What ever happened to...

Other talk that doesn't fit elsewhere.
This is for General Discussion, not General chat.
User avatar
NucNut
Trained
Trained
Posts: 92
Joined: 04 Sep 2007, 12:58
Location: Western Australia, Earth, Sol, Milky Way

What ever happened to...

Post by NucNut » 09 Sep 2007, 11:36

The transports? In the original WZ2100, you could get several different types of cyborg transports, culminating in a campaign-sized transport that shot multiple pulse lasers at the targets it overflys.

Also, is it possible to create a transport that can carry any units (campaign style) or a heavily armed, multiple turret air-battleship? That would be so cool, and it would give the SAM AA a purpose...
REDAC Aerospace: Proving that aircraft and terrain don't mix since 2005 :D
XFire Profile: dogzeroonefox

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

Re: What ever happened to...

Post by Chojun » 09 Sep 2007, 15:31

Those transports, I believe, were in the unofficial NEWST patch 1.11.  They were unbelievably overpowered.  Two or three of those multi-pulse laser transports could take out an enemy base in one pass.  Most of the NEWST mods or patches were quite unbalanced.  But, then again, the game as it is currently is unbalanced :)  But that is a topic for a different thread.

I agree with you about SAM AA though.  I think the solution to SAM is to give them an extreme radar sight range, and a long reload time.  For the Vindicator SAM it would be nice if they could lock-on to 1 target for each missile.  That would be pretty easy to implement in the source.
The best thing to do when your philosophies don't stand up to debate is to lock the thread and claim victory.

themousemaster
Regular
Regular
Posts: 607
Joined: 10 Nov 2006, 16:54

Re: What ever happened to...

Post by themousemaster » 09 Sep 2007, 19:32

Chojun wrote: Those transports, I believe, were in the unofficial NEWST patch 1.11.  They were unbelievably overpowered.  Two or three of those multi-pulse laser transports could take out an enemy base in one pass.  Most of the NEWST mods or patches were quite unbalanced.  But, then again, the game as it is currently is unbalanced :)  But that is a topic for a different thread.

I agree with you about SAM AA though.  I think the solution to SAM is to give them an extreme radar sight range, and a long reload time.  For the Vindicator SAM it would be nice if they could lock-on to 1 target for each missile.  That would be pretty easy to implement in the source.

Perhaps, but heavier bodies can take a full bettery of even the vindicator SAM's and survive.  If they spread their missles, odds are a large air force would decimate a whole base, and all go home at half health.


The only reason SAM's are so... shall we say, guaranteed kills as they are now, is because teching up the SAM (AKA rocket) tree is so much easier than getting VTOLs, upgrading them, AND getting them their HEAP's.


I believe a good solution (other than increased range, which I'll believe), is to make them take longer to develop, so that they are "coming around" right when those heavier body VTOL's are/can be.

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

Re: What ever happened to...

Post by Chojun » 09 Sep 2007, 20:53

Thinking in terms of real-world mechanics, a VTOL should not survive a hit from a SAM.  A heavy body VTOL should probably be able to take 1 hit from a SAM.  So in that sense, SAMs as they currently stand are probably OK aside from some minor tweaking.  But the problem is that it's too easy for a turtling player to just build many SAM sites and make their base impervious to air assault.  Consider:  If Warzone's targeting AI were efficient, and it defenses fired at separate targets based on what would deal the most damage and how much damage it takes to kill a target (as opposed to how it is now where 5 SAM sites will fire their missiles at the first target they see, while the rest of the targets will pass through unobstructed), then a player could feasibly make their base impenetrable to attack.

Long ramble short, I think it's too easy to spam VTOLs and win.  And SAMs, as you said earlier, NucNut, really are quite useless.
The best thing to do when your philosophies don't stand up to debate is to lock the thread and claim victory.

User avatar
NucNut
Trained
Trained
Posts: 92
Joined: 04 Sep 2007, 12:58
Location: Western Australia, Earth, Sol, Milky Way

Re: What ever happened to...

