What ever happened to...

Other talk that doesn't fit elsewhere.
This is for General Discussion, not General chat.
Troman
Trained
Trained
Posts: 424
Joined: 12 Aug 2006, 15:40
Contact:

Re: What ever happened to...

Post by Troman » 12 Sep 2007, 14:29

2_Late wrote: If you want a solution to threat assessment when you can only see one of fifty that are coming? My suggest: You have a intelligence failure, Stop focusing on threat probability of one target and start working on a way to either link up existing sensors into a larger network or adjust existing sensors to detect air threats further out then you can fire at. This would solve the problem of decoying. This way you would know about a threat and it's composition long before you're faced with what target to shoot at first.
You are talking about a different problem and about different level of AI. You talk about intelligence failure of the game AI (higher level, scripted) and what we discussed is unintelligent behavior of unit AI that controls SAMs behavior (on a lower level, hardcoded).

Unit AI doesn't build sensors and game AI doesn't control SAMs.

That put aside consider a situation where you have no sensors built by game AI (or human player) for some reason and you have the probabilities again. When working with unit AI we can't rely on game AI and can't make any assumptions about it, defenses should behave intellegent and do the best out of any situation no matter if they have sensors nearby or not.

However it is a good suggestion fo the game AI, but as it is now even if you build sensors nearby, all SAMs will stil fire at the first enemy VTOL that will come into their weapon range.

BTW I have no plans to work on any of this problems in the near future, although I'd like to see them fixed some day.
Sign Up for Beta-Testing:
?topic=1617.0

Chojun
Regular
Regular
Posts: 518
Joined: 25 Nov 2006, 17:49
Contact:

Re: What ever happened to...

Post by Chojun » 12 Sep 2007, 15:19

2_Late wrote: If you want a solution to threat assessment when you can only see one of fifty that are coming? My suggest: You have a intelligence failure, Stop focusing on threat probability of one target and start working on a way to either link up existing sensors into a larger network or adjust existing sensors to detect air threats further out then you can fire at. This would solve the problem of decoying. This way you would know about a threat and it's composition long before you're faced with what target to shoot at first.
Quote from: Order number 4E6F7721, April 1rst 4:30am, 2101
To: R&D

By order of Air Guard Strategic Command.

You are to develop early detection and analysis of air born threats for SAM, and a means to network existing sensors to coordinate fire batteries for attack or counter attack.

Failing this you are order to create a Manual fire interface for all ground base artillery and SAM facility's...
I'll assign a research facility right away ;)
The best thing to do when your philosophies don't stand up to debate is to lock the thread and claim victory.

Troman
Trained
Trained
Posts: 424
Joined: 12 Aug 2006, 15:40
Contact:

Re: What ever happened to...

Post by Troman » 12 Sep 2007, 15:51

Rman Virgil wrote:
* That abstract was a real tease.
That's what I thought too.
Rman Virgil wrote: * "Lanchester-Type Equations" caught my eye - what the heck was that ?
Wikipedia has a page about it: http://en.wikipedia.org/wiki/Lanchester's_laws
Rman Virgil wrote: * Well researching that opened up some interesting doors of perception for me - also new.

* And found the "Rand Foundation" which DOES offer free .pdfs of their docs and what a collection along these lines it is.

* So I downloaded THIS publication & am learning about these "Lanchester Equations" applied to various combat situations... eye-opening, to say the least.
Those equations are interested indeed, and the trick seems to be to model the efficiency constants as well as possible.

We could have applied many interesting concepts, one thing that's making everything more complicated is fog of war.
Sign Up for Beta-Testing:
?topic=1617.0

User avatar
Rman Virgil
Professional
Professional
Posts: 3812
Joined: 25 Sep 2006, 01:06
Location: USA

Re: What ever happened to...

Post by Rman Virgil » 12 Sep 2007, 21:57

Troman wrote:
That's what I thought too.
Wikipedia has a page about it: http://en.wikipedia.org/wiki/Lanchester's_laws

Those equations are interested indeed, and the trick seems to be to model the efficiency constants as well as possible.

We could have applied many interesting concepts, one thing that's making everything more complicated is fog of war.
* Shucks..... why didn't I check-out Wikipedia last night ... That would have been a better place to begin than going straight into disertations - thanks for bringing it up.

