DyDo skirmish AI: ver 2.1.1 (21 Apr 2010)

[guciomir]

http://developer.wz2100.net/changeset/9388

this also seems interesting, is something like that incorporated already in Dydo?

[Per]

http://developer.wz2100.net/changeset/9388

this also seems interesting, is something like that incorporated already in Dydo?

Requires r9387 which is currently only in trunk.

[DylanDog]

http://developer.wz2100.net/changeset/9388

this also seems interesting, is something like that incorporated already in Dydo?

<<If AI loses a race for an oil resource, it will now instead build a cheap defense building there as revenge.>>

DyDo already does this but only if no other OilRes are left.

[guciomir]

1. Dydo does not defend its derrics too well. At the same time a lot of its units are inside the base. Maybe defence range of defence group needs to be expanded? Also there are not enough defence towers near the derricks. Especially for derics near enemy.

2. Dydo is not agressive enough at the beginning of the game, it is easy to secure most of map using only cyborgs and a few minirocket pods towers.

3. Dydo does not respond to artillery fire well.

4. Tons of sensor towers are buing built but what for? Less would be enough.

5. Aircrafts come pretty late. When I have map secured they should attack my defences or my artillery near Dydo base. Why they are flying for the whole map and try to destroy my factories? It is impossible to fly to my factories when I control all the map and have a lot of defences.

6. I would like to attach some saved games, but where are they stored? I cant find them in my warzone folder (windows)

7. I would also suggest making the attack groups bigger, more units inside.In the early game the attack group does not have to be numerous, but after some time it should be bigger. Some scaling etc.

8. Dydo does not handle well when I have a lot of cyborgs (in the beginning of the game). Maybe it would be a good idea to concentrate on the machine gun technologies at the beginning of the game? It is effective vs cyborgs, structures and also wheeled vehicles.

[KukY]

6. I would like to attach some saved games, but where are they stored? I cant find them in my warzone folder (windows)

My Documents\Warzone 2100 <game version>\savegame
Place <savegame>.gam, <savegame>.es and <savegame> folder with everything in it into a ZIP(Or any other...)archive and upload.

[guciomir]

Thanks!

http://www.mediafire.com/?wwdzjknyyio

this is from the trunk snapshot




Also, I don't know why did it happen, but in these games defence group was gigantic, but it stayed in the Dydo base while my artillery was destroying everything (including the defence group)

[KukY]

Thanks!
http://www.mediafire.com/?wwdzjknyyio
this is from the trunk snapshot
Also, I don't know why did it happen, but in these games defence group was gigantic, but it stayed in the Dydo base while my artillery was destroying everything (including the defence group)

I thought "Upload on forum.". Try avoiding placing things on upload sites.

[guciomir]

it would be great if Dydo was faster at research
a) maybe use more research centres? either use fixed number like 4 or 5 (5 better) or make it dependable on the oil derrics owned. But i would seriously go with 5 centres all the time.
b) is it possible that the current research systems blocks research? as far as I understand you have 3 research trees and Dydo decides based on some factors which tree should it follow. So now imagine we have 2 research centres which just ending researching and are ready for a new topics. 1st centre choose technology "A" from "z" tree and 2nd centre tries to choose technology "A2" from the same tree, but fails because "A2" needs "A" completed.
c) right now Dydo researches a bit of everything, it would be far superior if it specializes is some technologies and do not waste time and money in not usefull stuff

[DarkCheetah]

i'v played dydo as well, i think GEN 0,1,2 and 3 are all the same

i tried dydo's Turtle perosnality and they dont even build a fortress, maybe like 4 towers on 1 spot and thats it, and they also build to much untits as a tutle, wich 1% of them attack, and rest stay home, staying in the way of his trucks

come on dydo you can do it!! upgrade ur ai!!

[DylanDog]

guciomir, thanks for this post!!

1. Dydo does not defend its derrics too well. At the same time a lot of its units are inside the base. Maybe defence range of defence group needs to be expanded? Also there are not enough defence towers near the derricks. Especially for derics near enemy.

Actually the defend range is half way between DyDo base and nearest enemy Base with the limitations to 45 tiles. If the Derricks attacked by you are outside that reange no droids are sent there...increase the defend range is not fine, maybe DyDo could defend derricks with VTOLs, I think I try to follow this way.

2. Dydo is not agressive enough at the beginning of the game, it is easy to secure most of map using only cyborgs and a few minirocket pods towers.

See my answer on point 7.

3. Dydo does not respond to artillery fire well.

The feature of defending against artillery is quite new and I recognize it requires some improvment, I will work on this but it is really a difficult thing to implement.