Post by NucNut » 10 Sep 2007, 11:07

That is definitely one thing that should be tweaked in warzone: the targeting AI, and even the AI that controls your units when you're not specifically controlling them. It is so easy to create a fast unit, drain all your opponents ammo, and follow with the heavies to literally NUKE your target.

What would be good would be another page to the unit commands screen, telling your units what they should direct their fire at based on a preferential system of the target's body, propulsion and (maybe) turret type (I'm thinking specifically of the warhead type used).
EG:
1. Heavy Body Vehicles
2. Half-tracked propulsion
etc.

It would consist of several rows of buttons,
[Super Heavy Body] [Heavy Body] [Medium Body] [Light Body]                                  <-- Target's Body Type
[Wheeled] [Tracked] [half-Tracked] [hover] [VTOL] ({Propellor} {ski}  Possibly?)          <-- Targets Propulsion
[Full] [Damaged] [Critical]                                                                                  <-- Targets Health
[Weapon] [Sensor/Truck/Etc]                                                                              <--Turret Type

An in game customisable interface may not be the way to go, but at the LEAST the targetting AI must be able to check that every unit/battery is not shotting at the same target

Do this, get rid of Vista problems, and make it easier to get the nightly builds (multi-turrets) working, and i'd be very happy indeed  ;)

Remember the Archangel - beautiful range, good damage, intelligence of a peanut.

Just quickly, maybe the targetting AI could start out as being dumb (no change from current), but as you research certain (new?) topics, it gets smarter and can pick out high risk or easy to kill targets WITHOUT wasting all it's ammo. Having a 'Targetting Control Centre' structure could really help this idea, loose the building and your defences are stupid again.

Certainly something to think about...
REDAC Aerospace: Proving that aircraft and terrain don't mix since 2005 :D
XFire Profile: dogzeroonefox

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

Re: What ever happened to...

Post by Troman » 10 Sep 2007, 15:04

NucNut wrote: That is definitely one thing that should be tweaked in warzone: the targeting AI
Agreed and it was tweaked already. The units are not blindly targeting a random unit, they choose their target using several criterias: effectiveness of the unit weapon against different propulsion types, distance to the target, damage factor of enemy units, type of the unit (weapon droid, truck, repair, commander etc).
NucNut wrote: What would be good would be another page to the unit commands screen, telling your units what they should direct their fire at based on a preferential system of the target's body, propulsion and (maybe) turret type (I'm thinking specifically of the warhead type used).
Sounds interesting.
NucNut wrote: An in game customisable interface may not be the way to go, but at the LEAST the targetting AI must be able to check that every unit/battery is not shotting at the same target
Usually it is exactly what you want though. In a tank slaughter to maximize the overall health of your units left after a battle you minimize the number of weapons enemy has.

It is the same with the SAMs:
Chojun wrote: If Warzone's targeting AI were efficient, and it defenses fired at separate targets based on what would deal the most damage and how much damage it takes to kill a target (as opposed to how it is now where 5 SAM sites will fire their missiles at the first target they see, while the rest of the targets will pass through unobstructed), then a player could feasibly make their base impenetrable to attack.
It is a special case with SAMs are VTOLs.

If enemy VTOLs come as an ant tail and jump out of the fog of war in the target range of all SAMs one by one then it is even theoretically impossible to ideally spread targets among all SAMs to maximize the number of killed VTOLs.

If we have x SAMs and 1 VTOLs just appeared, how many SAMs should fire at it?

It might be a single enemy VTOL or there might be 50 VTOLs left in the fog of war. Considering SAMs accuracy, damage and health of the VTOLs we can find some acceptable solutuion.
If accuracy is 50% we will need firepower capable of killing that VTOL at least twice. But we work with probabilities here, we can't risk too much otherwise too many VTOLs will survive and cause trouble in the base.
The problem is to find an optimal number of SAMs for a VTOL under uncertainty and even if we don't really need an optimal solution that's still not trivial and there are more subproblems to solve, the problem was simplified.
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 » 10 Sep 2007, 18:06

Troman wrote: Agreed and it was tweaked already. The units are not blindly targeting a random unit, they choose their target using several criterias: effectiveness of the unit weapon against different propulsion types, distance to the target, damage factor of enemy units, type of the unit (weapon droid, truck, repair, commander etc).