* "Fog of War"..... hmmm, I've thought about it some & I've come to the conclusion that it's implementation in the WZ world is  basicaly a KISS solution, more sledge than scapel.

* IF it can be approached in a more sophisticated way by takeing your lead from RL military tacs then the following is a possibility.

* Like have you ever heard of any military op that doesnot include the very latest in recon & proceeds any impending conflict with the best scouting asset .... and in the year 2100 with the postulated resources and tech in the WZ world .... especially by T-3 don't you think there would be a dedicated, very fast, weaponless, electronic survey air drone that could send back "Snap-shot" intel before being destroyed... then that snap-hot captured  "force strength intel" being fed back to your CB  ?

* I guess what I'm saying is that "Fog of War" is presently only punctured if the Gamer can eyeball the enemy by occuping adjacent territory with an asset or if the enemy's army encrouches your sensor range. And that's OK for T1 & T2 but by T3 it no longer makes sense within the context of all else that becomes available tech-wise. In other words by T3 the "Fog of War" parameters should change and the fact that they don't is more a reflection that WZ had to go to market and start generating revenues.

* However, changeing the "Fog of War" implementation at T3 without damageing balanced gameplay is a great & time-consuming task.

* I have no doubt about the scale of resources needed to make that work well.

* For part of this T-3 altered "Fog of War" would include those unimplemented ECMs which would be the natural counter to a spy-plane drone....

* But oh my gosh, looking at  re-balancing the GPMs, that  alone would be a hellish feat.... which could very well be the reason Pumpkin left that task for the WZ sequel... Not that implementing the ECMs & also creating the necessary postulated Electronic Recon Schema/Tech/ Spy Plane Drone unit (for which it would be the counter) is any minor task in itself.

- RV :)
Last edited by Rman Virgil on 13 Sep 2007, 06:13, edited 1 time in total.
.

Impact = C x (R + E + A + T + E)

Contrast
Reach
Exposure
Articulation
Trust
Echo
.

Troman
Trained
Trained
Posts: 424
Joined: 12 Aug 2006, 15:40
Contact:

Re: What ever happened to...

Post by Troman » 13 Sep 2007, 16:17

You overlooked uplink center, that reveals the entire map. AIV does strive for it. But it is not an ideal solution for both the game and AI. Just revealing the whole map is too simple (ECM could be one of the possible counter measures) and waiting for T2/T3 until AI can get uplink center to be able to see enemy forces and judge about its strength is not acceptable.

There are some approaches that involve probabilities, for example when an enemy force is noticed near some derrick the force size is stored and when it disappears in the for of war the probability for it to exist drops with time, but this is too inaccurate.
And scouting to find out enemy force size is difficult even for the human player, so right now there's not much can be done about it, I guess.
Sign Up for Beta-Testing:
?topic=1617.0

Chojun
Regular
Regular
Posts: 518
Joined: 25 Nov 2006, 17:49
Contact:

Re: What ever happened to...

Post by Chojun » 13 Sep 2007, 20:23

* Like have you ever heard of any military op that doesnot include the very latest in recon & proceeds any impending conflict with the best scouting asset .... and in the year 2100 with the postulated resources and tech in the WZ world .... especially by T-3 don't you think there would be a dedicated, very fast, weaponless, electronic survey air drone that could send back "Snap-shot" intel before being destroyed... then that snap-hot captured  "force strength intel" being fed back to your CB  ?
I've kinda sat and observed the problem being discussed at length.  But I think the fundamental problem with FOW hasn't been discussed yet.

The problem of target selection is much easier if you know all the targets (that you care about at a particular moment).

Say there are 10 targets, 5 of which are recognized and identified (visible).  There are 10 SAM sites within range.  Threat Assessment (TA) selects 5 SAM sites to engage the targets, and TA determines that 1 SAM per target would be sufficient for a kill.  The 5 sites engage and the rest doesn't matter.  The other 5 targets don't matter because they can't be seen and for the SAM sites' intents and purposes, they don't exist.  But if they come into range they'll be prepared for them while the other 5 reload.

