match for actual battle) and call it a day.

If anyone has any related insight or otherwise, please let's hear it. Thanks!

.kju · October 20, 2001

see: https://community.bistudio.com/wiki/fleeing

1. courage only matters for the group leader

2. setting allowFleeing makes courage irrelevant

3. courage is influence by overall unit skill and cfgSkill settings (aka determine actual value with skillFinal)

snkman · August 29, 2006

@madrussian nice one!

Many thx for the very detailed test and sharing your results! 👍

madrussian · June 19, 2006

13 hours ago, .kju said:

see: https://community.bistudio.com/wiki/fleeing

1. courage only matters for the group leader

2. setting allowFleeing makes courage irrelevant

3. courage is influence by overall unit skill and cfgSkill settings (aka determine actually value with skillFinal)

^ Btw - I'm not disputing any of this (and thanks for weighing in!), just letting you all know my observations. Seems there is perhaps a bit more to the underlying fleeing formula(s), considering my test results.

Also, I should have mentioned I ran those tests with one single mod loaded (CBA), which made it easier to observe everything unfold.

allowFleeing default value?

Recommended Posts

madrussian 347

Share this post

Link to post

Share on other sites

pierremgi 4738

Share this post

Link to post

Share on other sites

madrussian 347

Share this post

Link to post

Share on other sites

.kju 3239

Share this post

Link to post

Share on other sites

snkman 351

Share this post

Link to post

Share on other sites

madrussian 347

Share this post

Link to post

Share on other sites

.kju 3239

Share this post

Link to post

Share on other sites

madrussian 347

Share this post

Link to post

Share on other sites

.kju 3239

Share this post

Link to post

Share on other sites

snkman 351

Share this post

Link to post

Share on other sites

madrussian 347

Share this post

Link to post

Share on other sites

Please sign in to comment

Browse

Activity