4. Tons of sensor towers are buing built but what for? Less would be enough.

DyDo build CB sensors towers when attacked by long range weapons. It doeas check if some CB is already in range before building a new one but I understand this is not enough, I will add another type of check to avoid too many CB are built.

5. Aircrafts come pretty late. When I have map secured they should attack my defences or my artillery near Dydo base. Why they are flying for the whole map and try to destroy my factories? It is impossible to fly to my factories when I control all the map and have a lot of defences.

VTOL management has been improved a lot in the version I am working on, you are right in DyDo 1.2.7 and lower VTOLs management was not that good.

6. I would like to attach some saved games, but where are they stored? I cant find them in my warzone folder (windows)

Someone else answered this, see post below.

7. I would also suggest making the attack groups bigger, more units inside.In the early game the attack group does not have to be numerous, but after some time it should be bigger. Some scaling etc.

good point but I desagree a bit that at the beginning the attack groups could be not that big. Attacking with few units is not ok if the target is a base as it has defences + droids for sure. I think what I can improve here is that DyDo should attack enemy derricks with few units asap. This is definitely something I can introduce but only as a random option to be executed every now and then.

8. Dydo does not handle well when I have a lot of cyborgs (in the beginning of the game). Maybe it would be a good idea to concentrate on the machine gun technologies at the beginning of the game? It is effective vs cyborgs, structures and also wheeled vehicles.

mmm...DyDo concentrate on flamers at the very beginning, they should be effective as well as MG against cyborgs. This type of balances are not easy to implement in an AI. If the AI concentrates too much on MG & flamers then the human player develops tanks if DyDo concentrates too much on rockets and lancer then the human player develops cyborgs. If the AI concentrates to develop all techs then the general weapon development is very slow.

I think you have suggested some good things and given me some work!!!

[DylanDog]

it would be great if Dydo was faster at research
a) maybe use more research centres? either use fixed number like 4 or 5 (5 better) or make it dependable on the oil derrics owned. But i would seriously go with 5 centres all the time.
b) is it possible that the current research systems blocks research? as far as I understand you have 3 research trees and Dydo decides based on some factors which tree should it follow. So now imagine we have 2 research centres which just ending researching and are ready for a new topics. 1st centre choose technology "A" from "z" tree and 2nd centre tries to choose technology "A2" from the same tree, but fails because "A2" needs "A" completed.
c) right now Dydo researches a bit of everything, it would be far superior if it specializes is some technologies and do not waste time and money in not usefull stuff

a)ok, what I was thinking to implement is to relate the number of resLabs & Factories to the number od derricks in the map. This would be the best thing to do. Do you agree?

b)No it is not blocking and there is one research tree (split into three arrays) for each Personality, if DyDo tries to pursue something with cannot be researched (because of pre-requisites or because an ally is already researching) then a random research is done. Also DyDo checks with researchStarted() and researchFinished() before doing a research.

c) correct! it researches a bit of everything. Put together some research trees for DyDo and let`s see if I can implement it and if this makes DyDo stronger.

You do not need to put all researchers from T1 to T3, let`s start with T1 level. Put three different research trees together (20 researches each) in the following format:

R-Wpn-MG1Mk1
R-Wpn-MG-Damage04
R-Wpn-MG-ROF02
...and so on

Or you can try it on your own, it is not that difficult to put the research in DyDo. For example to put the researches for General1 in DyDo player #0 you have to work on player0.vlo file and replace the value between "", in the below showed section of the player0.vlo file, with the ones you want DyDo to research:

Code: Select all

  research[0][0]	RESEARCHSTAT	"R-Sys-MobileRepairTurret01"
  research[0][1]  RESEARCHSTAT  "R-Wpn-Flamer-ROF03"
  research[0][2]  RESEARCHSTAT  "R-Wpn-Flamer-Damage03"
  research[0][3]	RESEARCHSTAT	"R-Struc-Research-Upgrade03"
  research[0][4]	RESEARCHSTAT	"R-Struc-Power-Upgrade01b"	  

  //Cobra & Hover ---------------------
  research[0][5]  RESEARCHSTAT  "R-Vehicle-Body05"
  research[0][6]	RESEARCHSTAT	"R-Vehicle-Prop-Hover"	
  research[0][7]  RESEARCHSTAT  "R-Defense-WallTower01" 
  research[0][8]  RESEARCHSTAT  "R-Defense-Tower06" 
  research[0][9]  RESEARCHSTAT  "R-Wpn-Flame2" 
  research[0][10]  RESEARCHSTAT  "R-Defense-MRL" 
  research[0][11]  RESEARCHSTAT  "R-Defense-WallTower03" 

[DylanDog]

i'v played dydo as well, i think GEN 0,1,2 and 3 are all the same

i tried dydo's Turtle perosnality and they dont even build a fortress, maybe like 4 towers on 1 spot and thats it, and they also build to much untits as a tutle, wich 1% of them attack, and rest stay home, staying in the way of his trucks

come on dydo you can do it!! upgrade ur ai!!

