DyDo skirmish AI: ver 2.1.1 (21 Apr 2010)

[DylanDog]

I think someone had mentioned that artillery was hard to work into AI, but to counter this

- it can not defend against artillery protected by some droids (it ries to counterattack, but my droids are enough to protect artillery)

DyDo counter-attacks with all droids available, if the attacker has more droids than DyDo, well there is nothing DyDo can do.

and to improve this

- it counterattacked when my artillery started attacking its base (failed, but tried!)

make use of, and strategicly build CB towers.

This is already implemented in DyDo, possibly needs to be improved. Actually DyDo builds CB and artillery to defend against enemy artillery.

Why is it hard to use artillery in AIs?

It is really hard to use artillery on tanks because you have to manage sensors and tanks in a coordinate way.

[DylanDog]

As always thanks a lot for the post!

I downloaded newest beta and played DyDo there. General 1, 2player map in Arizona. advanced bases

I already posted somewhere that in 1vs1 games DyDo is still too weak. You better test it in 4/8 players game.
My idea is to slowly re-balance DyDo for 1vs1 games too but really it is not an easy thing make an AI able to beat a human player in 1vs1 games without using cheats.

Please test DyDo 2.1 using the DyDo challenges on medium/hard level and let me know.

Things that Dydo does wrong:
- it does not recognize that my army is mostly cyborg and it builds against them minirockets

Fixed in 2.1, when the human player is the last enemy left (no other enemies AI are left) DyDo builds droids with the best weapon to beat the human player. This means that if the human player has MAINLY cyborgs DyDo build 70% of the droids with AP weapon, 30% is left to AR to destroy defenses (as you suggested to myself via PM !!).

- I do not "research all" but have a quick research path that leads me to more advanced technology faster. Dydo is slow at research, because it still uses t1-t2 weaponry when I have t3 already

I know, this is the biggest issue I have with DyDo actually. Actually it pursue almost all type of researches. Change this is also a complicated thing but is a must have.
Added in the to-do list.

- it can not defend against artillery protected by some droids (it ries to counterattack, but my droids are enough to protect artillery)

If your droids are enough there is nothing DyDo can do as it counter attacks with all available droids. This means that at that point you are already too strong for DyDo. This again because in 1vs1 games DyDo is still too weak (even if stronger then the standard AI)

- i did not see a single vtol during the game, don't know why It would be great if vtols attack my artillery

This is strange, cannot tell why. Maybe because VTOL tech was still not researched? Anyway when DyDo has VTOLs and is attacked by artillery then it counter-attacks with VTOLs too.

- dydo still sends trucks to death, when my artillery has oil in range and dydo wants to build derrick there

Added in the to-do list

[Berg]

Tried to email these to you ..example of new logo's

[guciomir]

I played 1-1 on 2.1 Dydo. I will play 4 players game and challenges too

1. Dydo is even slower at research (compared to some previous versions). I had autorepair, tank killers, hovitzers while Dydo still had no heavy bodies.
I think it is because Dydo build more defense (etc) so it has fewer cash for research.
2. Choose best weaponry feature works nice, but I think it still counts my trucks, because I see minirockets while my army is cyborg only.
3. Usually the fight looks like:

S
S
SS
SS - Dydo forces
S

XXXXXX - my defences
YY - my trucks
OOOO - my mortars

Dydo forces could have been less scattered, so they could concentrate fire better.
And VERY important, Dydo forces do not concentrate fire at 1 of mine towers, but they attack many of them. They damage all of my defences, but do not kill any of them. I think it would be better if Dydo attack 1 tower at the same time, kill it and then attack 2nd tower. (same with units).
4. I build sensor near Dydo base and saw the following:
When Dydo builds a tower, it sends a lot of cyborgs and trucks to help building (10 or more). So this building forces go from 1 point, build a tower there and then goes to other side of the base, build another tower etc. Dydo lose a lot of time by not dividing his trucks.


my army of cyborgs was killed easily by Dydo when I sent it to attack - nice! now this simple strategy is not so good anymore


edit:
5. I played a game with Dydo allied. It was mainly testing as I did not build anything, I was looking what my ally will do. He had very good access to power so this was not an issue. He had 5 research centres although 2-3 were used all the time. Sometimes four. I believe you have some conditions that decide when Dydo starts the research, maybe they are too severe?
6. I can not play a challenge when I launch dydo challenge the screen freezes before the map load and I have to ctr alt del .

edit2:
7. I played a nice 4p free for all game. Main issue is that Dydo does not research ANY heavy body until very late game. I had lasers, twin assault guns and first needles when I saw first python in Dydo. I think this is a bug, because Dydo was using super cyborgs, lasers and still medium + light bodies on tanks.

[DylanDog]

Tried to email these to you ..example of new logo's

I like it!! added on 2.1.1

[DylanDog]

to play the Dydo challenges you obviously need the maps. This will be fixed in the next version of DyDo, when such maps will be part of the DyDo package.

[guciomir]

I played one of them - game was very fun, Dydo is powerfull there.

I think you should ask a map maker to help you with creating those maps. Their layout is entertaining, but they look quite ugly. This has nothing with the game itself (it is great to play vs Dydo there), but I guess they need to be improved in artistic way before they are put into official release.