The fundamental problem that I referred to earlier is the fact that many of Warzone's weapons have an effective range that is much farther beyond their sight range.  So this is what compounds the AI prediction problem where "there could be more targets" so how many missiles do we fire if we can see only one target?

The simple answer to that question is "enough to kill the target."  If a target somehow escapes undamaged or otherwise intact, hopefully the Commander installed some 2nd layer defenses that will mop up anything that gets through.  Not all of the United States' air defenses are located on its perimeter (but I digress).

The answer to the questions that have been discussed at length is two fold:  First, give the weapons an ability to "see" their targets before they move into range.  This gives the targetting AI a chance to determine and assign targets.  Secondly, give the AI an ability to determine how much ordinance is needed for a takedown and fire only that much ordinance.  There are no worst-case scenarios for targetting, which is why a good Commander will plan for defense failures and install 2nd/3rd layer defenses.

Case-in-point:  The Patriot missile batteries, as less-effective as they were, would only shoot 1 interceptor per battery, not all 4 that they had available...
The best thing to do when your philosophies don't stand up to debate is to lock the thread and claim victory.

Troman
Trained
Trained
Posts: 424
Joined: 12 Aug 2006, 15:40
Contact:

Re: What ever happened to...

Post by Troman » 13 Sep 2007, 21:39

Chojun wrote: The simple answer to that question is "enough to kill the target."
How many are enough to kill the target?
Sign Up for Beta-Testing:
?topic=1617.0

Chojun
Regular
Regular
Posts: 518
Joined: 25 Nov 2006, 17:49
Contact:

Re: What ever happened to...

Post by Chojun » 14 Sep 2007, 02:07

That's a question of balance, but it should probably be 1 or 2 depending on body size.
The best thing to do when your philosophies don't stand up to debate is to lock the thread and claim victory.

User avatar
Rman Virgil
Professional
Professional
Posts: 3812
Joined: 25 Sep 2006, 01:06
Location: USA

Re: What ever happened to...

Post by Rman Virgil » 14 Sep 2007, 03:52

--------------->
* Being so preoccupied with the real danger of breaking WZ gpms by introducing this tech I believe I did a poor job of explaining the tech itself.
* First - this is apart from the Sat UpLink. It would come well before and would be surgical / electronic analytic whereas the Sat UL is blunt / player eyeball.
* Here are some links that do a better job of explaining Airborn Drone Spy Tech:
http://www.pbs.org/wgbh/nova/spiesfly/uavs.html
http://www.livescience.com/technology/s ... 41130.html
http://www.msnbc.msn.com/id/7408216/
* Now let me say that WZ 2100 tech has to deal with 2007 tech in some way, IMHO, otherwise the willing suspension of disbelief is compromised.
* Take for example The Wasp air spy drone which has a 13-inch wingspan and weighs 6 ounces. Instead of getting a bead on enemy naval as it does in RL why not WZ vtols ? And? taking the resultant enemy "Force -strength intel" aka # of enemy vtols data & transmitting from enemy airspace to your AA emplacments (thus creating a different kind of 'Fog of War' than presently exits in WZ). * Then ECMs counter could be to "jam" or "black-out" that data transmission which would place it back full-circle & squarely into the present challenge. ;)
* Funny how that works, heh.
- RV :)
* EDIT: The Military's use of Air Drone Tech is growing & indeed has already become absolutely? central to the tacs & strats of all the armed forces moving foward through the 21st Century. There are many reasons for this which I haven't even mentioned.
* If your interested in the details of RL armed forces commitment to Drone Tech the Wall Street Journal article of Feb. '06 on the famous Lockheed Martin's "Skunk Works" provides a great overview with key insights: HERE
* In conclusion, IMHO, Drone Tech is a veritable Goldmine when it comes to WZ 2100 evolution. :)
--------------
Last edited by Rman Virgil on 14 Sep 2007, 16:40, edited 1 time in total.
.

Impact = C x (R + E + A + T + E)

Contrast
Reach
Exposure
Articulation
Trust
Echo
.

Troman
Trained
Trained
Posts: 424
Joined: 12 Aug 2006, 15:40
Contact:

Re: What ever happened to...