Usually it is exactly what you want though. In a tank slaughter to maximize the overall health of your units left after a battle you minimize the number of weapons enemy has.
* Ahh... Threat Assessment-Analysis.... kwel :) IMHO, absolutely a cornerstone foundation to implementing more advanced Command & Control Gpms for us who hanker after true field general status as WZ Commanders.

* If I may ask - How exactly was it tweeked ? (seems it would have involved more work than the word "tweeking" would suggest..)

If accuracy is 50% we will need firepower capable of killing that VTOL at least twice. But we work with probabilities here, we can't risk too much otherwise too many VTOLs will survive and cause trouble in the base.

The problem is to find an optimal number of SAMs for a VTOL under uncertainty and even if we don't really need an optimal solution that's still not trivial and there are more subproblems to solve, the problem was simplified.
* Ahh... I believe we are looking at what is called - Bounded Optimality....interesting.... If I recall Alex McLean wrote an A.I.  paper on it at University of Eddinberg. It was also something demonstrated beautifully in the EinStein Artificial War Lab.

- RV :)
Last edited by Rman Virgil on 10 Sep 2007, 19:00, edited 1 time in total.
.

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

Contrast
Reach
Exposure
Articulation
Trust
Echo
.

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

Re: What ever happened to...

Post by Chojun » 10 Sep 2007, 20:03

Unfortunately, threat assessment analysis already exists in a number of different games.  Our implementation of it would only be bringing WArzone up-to-date in that regard..

For instance, Warcraft 3 has a limited amount of it.  Units will target other units depending on how optimally they can damage them.  Also, the weaker units will be targeted, and that is balanced on the weakest unit that is capable of the most damage (if that makes sense)
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 » 10 Sep 2007, 21:10

Chojun wrote: Unfortunately, threat assessment analysis already exists in a number of different games.  Our implementation of it would only be bringing WArzone up-to-date in that regard..
And what do we need instead?
Rman Virgil wrote: * If I may ask - How exactly was it tweeked ? (seems it would have involved more work than the word "tweeking" would suggest..)
If you mean the algorithm itself, then by finding the most important criterias that influence the outcome of a battle. And the weights for the criterias with the usual trial and error plus game experience.
Rman Virgil wrote: * Ahh... I believe we are looking at what is called - Bounded Optimality....interesting....
I think you confused something here. Bounded Optimality deals with problems under computational constrains.
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 » 10 Sep 2007, 21:50

Troman wrote:
If you mean the algorithm itself, then by finding the most important criterias that influence the outcome of a battle. And the weights for the criterias with the usual trial and error plus game experience.
* Expert Sys MO, I figured. What I wondered was what was there originally to support it & what had to added.

I think you confused something here. Bounded Optimality deals with problems under computational constrains.
* Well I'm going by Russell who extended the '89 def within the context of a.i. strats in '95. HERE

* The same term can be used in 2 contexts.

* This is the context i'm useing:

Russell proposes four candidate definitions of intelligence: perfect rationality, calculative rationality, metalevel rationality, and bounded optimality.



-Cheers, RV :)
Last edited by Rman Virgil on 10 Sep 2007, 22:24, 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 » 11 Sep 2007, 14:43

Rman Virgil wrote: * Expert Sys MO, I figured.
You puzzled me here.  ???
To be honest I don't know how it is related. An expert system usually replaces a human expert, using a knowledge database and some form of interaction with the end user to get input, for example for a medical diagnosis.