[DylanDog]

Feel free to suggest maps to be used in challenges, I will then test them and will put in the DyDo package. I need to test them to be sure that droids do not stuck too often. There are a lot of good maps but with issues which causes droids to stuck in certain positions. You can also test them, just use the cheats every now and then to look an aI playing and check if it has any droid stuck.

[DylanDog]

Guys, I have modified a bit the scavenger AI. It will be part of the next DyDo version but you have play with it now too, see this post.

[guciomir]

Dydo, did you create those maps for challenges yourself? If yes, you could create a topic somewhere in the forums to ask map makers, so they help you with making those maps nice looking.

[DylanDog]

Dydo, did you create those maps for challenges yourself? If yes, you could create a topic somewhere in the forums to ask map makers, so they help you with making those maps nice looking.

I donnow how to create MAPs and wouldn' t have the time anyway for that

[kringled]

I played a couple of games against DyDo 2.1.1 over the last week. Game 1: Map was Highground, T1 advanced bases, with scavengers. The aggressive scavengers did not allow DyDo to shine, and it was not much harder to defeat than the standard AI. It did use repair facilities intelligently, and most of the mobile units that defended the base were Cobra bodies.
The second game was on the map "Mountain" , 4 player Free For All, T1 simple bases, with scavengers. The DyDo personalities came out randomly at Orange: Turtle AI; Black, Turtle AI, White/Gray Cyborg AI. Overall, DyDo put up more of a fight than the standard AI. I still won in about the same time, but I took a lot more unit losses.

What DyDo did well: Groups defended their bases in large numbers and to strong effect. When they had them, repair facilities were used. Gray brought a squad to attack my base (from a direction I hadn't anticipated!) and they proved very stubborn to get rid of. (First time I've ever built cyborg machinegunners with the intention of actually using them.) DyDo does not seem to overbuild factories and underbuild units (which is what stock AI
seems to look like to me.
What DyDo did not do so well with: All 3 of the DyDo players were using primarily viper bodies until the very end of the game (or their end as a player). The cyborg attack squad was equipped primarily with flamers and machine guns, which was why they were so effective at defending themselves from my mostly cyborg based group that was in position to defend the home base, but it also made them quite slow to destroy structures. Beyond the bodies, it seemed like for the most part, DyDo research was lagging far behind. Orange did manage to get to VTOL when I hadn't quite started, but was only able to put out 3 units (with low-yielding Lancers) before I got to their rearming pads and factories. I would suggest that you might consider some more narrowly defined (and randomly selected) weapon research priorities - a couple of combos I've done are (rocket/machine gun) and (cannon/flamer).


As an aside, I think I was able to defeat the enemy in the same time (more or less) as with stock AI,despite the stiffer resistance that DyDo put up. Making more units means making less factories, and with the stock AI, I often spend more time just shooting up an enemy base than I do attacking enemy units. So it took more to break DyDo's defenses, but once I did, the base went down quicker.

Does Dydo still have real problems with the maps Valley and UrbanDuel?
I'd like to try it out on another map - I have in mind GreatRift, as I played that recently vs. stock AI and found the stock AI very placid.

Keith

[kringled]

Played against DyDo 2.1.1 on GreatRift, T1 simple bases FFA. All three DyDo AIs came out as Standard AI, the [difficulty] was at 5 for 2 of them and 3 for one. Standard AI did a bit better at defending its base with medium body units; more than one of them had Inferno units, which definitely sent my cyborg group running back to it's repair facility. DyDo was able to launch several attacks on my home base or more accurately, the oil resources near the entrance to it, where they got kind of hung up until I brought units back to defend. I also lost a few units who were coming out of a factory to join one of my combat groups, in an ambush from one of the AIs. Overall, though, it still seems like the AI lags in researching. For example, in the games I've played, Lancers don't seem to be built very often. I may have to try some maps with 1vs1vs2 allied DyDo players (with me as the 1). I think the things I use to win are: 1) I tend to use command tanks, which allow me to build up experience over time and make my units much more effective. 2) I seem to be more effective at researching critical tech than DyDo. Couple these two factors together and while I may be just able to hold my own early on, as the game progresses I tend to build a pretty big advantage. 3) I tend to specialize in certain weapons combos, which adds to my tech advantage. 4) I'm less likely to order units into do or die situations, which maximizes my economic efficiency. (Which I need, as I suspect I am putting a lot more of my energy into research than DyDo.

Keith

[DylanDog]

I may have to try some maps with 1vs1vs2 allied DyDo players (with me as the 1).
Keith

Hello Keith

Thanks for testing DyDo.

I am trying to find a way to make the scavenger AI allied with DyDo. Actually the do help more the human that DyDo.

Yes for the most of the players playing 1vs1 or 1vs1vs1vs1 could not be that interesting. This because DyDo actually researches a bit of everything, making some research more specific is on the to-do list (last point) but it is not that easy because I have to re-balance also the part of the script which decides which templates has to be built.

Please use the last version you can find on the DyDo homepage.

I would like to see your comment for example in DyDo-challenge on the map CNC-Assault medium level where you play 1(you)vs1(DyDo)vs2(DyDo). (DyDo-challenges are included in the WZ file you download from the website)

[guciomir]

I played the challenge and liked it (1v1v2). I lost my first game because Dydo launched nice attack in the early game. In the second I was prepared better and killed Dydo thanks to artillery and thanks to researching faster. (1v3 would be much harder I guess).

One question:
when 2 Dydo's are allied, how they cooperate with research?