Post by Troman » 14 Sep 2007, 16:41

Chojun wrote: That's a question of balance, but it should probably be 1 or 2 depending on body size.
Chojun, that is the key problem we were discussing all the time:
Troman wrote:The problem is to find an optimal number of SAMs for a VTOL under uncertainty
The only difference is that you think uncertainty doesn't matter.

Not sure what you mean with 'balance' though.
Sign Up for Beta-Testing:
?topic=1617.0

Chojun
Regular
Regular
Posts: 518
Joined: 25 Nov 2006, 17:49
Contact:

Re: What ever happened to...

Post by Chojun » 14 Sep 2007, 20:17

A SAM should fire only as many as it takes to destroy its current target.  Any uncertainty should be taken care of by the Commander as they plan their base defenses.

Since there are so many Linux lovers around here, you should take the UNIX approach.  "Everything should do one thing and do it well."  The only thing that the SAM site should care about is to be assigned a unique target and to fire enough SAMs to take it down given a best-case scenario.  Worst-case scenario and all of the SAM-sites fire their missiles at one target.
The best thing to do when your philosophies don't stand up to debate is to lock the thread and claim victory.

User avatar
Rman Virgil
Professional
Professional
Posts: 3812
Joined: 25 Sep 2006, 01:06
Location: USA

Re: What ever happened to...

Post by Rman Virgil » 14 Sep 2007, 21:35

Chojun wrote: ..........

Worst-case scenario and all of the SAM-sites fire their missiles at one target.
*  A waste of ordinance that can be deliberately triggered in a 2 wave attack wherein the 1st wave is a sacrificial decoy and the timed 2nd wave is your main force which has been given something of a safety buffer simply because your SAMs have all blown their first salvo like blind fools on the decoy (I think Troman is trying to minimize the "blind-fool" dependency).

* There was / is a very clever Nexus Mission towards the end of the last CAM wherein you have to take advantage of this state of affairs (over & over w/cheap Bugs as your 1st wave) in order to advance-prevail to next mission

------------>

* BTW Chojun - This wk has played-out a little dif than I expected when I got back from my SF trip.... I was downtown but only in the early a.m. with the rest of my day scattered between Ogden-Orem-Magna tieing up lose ends before moth-balling our Project next week till Feb '08.....

* However I now know I will definitely be in the office (which is just 5-Min N from downtown) all day Tues-Weds next wk & free to do lunch anytime-where (your choice) just give me a holler & let me know if those days are good for you. My addy is: rman2120 AT fastmail DOT fm

L8r, RV :)
.

Impact = C x (R + E + A + T + E)

Contrast
Reach
Exposure
Articulation
Trust
Echo
.

2_Late
Rookie
Rookie
Posts: 29
Joined: 06 Sep 2007, 23:21

Re: What ever happened to...

Post by 2_Late » 14 Sep 2007, 21:59

May I look stupid again and suggest another approach entirely?

Instead of extending the sensor network, Plan A, extend the arsenal. Don't have the SAM sites pay attention purely to the targets, and other SAM sights. Have it evaluate a target's range and damage capabilities vs if it took the target out directly or let a much faster firing AA gun do the work, taking out the chaff.

The idea being that the target (decoy) has neither the range or damage capabilities take either of the weapons out just yet. So the SAM hold it's fire for any other threats that are just out of sensor range, and lets the AA turrets do it's work. Then if it become absolutely necessary that the SAM open fire it can then, but not before. This way the cheap stuff that would be a waste of the SAM's attention is taken care of and the SAM is still ready to deal with the big guns.

Edit: Before you bring up what the AI might have to do for all of this, may I suggest you make a simple GUI that allows the user to link all the needed firepower as a group and give them a simple slider to decide what conditions the turrets should fire on. This way you've spared your self hours of tweaking, and allowed the users to create much more intelligent networks.
Last edited by 2_Late on 14 Sep 2007, 22:04, edited 1 time in total.

User avatar
Deathguise
Trained
Trained
Posts: 85
Joined: 06 Jul 2007, 20:08
Location: UK

Re: What ever happened to...

Post by Deathguise » 14 Sep 2007, 22:54

Troman wrote:How many are enough to kill the target?
Under current game rules: Anywhere from 1 to 28 missiles for a worst case scenario and that's assuming that all missiles are 100% accurate.