The problem I was solving is a classical Multi-Criteria Decision Making  problem. My approach was similar to S.M.A.R.T: finding common currency, setting up weights, finding alternatives and finally calculating the best alternative. But not really the same.
Rman Virgil wrote: What I wondered was what was there originally to support it & what had to added.
Well they just grabed the closest target if I recall correctly.
Rman Virgil wrote: * Well I'm going by Russell who extended the '89 def within the context of a.i. strats in '95. HERE

* The same term can be used in 2 contexts.
The definition in your link is the same definition.

The main reason for Bounded Optimality as a class of problems is mainly limited computational power.
The word 'bounded' refers to computational resources, like CPU power, memory.
Bounded program is a program that simply can't afford longer computational time for some reason, be it because of the cost of the computational time or the real-time nature of the program (think chess or go).

I can't really classify SAM problem as a BO problem.

PS: I guess that's another hi-jacked topic. ;)
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 » 11 Sep 2007, 16:45

Troman wrote: You puzzled me here.  ???
To be honest I don't know how it is related. An expert system usually replaces a human expert, using a knowledge database and some form of interaction with the end user to get input, for example for a medical diagnosis.

The problem I was solving is a classical Multi-Criteria Decision Making  problem. My approach was similar to S.M.A.R.T: finding common currency, setting up weights, finding alternatives and finally calculating the best alternative. But not really the same.
Well they just grabed the closest target if I recall correctly.
* Forgive the novice in me - more artist than engineer in my thinking.

* Let me explain my process: an ES ai captures expertise, weights it for the  interaction with an end-user as you said - seperated occassion-events and brains.

* But what if those processes are NOT seperated: the expertise of a brain captured, that brain weighting the alternatives as well as correlating the prioritzed inputs that will determine the various outcomes.

The definition in your link is the same definition.

The main reason for Bounded Optimality as a class of problems is mainly limited computational power.
The word 'bounded' refers to computational resources, like CPU power, memory.
Bounded program is a program that simply can't afford longer computational time for some reason, be it because of the cost of the computational time or the real-time nature of the program (think chess or go).

I can't really classify SAM problem as a BO problem.
* Here's my process on this:

* I extended BO into a game program via the concept of "prioritized allocation of state resources" in a time-constrained decision-event.

* Anyway that was what was in my novice brain.

* Thank you for taking the time to share your expertise.

* As for hijacking the thread..... sorry... chit happens & hopefully we can chalk it up as a learning opportunity.

- Cheers, /R\m/a\n/ \V/i\r/g\i/l\  :)
.

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 » 11 Sep 2007, 19:04

SAM problem is an "Assignment Problem", under uncertainty in our case. Should have classified it sooner, but at first it seemed to be some kind of an unfamiliar problem to me.
Rman Virgil wrote: * As for hijacking the thread..... sorry... chit happens & hopefully we can chalk it up as a learning opportunity.
Well we sure can.

BTW when looking for examples I stumbled upon this.
Too bad the link to the full text of this one is not available
Sign Up for Beta-Testing:
?topic=1617.0

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

Re: What ever happened to...

Post by 2_Late » 12 Sep 2007, 01:46

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.
Order number 4E6F7721, April 1rst 4:30am, 2101 wrote: 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 Manuel fire interface for all ground base artillery and SAM facility's...

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, 07:23

Troman wrote:
SAM problem is an "Assignment Problem", under uncertainty in our case. Should have classified it sooner, but at first it seemed to be some kind of an unfamiliar problem to me.
* Kwel :)

BTW when looking for examples I stumbled upon this.
* Facinating. Was a wee bit familiar with the Monte Carlo applied within biz sim decision modeling but in a military context not at all. Kwel stuff. :)

Too bad the link to the full text of this one is not available
* That abstract was a real tease. I couldn't see how to get any of those docs from that repository....annoying.... but....

* "Lanchester-Type Equations" caught my eye - what the heck was that ?

* 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.

- RV :)


[/quote]
.

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

Contrast
Reach
Exposure
Articulation
Trust
Echo
.

Post Reply