I know I have to improve VTOL, Cyborg and Turtle AI. I am concentrated on General 1 for now. This is the core AI. Next step will be the Turtle one.

[guciomir]

I have some other comments. I think you are making a great job, but still there is a place for progress.

Actually the defend range is half way between DyDo base and nearest enemy Base with the limitations to 45 tiles. If the Derricks attacked by you are outside that reange no droids are sent there...increase the defend range is not fine, maybe DyDo could defend derricks with VTOLs, I think I try to follow this way.

Yes, defending derriks with VTOLs is a good idea.
I would also suggest creating an event when Dydo decides to split its defend group into subgroup A and subgroup B. Defend range for "A" is still 45 tiles, while range for "B" is bigger so Dydo defends derricks with this subgroup. I have no idea if it is good solution, what is your opinion?

The feature of defending against artillery is quite new and I recognize it requires some improvment, I will work on this but it is really a difficult thing to implement.

In one of the attached savegames my artillery was killing Dydo defend group and it did not try to attack my mortars/hovitzers. Vtols should attack my artillery and also what is the purpose of defend group if it defends only vs short range weapons? Maybe defend range should be increased when player builds artillery near Dydo? But in this case it would be easy to lure Dydo into trap. What is your plan here?

a)ok, what I was thinking to implement is to relate the number of resLabs & Factories to the number od derricks in the map. This would be the best thing to do. Do you agree?

I am no expert here I think someone like Zarel would be a good person to advice. From my point of view 4-5 research centres are always very important, while the number of factories can depend on the size of map (number of derrics)

c) correct! it researches a bit of everything. Put together some research trees for DyDo and let`s see if I can implement it and if this makes DyDo stronger.

I'd love to, but I do not feel to be enough experienced for this. C'mon skirmish/multiplayer guys, you should be able to help Dydo with that!!!!! If more experienced players does not suggest any research trees, I will try to produce my own

Few hints (maybe more experienced players will disagree)
- skip flamers but use that research time for machine guns / cannons / rockets . It will make them stronger.
- its good to have "energy/research" tech as fast as possible

More ideas:
1. implement event when Dydo builds cyborg transports, puts infantry inside and then send transports near the edge of the map and attack player base from the behind
2. I think that you can make an event when defend group resupplies attack group. In the games I play the Dydo attack group was always small but with good tech and defend group was huge in numbers but weak in technology. I think this is because Dydo sends attack group right away and its is being constantly "recycled" because units die, units are created. On the other hand defend group sits in Dydo base, does nothing and is becoming bigger and bigger but from the time perspective many units are very old. What I suggest is to from time to time, swap units from Defend group into attack group. This way units will be better used, because their technology will still be able to do something.
3. maybe you could create new topic where you ask for help with research tree. I think many good players does not read this topic because it may be too technical for them. Maybe they will post their research trees or at least make some debate over which technologies are the best.

[DarkCheetah]

I know I have to improve VTOL, Cyborg and Turtle AI. I am concentrated on General 1 for now. This is the core AI. Next step will be the Turtle one.

i like that that you will do turtle next! you must give them a base template like the orginal turtle ai, so they will grow and grow and grow, and build wals and more variable weapons

[Arreon]

I have a suggestion.

I don't know how to do programming, but I want to offer a way to improve the AI.

Maybe you can reconfigure the AI so that if their trucks get stuck to each other, maybe you could have the AI move their trucks in the opposite direction. This could be triggered by a player command like "move your trucks". This may help the AI as they can now assign more trucks to tasks, rather than having them try to drive through each other.

Also, I think the WZ community would appreciate it if the AI uses every defensive structure available by research. I'm glad you've put the ripple rockets in (I'm a big fan of those), but if you haven't already, go ahead and add in the Archie Missiles, Seraph, Heavy Laser Emplacements, Twin Assault Gun hardpoint, and maybe even the Flashlight Emplacement.

Some new templates (like Plasmite-Retribution-Hover) would be great as well. A few I have in mind are:

1: Heavy MG-Scorpion-Hover
2: Pulse Laser-Vengeance-Hover
3: Seraph-Vengeance-Hover
4: HPV Cannon-Scorpion-Half-tracks
5: Medium Cannon-Scorpion-Half-tracks
6: Heavy Flamer-Cobra-Hover

For extra points you could implement mobile sensor and artillery units.