Another thing to think about is the flight speed of the missiles themselves: at it stands the avenger sam has a flight speed of 800 and the vindicator a flight speed of 700, i should imagine most of the widely used vtols also have a flightspeed of 700. If the SAM's had a faster flight speed they might be able to destroy a decoy before its gets in range of all your other SAM's.
Last edited by Deathguise on 14 Sep 2007, 23:09, edited 1 time in total.
"Abandon All Hope" - Chiggy von Richthofen

Troman
Trained
Trained
Posts: 424
Joined: 12 Aug 2006, 15:40
Contact:

Re: What ever happened to...

Post by Troman » 14 Sep 2007, 23:09

Chojun wrote: A SAM should fire only as many as it takes to destroy its current target. 
That's not something C understand though. ;) The question still remains: "how many AA Defenses (SAMs in this case) are necessary to destroy a VTOL". I'm interested in it just out of curiosity. We know probability for SAMs to hit the target, what is problematic, is that this probability is not the same as a probability for a VTOL to be eventually hit, since AA defenses predict VTOL location based on their current speed and direction. If a VTOL will change its speed and/or direction AA defenses will miss.

As for the second line of defenses, consider this situation: a player has built 50 SAMs near each other. Now 10 enemy VTOLs come. Let's say 20 SAMs start targeting the VTOLs, since an algorithm has calculated that 20 SAMs are "enough to kill the targets". Let's say 10 SAMs were lucky and fired at the right predicted location of the VTOLs (since hit chance has probabilistic nature). 5 of the 10 VTOLs targeted by 'lucky' SAMs has changed their speed or flying direction. Ergo only 5 out of 15 VTOLs get taken down, other 15 fly through and wreak havoc in the base. The advantage is that many SAMs haven't wasted their ammo and are ready and patiently waiting for the second wave, but the enemy is stock AI so it won't come.

I think the player will be cursing like a drunk sailor wondering why 30 of 50 costy SAMs didn't even try to move. And he will be right. He can't know that we assume that the player has a second or even a third line of defenses.
Chojun wrote: Since there are so many Linux lovers around here, you should take the UNIX approach.
I have a rough idea of the algorithm. Main thing is it should be effective and it should be able to handle any combination of different AA sites: SAMs, Whirlwind, Hurricane etc. This is probably going to be the hardest part. I'm not working on it though, just taking part in the collaborative musing about a non-trivial problem and I think it was my last contribution, it takes too much time. ;)
Rman Virgil wrote: *  A waste of ordinance that can be deliberately triggered in a 2 wave attack wherein the 1st wave is a sacrificial decoy and the timed 2nd wave is your main force which has been given something of a safety buffer simply because your SAMs have all blown their first salvo like blind fools on the decoy (I think Troman is trying to minimize the "blind-fool" dependency).
I think it is actually the opposite, as I understand it Chojun is trying to do it.
What I think is that there's almost no way for AI to deal with such a trick. Even humans will have a hard time recognizing it.

Our main goal is to minimize number of structures destroyed by the VTOL attack. Therefore we must minimize number of VTOLs that will get through our AA defenses (not their damage, although there is a relation, of course).
Because of this it is better to spend an extra shot(s) at the enemy VTOL to maximize the chance for it to get killed than to fire just enough to take down a VTOL (taking into account SAM's chance to hit), since 50% of undamaged VTOLs getting through is better than all VTOLs with 1% health left dropping all their bombs at the base.

It is actually what Chojun said:
Chojun wrote:"Everything should do one thing and do it well."
What I just described makes SAMs doing one thing: killing VTOLs and they will be doing it well under any situation.

If the enemy uses the trick with two waves, then he has used a smart strategy, (which is not necessary now, since SAMs are not intelligent enough) and has outsmarted the game AI - the strategic AI. This is not a fault of unit AI of the SAMs and this, the strategical part, is what SAMs should not care about IMHO, otherwise they will be doing a second thing, which is actually a game AI's task.

My 2 cents.


EDIT: looks like there are two new posts posted while I was writing my novel, I guess I'm too slow.
Sign Up for Beta-Testing:
?topic=1617.0

Post Reply